For DXJ Writer: Read Me First!
Week One Read Me First
INTRODUCTION TO RESEARCH IN CRIMINAL JUSTICE
This course introduces the student to concepts of research in criminal justice. In this section, we explore the basics of what research is and what it can do in the criminal justice system. This week, the student is exposed to a variety of new terminology relating to research and meet the following objectives.
This Week in Relation to the Course
CJA/334 addresses various questions and problems of various components of the criminal justice system. As with other disciplines, criminal justice is sometimes difficult to define, as it involves issues that involve cross-discipline issues from psychology and public administration to sociology and law. Criminal justice research addresses all aspects from police, courts, corrections, criminal procedures, juveniles, and a host of other mundane problems faced by stakeholders.
Heffner (2004) states that research is the foundation of all sciences. Research is a controlled effort to acquire knowledge about a problem. Heffner writes that research uses the process of deductive reasoning which “refers to a structured approach utilizing an accepted premise (known as a major premise), a related minor premise, and an obvious conclusion” (2004).
According to Heffner (2004), the research process may fluctuate, given the type of research being done, the hypothesis being tested, and the questions being answered. For the physical scientist, the process would involve a more comprehensive structured approach, whereas the social sciences may use a less structured approach, such as surveys, interviews, and other nonintrusive measures. A universal method for accomplishing productive research is in the table below.
Not only are steps in research important, but the ethics of the endeavor are also important. What are the main ethical issues in criminal justice research? Most criminal justice research involves using human subjects as participants and presents a variety of issues that must be addressed by investigators. The primary concern should be that no harm comes to participants. Investigators must secure participants’ informed consent and objectively weigh the risks and benefits of the research. Another issue that often plagues research is information confidentiality. How will the information be protected from curiosity seekers or nonethical researchers? Above all, procedures must be in place to address any exigent circumstance that may arise.
Table 1: Steps Involved in the Research Process
(adapted from Heffner, 2004)
Heffner, C. (2004). Research methods: Chapter one: Introduction to research. AllPsychOnline. Retrieved from http://allpsych.com/researchmethods/introduction.html
Practical Applications and Questions
What is research? What provides discipline in the research process? Why is this discipline important? What are the similarities and differences between basic and applied research? What are the similarities and differences between qualitative and quantitative research? What distinguishes scientific research from research or problem solving in everyday life? What are the similarities and differences between inductive and deductive logic? Why should someone know about research methods? Compare and contrast subjectivity and objectivity. What parts do replication and verification play in scientific research? Why should researchers share results with their subjects? From what sources might research topics emerge? What are research ethics? What is a code of conduct? What is a paradigm? How does this contribute to building theory? What is a variable? Explain the difference between an independent variable and a dependent variable. Provide an example of each. What is the Lucifer Effect? Provide a real-life example. What is an institutional review board? Why is this important for the academic and research communities? What is a research population? What is a research sample? Provide an example of each. What is a random sample? Why is this important to the efficacy of an effective research study?
How Tools, Readings, and Simulations Help Solidify Concepts
This week, there are three chapters from the textbooks and several additional readings. As you review the textbooks, remember that we are talking about research methods. Research is the mechanism by which the efficacy of theories is usually tested. Theory almost always provides guidance for research and how it is implemented.
Ethics are standards of professional groups or organizations that spawn ethical codes. Several different problems arise in criminal justice research from plagiarism to statistical manipulations to research harmful to human subjects. Those issues must also be reviewed and discussed.
Criminal justice research is the same as all other research disciplines. The discipline is an organized and systematic way of finding answers to relevant questions and problems. The process is well thought out and preplanned to achieve the outcomes. In the quest for finding the answers and solutions to questions and problems, the results often lead to other problems and questions. Research must have focus and a purpose.
In the ethics of research, we have an obligation to distinguish between what is ethics and what is law, because there are common characteristics. As Goodman & Miller (2000) states, “Ethics offers conceptual tools to evaluate and guide moral decision making. Ethical considerations apply in attempts to determine what is good or meritorious and which behaviors are desirable or correct in accordance with higher principles” (pp. 370-402). Laws, on the other hand, allow society a rule to live by under specific circumstances and prescribe legal sanctions, such as fines or prison time, for those who do not conform to the law.
The legal philosophy frequently evolves from ethical issues. Whereas legal principles deal with the realistic control of morality, ethical principles deal with the individual’s approach to morality. These two principles are important for the researcher to abide by. They must know the legal ramifications of illegal behavior and the ethical ramifications of various actions. In criminal justice research, we must adhere to the process of self-regulation, lest the imposition of laws to regulate ethical behavior in research.
Goodman, K. W., & Miller, R. A. (2000). Ethics and Health Informatics: Users, Standards, and Outcomes. In E.H. Shortliffe & L. E. Perrault (Eds.), Medical Informatics: Computer Applications in Health Care and Biomedicine (pp. 379–402), New York, NY: Springer
Heffner, C. (2004). Research methods: Chapter one: Introduction to research. AllPsychOnline. Retrieved from http://allpsych.com/researchmethods/introduction.html
Prepare a 1,050- to 1,750-word paper in which you describe the research process. Include the following: · Include new terminology learned from the reading. · How will this new terminology and knowledge apply to a career in criminal justice? · How can not knowing the proper terminology affect you as you conduct criminal justice research? · How will knowing these terms be an asset to you when evaluating and analyzing research studies or data? Format your paper consistent with APA guidelines
Source for this class-(5 weeks):
Research Methods in
Criminal Justice and
E i g h t h E d i t i o n
Frank E. Hagan
Upper Saddle River, New Jersey
Research Methods in Criminal Justice and Criminology
, Eighth Edition, by Frank E. Hagan. Published by Prentice Hall. Copyright © 2010 by Pearson Education, Inc
The first edition of
Research Methods in Criminal Justice and Criminology was prepared in theearly 1980s, when no comprehensive research text existed that directly addressed the areas ofcriminal justice and criminology.The text remains a comprehensive one, emphasizing sources and resources of classic andcontemporary research in the field. There continues to be an acceleration of publications inthe field, employing increasingly sophisticated and esoteric research designs and statisticalanalysis. The intent of the eighth edition remains the same as the first seven: to reduce the gapthat exists between the types of materials appearing in professional journals and publicationsin the field and the ability of students and professionals to understand them. The approach is touse criminological and criminal justice studies to illustrate research methods, because it isas important to become familiar with examples of research in the field as it is to learn fundamentalresearch skills.This edition features revisions throughout, while retaining a vital core of material from thefirst seven editions. The organization of the work will carry the student through the sequence ofthe research process. Instructors may wish to shuffle the order of the chapters, however, to suittheir syllabus or research style.The first chapter introduces the reader to the area of criminological and criminal justiceresearch while attacking commonsense approaches to research. Chapter 1 also outlines the stepsin research elaborated on in Chapters 3 through 11. Following the issue of problem formulationin the first chapter, Chapter 2 examines the important issue of research ethics. Research designsand the experimental model, the latter being a benchmark with which to compare all otherresearch in criminal justice, are detailed in Chapter 3.In Chapter 4, the Uniform Crime Reports and its major revisions are examined, as are thevarious sampling strategies used in research. Chapter 5 looks at survey research, particularlymail questionnaires and self-report studies. Chapter 6 concentrates on interviews and telephonesurveys, particularly recent developments in victim surveys. Also featured are Internet surveys.Participant observation and case studies are the subject of Chapter 7. Such field studies representsome of the most fascinating literature in the field.Chapter 8 explores the interesting world of nonreactive or unobtrusive techniques, whichinclude criminal justice and criminological applications involving secondary and content analysis,physical trace analysis, the use of official data, and observational strategies—all of which areuseful, cost-effective means of gathering data. Alternative means of data gathering such as surveys,field studies, and unobtrusive methods often contain strengths missing in experimental research.The important issues of validity and reliability are detailed in Chapter 9; the triangulated strategiesare proposed as the single most logical path by which to resolve these questions. In all of thesechapters, examples of both classic and contemporary research in criminal justice and criminologyare used as illustrations. In addition to providing an overview of research methods, this text alsopresents a review and analysis of research literature.Chapter 10 discusses scaling and index construction and features new and expanded coverageof crime severity scales, salient factor scores, and prediction scales. Chapter 11 discusses evaluationresearch and policy analysis that reflects the growing interest of the social sciences in these subjectsin the past few years.
was prepared in theearly 1980s, when no comprehensive research text existed that directly addressed the areas ofcriminal justice and criminology.The text remains a comprehensive one, emphasizing sources and resources of classic andcontemporary research in the field. There continues to be an acceleration of publications inthe field, employing increasingly sophisticated and esoteric research designs and statisticalanalysis. The intent of the eighth edition remains the same as the first seven: to reduce the gapthat exists between the types of materials appearing in professional journals and publicationsin the field and the ability of students and professionals to understand them. The approach is touse criminological and criminal justice studies to illustrate research methods, because it isas important to become familiar with examples of research in the field as it is to learn fundamentalresearch skills.This edition features revisions throughout, while retaining a vital core of material from thefirst seven editions. The organization of the work will carry the student through the sequence ofthe research process. Instructors may wish to shuffle the order of the chapters, however, to suittheir syllabus or research style.The first chapter introduces the reader to the area of criminological and criminal justiceresearch while attacking commonsense approaches to research. Chapter 1 also outlines the stepsin research elaborated on in Chapters 3 through 11. Following the issue of problem formulationin the first chapter, Chapter 2 examines the important issue of research ethics. Research designsand the experimental model, the latter being a benchmark with which to compare all otherresearch in criminal justice, are detailed in Chapter 3.In Chapter 4, the Uniform Crime Reports and its major revisions are examined, as are thevarious sampling strategies used in research. Chapter 5 looks at survey research, particularlymail questionnaires and self-report studies. Chapter 6 concentrates on interviews and telephonesurveys, particularly recent developments in victim surveys. Also featured are Internet surveys.Participant observation and case studies are the subject of Chapter 7. Such field studies representsome of the most fascinating literature in the field.Chapter 8 explores the interesting world of nonreactive or unobtrusive techniques, whichinclude criminal justice and criminological applications involving secondary and content analysis,physical trace analysis, the use of official data, and observational strategies—all of which areuseful, cost-effective means of gathering data. Alternative means of data gathering such as surveys,field studies, and unobtrusive methods often contain strengths missing in experimental research.The important issues of validity and reliability are detailed in Chapter 9; the triangulated strategiesare proposed as the single most logical path by which to resolve these questions. In all of thesechapters, examples of both classic and contemporary research in criminal justice and criminologyare used as illustrations. In addition to providing an overview of research methods, this text alsopresents a review and analysis of research literature.Chapter 10 discusses scaling and index construction and features new and expanded coverageof crime severity scales, salient factor scores, and prediction scales. Chapter 11 discusses evaluationresearch and policy analysis that reflects the growing interest of the social sciences in these subjectsin the past few years.
early 1980s, when no comprehensive research text existed that directly addressed the areas of
criminal justice and criminology.
The text remains a comprehensive one, emphasizing sources and resources of classic and
contemporary research in the field. There continues to be an acceleration of publications in
the field, employing increasingly sophisticated and esoteric research designs and statistical
analysis. The intent of the eighth edition remains the same as the first seven: to reduce the gap
that exists between the types of materials appearing in professional journals and publications
in the field and the ability of students and professionals to understand them. The approach is to
use criminological and criminal justice studies to illustrate research methods, because it is
as important to become familiar with examples of research in the field as it is to learn fundamental
This edition features revisions throughout, while retaining a vital core of material from the
first seven editions. The organization of the work will carry the student through the sequence of
the research process. Instructors may wish to shuffle the order of the chapters, however, to suit
their syllabus or research style.
The first chapter introduces the reader to the area of criminological and criminal justice
research while attacking commonsense approaches to research. Chapter 1 also outlines the steps
in research elaborated on in Chapters 3 through 11. Following the issue of problem formulation
in the first chapter, Chapter 2 examines the important issue of research ethics. Research designs
and the experimental model, the latter being a benchmark with which to compare all other
research in criminal justice, are detailed in Chapter 3.
In Chapter 4, the Uniform Crime Reports and its major revisions are examined, as are the
various sampling strategies used in research. Chapter 5 looks at survey research, particularly
mail questionnaires and self-report studies. Chapter 6 concentrates on interviews and telephone
surveys, particularly recent developments in victim surveys. Also featured are Internet surveys.
Participant observation and case studies are the subject of Chapter 7. Such field studies represent
some of the most fascinating literature in the field.
Chapter 8 explores the interesting world of nonreactive or unobtrusive techniques, which
include criminal justice and criminological applications involving secondary and content analysis,
physical trace analysis, the use of official data, and observational strategies—all of which are
useful, cost-effective means of gathering data. Alternative means of data gathering such as surveys,
field studies, and unobtrusive methods often contain strengths missing in experimental research.
The important issues of validity and reliability are detailed in Chapter 9; the triangulated strategies
are proposed as the single most logical path by which to resolve these questions. In all of these
chapters, examples of both classic and contemporary research in criminal justice and criminology
are used as illustrations. In addition to providing an overview of research methods, this text also
presents a review and analysis of research literature.
Chapter 10 discusses scaling and index construction and features new and expanded coverage
of crime severity scales, salient factor scores, and prediction scales. Chapter 11 discusses evaluation
research and policy analysis that reflects the growing interest of the social sciences in these subjects
in the past few years.
, Eighth Edition, by Frank E. Hagan. Published by Prentice Hall. Copyright © 2010 by Pearso
Data analysis is the subject of Chapters 12 and 13, with Chapter 12 examining data
management activities such as coding, keyboard entry, and table reading and Chapter 13 providing a
user’s guide to statistics. The latter is intended as a quick reference guide to many of the major
statistical techniques presented in the literature. Although the primary goal of this chapter is to
provide the reader with the ability to recognize and interpret the meaning of statistics, “pop quizzes”
and additional appendixes have been added to assist in improved comprehension. It is hoped that the
style of presentation will convert many readers who may begin the course with apprehension into
relatively fluent users of “researchese,” a valuable and useful international language.
In addition to updating tables, figures, references, and examples, some principle changes
have been made in this edition in response to reviewer and user suggestions. Useful Web sites
have been provided in all chapters. New to this edition is discussion of Zimbardo’s “Lucifer
effect,” controversies related to the Human Terrain System and Minerva Consortium, advice on
interviewing active offenders and gaining entry to correctional facilities, Steffensmeier and
Confessions of a Dying Thief, and discussion of the violent and property crime indexes.Also featured are Sherman’s Scientific Methods Scale, visual criminology, the Scarlet M incorrections research, resolution of the Iowa “Monster study,” the current status of shield laws,and telephone focus groups.I would like to thank the many people who assisted me in various ways in writing theeditions of this text. I would like to express my appreciation to those at Pearson Prentice Hall fortheir encouragement and assistance on this project. Tim Peyton, Senior Acquisitions Editor;Jessica Sykes, Project Manager; Alicia Wozniak, Senior Marketing Coordinator; and DanTrudden, Developmental Editor, were all instrumental in getting this project done. I would alsolike to thank past reviewers, Howard Abadinsky, John Hudzik, and John Smykla for their helpfulreviews of the first edition, as well as James A. Adamitis, the University of Dayton; Rosy A.Ekpenyong, Michigan State University; RXXXXX XXXXX, Indiana University of Pennsylvania;Robert J. Mutchnick, Indiana University of Pennsylvania; Shirley R. Salem, Southern Universityat New Orleans; and Frank Schmallenger, Ph.D., Editor of
, and discussion of the violent and property crime indexes.Also featured are Sherman’s Scientific Methods Scale, visual criminology, the Scarlet M incorrections research, resolution of the Iowa “Monster study,” the current status of shield laws,and telephone focus groups.I would like to thank the many people who assisted me in various ways in writing theeditions of this text. I would like to express my appreciation to those at Pearson Prentice Hall fortheir encouragement and assistance on this project. Tim Peyton, Senior Acquisitions Editor;Jessica Sykes, Project Manager; Alicia Wozniak, Senior Marketing Coordinator; and DanTrudden, Developmental Editor, were all instrumental in getting this project done. I would alsolike to thank past reviewers, Howard Abadinsky, John Hudzik, and John Smykla for their helpfulreviews of the first edition, as well as James A. Adamitis, the University of Dayton; Rosy A.Ekpenyong, Michigan State University; RXXXXX XXXXX, Indiana University of Pennsylvania;Robert J. Mutchnick, Indiana University of Pennsylvania; Shirley R. Salem, Southern Universityat New Orleans; and Frank Schmallenger, Ph.D., Editor of
Also featured are Sherman’s Scientific Methods Scale, visual criminology, the Scarlet M in
corrections research, resolution of the Iowa “Monster study,” the current status of shield laws,
and telephone focus groups.
I would like to thank the many people who assisted me in various ways in writing the
editions of this text. I would like to express my appreciation to those at Pearson Prentice Hall for
their encouragement and assistance on this project. Tim Peyton, Senior Acquisitions Editor;
Jessica Sykes, Project Manager; Alicia Wozniak, Senior Marketing Coordinator; and Dan
Trudden, Developmental Editor, were all instrumental in getting this project done. I would also
like to thank past reviewers, Howard Abadinsky, John Hudzik, and John Smykla for their helpful
reviews of the first edition, as well as James A. Adamitis, the University of Dayton; Rosy A.
Ekpenyong, Michigan State University; RXXXXX XXXXX, Indiana University of Pennsylvania;
Robert J. Mutchnick, Indiana University of Pennsylvania; Shirley R. Salem, Southern University
at New Orleans; and Frank Schmallenger, Ph.D., Editor of
The Justice Professional, for theirmany fine suggestions for the second edition. For their reviews of, suggestions for, and commentsconcerning the third edition, I would like to thank Pamela Tontodonato, Kent State University;Laure Weber Brooks, University of Maryland; James A. Adamitis, the University of Dayton;William E. Thornton, Loyola University; and Malcolm D. Holmes, University of Texas at El Paso.Mike Blankenship, Sean Gabiddon, Peter Benekos, and my colleagues in the Criminal JusticeDepartment are acknowledged, as are reviewers for the fifth edition: Wanda Foglia, RowanUniversity of New Jersey; William E. Thornton, Loyola University; Obie Clayton, MorehouseResearch Institute; XXXXX XXXXX, Jersey City State College; and Art Jipsom, MiamiUniversity. Reviewers for the sixth edition included Shaun Gabbidon, Penn State-Harrisburg; RayNewman, Polk Community College; John E. Eck, University of Cincinnati; Angela West,University of Louisville; and Robert Costello, Nassau Community College. The seventh editionbenefited from the reviews of Allan Y. Jiao, Rowan University; Stephen D. Kaftan, HawkeyeCommunity College; Sudipto Roy, Indiana State University; Shaun Gabbidon, Pennsylvania StateUniversity, Harrisburg; Lisa L. Sample, University of Nebraska, Omaha; Minerva XXXXX, XXXXXHouston State University; Amir B. Marvasti, Pennsylvania State University; Altoona, Debra E.Ross, Grand Valley State University; and Elizabeth L. Grossi, University of Louisville.Appreciation is extended to the reviewers of this edition: Desire Anastasia, San Diego StateUniversity; Jason Crow, California State University, Fresno; Carlos E. Posadas, New MexicoState University; Chad Trulson, University of North Texas; and Jeffrey A. Walsh, Illinois StateUniversity. Thanks is also extended to Vidisha Barua, Pennsylvania State University, Altoona.
, for theirmany fine suggestions for the second edition. For their reviews of, suggestions for, and commentsconcerning the third edition, I would like to thank Pamela Tontodonato, Kent State University;Laure Weber Brooks, University of Maryland; James A. Adamitis, the University of Dayton;William E. Thornton, Loyola University; and Malcolm D. Holmes, University of Texas at El Paso.Mike Blankenship, Sean Gabiddon, Peter Benekos, and my colleagues in the Criminal JusticeDepartment are acknowledged, as are reviewers for the fifth edition: Wanda Foglia, RowanUniversity of New Jersey; William E. Thornton, Loyola University; Obie Clayton, MorehouseResearch Institute; XXXXX XXXXX, Jersey City State College; and Art Jipsom, MiamiUniversity. Reviewers for the sixth edition included Shaun Gabbidon, Penn State-Harrisburg; RayNewman, Polk Community College; John E. Eck, University of Cincinnati; Angela West,University of Louisville; and Robert Costello, Nassau Community College. The seventh editionbenefited from the reviews of Allan Y. Jiao, Rowan University; Stephen D. Kaftan, HawkeyeCommunity College; Sudipto Roy, Indiana State University; Shaun Gabbidon, Pennsylvania StateUniversity, Harrisburg; Lisa L. Sample, University of Nebraska, Omaha; Minerva XXXXX, XXXXXHouston State University; Amir B. Marvasti, Pennsylvania State University; Altoona, Debra E.Ross, Grand Valley State University; and Elizabeth L. Grossi, University of Louisville.Appreciation is extended to the reviewers of this edition: Desire Anastasia, San Diego StateUniversity; Jason Crow, California State University, Fresno; Carlos E. Posadas, New MexicoState University; Chad Trulson, University of North Texas; and Jeffrey A. Walsh, Illinois StateUniversity. Thanks is also extended to Vidisha Barua, Pennsylvania State University, Altoona.
many fine suggestions for the second edition. For their reviews of, suggestions for, and comments
concerning the third edition, I would like to thank Pamela Tontodonato, Kent State University;
Laure Weber Brooks, University of Maryland; James A. Adamitis, the University of Dayton;
William E. Thornton, Loyola University; and Malcolm D. Holmes, University of Texas at El Paso.
Mike Blankenship, Sean Gabiddon, Peter Benekos, and my colleagues in the Criminal Justice
Department are acknowledged, as are reviewers for the fifth edition: Wanda Foglia, Rowan
University of New Jersey; William E. Thornton, Loyola University; Obie Clayton, Morehouse
Research Institute; XXXXX XXXXX, Jersey City State College; and Art Jipsom, Miami
University. Reviewers for the sixth edition included Shaun Gabbidon, Penn State-Harrisburg; Ray
Newman, Polk Community College; John E. Eck, University of Cincinnati; Angela West,
University of Louisville; and Robert Costello, Nassau Community College. The seventh edition
benefited from the reviews of Allan Y. Jiao, Rowan University; Stephen D. Kaftan, Hawkeye
Community College; Sudipto Roy, Indiana State University; Shaun Gabbidon, Pennsylvania State
University, Harrisburg; Lisa L. Sample, University of Nebraska, Omaha; Minerva XXXXX, XXXXX
Houston State University; Amir B. Marvasti, Pennsylvania State University; Altoona, Debra E.
Ross, Grand Valley State University; and Elizabeth L. Grossi, University of Louisville.
Appreciation is extended to the reviewers of this edition: Desire Anastasia, San Diego State
University; Jason Crow, California State University, Fresno; Carlos E. Posadas, New Mexico
State University; Chad Trulson, University of North Texas; and Jeffrey A. Walsh, Illinois State
University. Thanks is also extended to Vidisha Barua, Pennsylvania State University, Altoona.
, Eighth Edition, by Frank E. Hagan. Published by Prentice Hall. Copyright © 2010 by Pearson Education, Inc.
vI also once again express my gratitude to Marie Haug and Marvin Sussman for providing my earlytraining in research. Although much of what is good about this book is due to the many finesuggestions of the reviewers, the author is solely responsible for any shortcomings.Finally, I would like to thank my wife, MaryAnn, whose continuing support, editing, dataentry, and encouragement made completion of this new edition possible. I would like to dedicatethis edition to MaryAnn. I would like to encourage students as well as faculty to contact me withany questions, comments, or suggestions via e-mail:XXX@XXXXXX.XXX.F.E.H.
I also once again express my gratitude to Marie Haug and Marvin Sussman for providing my early
training in research. Although much of what is good about this book is due to the many fine
suggestions of the reviewers, the author is solely responsible for any shortcomings.
Finally, I would like to thank my wife, MaryAnn, whose continuing support, editing, data
entry, and encouragement made completion of this new edition possible. I would like to dedicate
this edition to MaryAnn. I would like to encourage students as well as faculty to contact me with
any questions, comments, or suggestions via e-mail:XXX@XXXXXX.XXX.
Introduction to Criminal Justice Research Methods:Theory and Method 1
Theory and Method 1
Scientific Research in Criminal Justice 2
Common Sense and Nonsense 3
Why Study Research Methods in Criminal Justice? 5
The Emergence of Science and Criminal Justice 5
The Probabilistic Nature of Science 6
Proper Conduct of Critical Inquiry 7
Approaches to Theory and Method in Criminal Justice 7
The Paradigm Shift in Policing 9
Pure versus Applied Research 10
The Project on Human Development: An AcceleratedLongitudinal Design Using Nine Spaced-Age Cohorts 11
Longitudinal Design Using Nine Spaced-Age Cohorts 11
Crime Analysis: Applied Criminal Justice Research 13
Qualitative and Quantitative Research 14
Researchese: The Language of Research 15
Dependent and Independent Variables 16
Examples of the Research Process 18
Recidivism among Juvenile Offenders 18
General Steps in Empirical Research in Criminal Justice 19
Problem Formulation: Selection of Research Problem 19
Feminist Perspectives and Research Methods 20
Problem Formulation: Specification of Research
The World Wide Web (WWW) 23
Summary 26 • Key Concepts 26
Review Questions 27 • Useful Web Sites 27
Ethics in Criminal Justice Research 28
Ethical Horror Stories 28
Biomedical Examples 28
Social Science Examples 31
AIDS Research in Africa and Asia: Is It Ethical? 31
The Minerva Consortium And The HumanTerrain System 35
Terrain System 35
Researcher Fraud and Plagiarism 36
Legendary Research Scams 37
The Researcher’s Role 38
Research Targets in Criminal Justice 39
Ethics and Professionalism 40
Ethics in Criminal Justice Research 40
History of Federal Regulation of Research 41
The Belmont Report 43
Institutional Review Boards (IRBs) 44
Research Activities Exempt from HHS Review 47
National Institute of Justice’s Human Subject Protection
Confidentiality of Criminal Justice Research 49
Codes of Research Ethics of the Academy of Criminal JusticeSciences (ACJS) 50
Sciences (ACJS) 50
Ethical Issues in Criminology/Criminal Justice Research 52
Avoid Research That May Harm Respondents 52
Honor Commitments to Respondents and Respect
Exercise Objectivity and Professional Integrity in Performing
and Reporting Research 53
Protect Confidentiality and Privacy of Respondents 54
Ethical Problems 55
The Brajuha Case (Weinstein Decision) 57
The Ofshe Case 58
The Hutchinson Case 58
The Scarce Case 59
Additional Ethical Concerns 61
Avoiding Ethical Problems 61
Summary 62 • Key Concepts 63
Review Questions 63 • Useful Web Sites 63
Research Design: The Experimental Modeland Its Variations 64
and Its Variations 64
The Experimental Model 65
Research Design in a Nutshell 65
Resolution of the Causality Problem 67
Hi, I am a Moderator with Just Answer. I sent your requested Expert a message to follow up with you here, when they are back online. If I can help further, please let me know. Thank you for your continued patience.
Many apologies for the delay in reply! I'm offline on Tuesdays (My college day.)I see the source information here but don't see the question or topic of the paper or its specifics. Could you please send those?
Hope your week is better! DXJ
I remembered you were teaching on Tuesday's, so don't worry about you getting back to me until now! It's all okay.....The info you are requesting is there: Right after the read me section I sent you! However, here it is again sweetie!
Prepare a 1,050- to 1,750-word paper in which you describe the research process. Include the following: · Include new terminology learned from the reading. · How will this new terminology and knowledge apply to a career in criminal justice? · How can not knowing the proper terminology affect you as you conduct criminal justice research? · How will knowing these terms be an asset to you when evaluating and analyzing research studies or data? Format your paper consistent with APA guidelines
I will begin sending you the chapter's
Hi! Thanks for posting this separately! You're such a sweetheart!
When do you need this? We'll work it out! DXJ
H A P T E R
1 Introduction to Criminal Justice
Theory and Method
Scientific Research in Criminal Justice
Common Sense and Nonsense
Why Study Research Methods
in Criminal Justice?
The Emergence of Science and Criminal Justice
The Probabilistic Nature of Science
Proper Conduct of Critical Inquiry
Approaches to Theory and Method
in Criminal Justice
Exhibit 1.1 The Paradigm Shift in Policing
Pure versus Applied Research
Exhibit 1.2 The Project on Human
Development: An Accelerated Longitudinal
Design Using Nine Spaced-Age Cohorts
Exhibit 1.3 Crime Analysis: Applied
Criminal Justice Research
Qualitative and Quantitative Research
Researchese: The Language of Research
Dependent and Independent Variables
Examples of the Research Process
Recidivism among Juvenile Offenders
General Steps in Empirical Research
Problem Formulation: Selection of
Exhibit 1.4 Feminist Perspectives and
Problem Formulation: Specification of
Exhibit 1.5 The World Wide Web (WWW)
Useful Web Sites
Most students of criminal justice or criminology approach a course in research methods with
the enthusiasm of a recalcitrant patient in a dentist's office. Even if the experience is not
going to be painful, it most certainly is not anticipated to be exciting or interesting. Being
primarily people oriented or pragmatically oriented, the criminal justice student expects to mildly
tolerate an experience that seems quite remote from the real-world practical, everyday problems in
Research Methods in Criminal Justice and Criminology, Eighth Edition, by Frank E. Hagan. Published by Prentice Hall. Copyright © 2010 by Pearson Education, Inc
2 Chapter 1 • Introduction to Criminal Justice Research Methods
SCIENTIFIC RESEARCH IN CRIMINAL JUSTICE
Critics of many applications of scientific research to the criminal justice system view such efforts
as either elucidation of the irrelevant, obscure jargonizing, or academic intimidation-a detailed
elaboration of what any person with common sense knows. To make matters worse, all of these
efforts take place at a substantial cost, usually to the taxpayer, at a time when action-oriented
programs that really count are being cut back.
Some recent research findings are illustrative:
1. Females and the elderly fear crime because they are the most heavily victimized of all
2. Victims of crime seldom know or recognize their offenders.
3. Crime is rising by leaps and bounds and is at an all-time high.
4. The larger the city, the greater the likelihood that its residents will be victims of crime.
5. In general, residents of large cities believe that their police are doing a poor job.
6. African Americans and Hispanics are less likely than the population as a whole to report
personal crimes to the police.
7. Most residents of large cities think that their neighborhoods are not safe.
8. African Americans are overrepresented on death rows across the nation; however, this
overrepresentation is more pronounced in the South than in other regions.
9. Crime is an inevitable product of complex, populous, and industrialized societies.
10. White-collar crime is nonviolent.
11. Regulatory agencies prevent white-collar crime.
12. The insanity defense allows many dangerous offenders to escape conviction.
What could be more obvious than these findings? If anything, all that we have learned is that
Uncle Sam continues to waste tax money on useless studies. The real purpose of presenting these
findings was to make a point: Sometimes common sense is nonsense. Each of the preceding
statements is incorrect and represents a myth about crime (U.S. Department of Justice, 1978;
Wright, 1985; Pepinsky and Jesilow, 1992; Walker, 2005; and Bohm and Walker, 2006). The actual
findings were as follows:
1. Rates of victimization are higher for males than for females and for younger rather than for
2. In a large percentage of violent crimes, particularly domestic violence, victims know and
recognize their offenders.
3. Beginning in the early 1990s, crime has been declining in the United States.
4. The residents of smaller cities have higher rates than those of larger cities for certain
crimes such as assault, personal or household larceny, and residential burglary.
5. The opinions of residents of numerous cities across the nation indicate that the vast majority
is satisfied with the performance of their police; four out of five residents of the twenty-six
cities surveyed gave ratings of good or average.
6. Crimes committed against African Americans and Hispanics are just about as apt to be
reported as are crimes against victims in general.
7. Nine out of ten persons living in twenty-six large cities surveyed felt very or reasonably
safe when out alone in their neighborhoods during daytime. A majority (54 percent) felt
the same at night.
8. African American overrepresentation on death row is less pronounced in the South than in
other major regions of the country.
9. Crime is not a major concern in some developed countries, and in Japan it has actually
decreased in the post-World War II period (Clinard, 1978; Adler, 1983).
10. Unsafe work conditions and the marketing of unsafe products kill and maim more
Americans each year than street muggers and assailants (Hills, 1987).
11. Regulatory agencies have been understaffed, underfinanced, and inadequate in controlling
white-collar crime (Clinard and Yeager, 1979).
12. Despite media attention, insanity defense cases are rare; successful ones are even more
rare (Morris, 1987).
COMMON SENSE AND NONSENSE
Nonsense about Crime and Drugs (Walker, 2005), Myths That Cause Crime (Pepinsky and
and Justice (Kappeler and Potter, 2005).
What was assumed to be obvious in our example appeared so after the results were
presented. If people agreed with the findings, they considered them obvious, whereas if they
disagreed, they viewed them as unscientific because common sense told them so. Hirschi and
Stark (1969) in "Hellfire and Delinquency" found a very weak relationship between church
attendance and nondelinquency. In speaking to a "damned if you do and damned if you don't"
phenomenon, they indicate that had they found a strong relationship as common sense would
have suggested, they would have been accused of wasting time on the obvious. Because
their study countered common sense, it was attacked as false, stupid, or an illustration of
Brown and Curtis (1987, p. 3) indicate that:
Many practitioners within criminal justice have met with repeated failure over the
years because they relied upon only their common sense. Thus, millions of dollars
have been spent on police patrol efforts that do not reduce crime, judicial practices
that are widely perceived as unfair, rehabilitation programs that do not rehabilitate
offenders and countless other failures.
In The Natural History of Nonsense (1958), Evans outlines numerous examples of
commonsense-nonsense issues that have hindered human progress. The beliefs that the earth
was flat and that it was the astrological center of the universe are but two of these examples. In
testimony before Congress, Herbert Simon, protesting threatened drastic cuts in funding for
social science research and in response to attacks on the social sciences as employing obscure
jargon and explicating the obvious or common sense, asserted that the common sense of the
social sciences is not a mirror of society but a result of research:
The social sciences are often discounted because much of what they learn seems to
be common sense. Well, it is common sense today to say that if you drop a feather
and a rock together in a vacuum, they will fall at the same pace. It was not common
sense before Galileo . . . [O]ne of the basic aims of the social sciences must be to
take knowledge that comes out of the laboratory-knowledge that may be stated in
language that is hard to understand-and make it part of the common sense of our
society. (Prewitt and Sills, 1981, p. 6)
4 Chapter 1 • Introduction to Criminal Justice Research Methods
Many of the terms used in everyday conversation originated in social science research;
however, little credit is given for these theoretical accomplishments because the discoveries,
once labeled, were quickly absorbed into conventional wisdom.
In an essay celebrating twenty-five years of criminal justice research sponsored by the
National Institute of Justice (the research arm of the U.S. Department of Justice), Blumstein and
Petersilia (1994, p. 36) state:
When compared to the public debate about these issues, which is still often focused
on ideological issues and simplistic solutions to complex problems, it is clear that the
thinking resulting from research, backed up by strong evidence, is very sophisticated,
but it still has a long way to go to become an important part of the public debate.
Researchers must also remember that the revolution in thinking brought about
by Galileo and Copernicus took far longer than twenty-five years to become widely
diffused to European thinking.
The most interesting phenomenon about those who propose a commonsensical approach
to criminal justice is that the debunker often simply substitutes his or her own subjective biases
and experience for the more scientific approach found lacking. Wilkins addresses those who
argue the antiscientific view (wherein each individual is unique and defies measurement and
The objection states that prediction is useless (or dangerous?) because the individual
is unique. Prediction is said to be either (or both) impossible or undesirable,
and this argument rests on the complexity of human relationship . . . If the case is
unique, what experience can the clinician use to guide him? . . . Statistical experience
can be based on samples of the population which we know to be unbiased.
A clinician has only his own sample to guide him with no guarantee of its lack of
bias. . . . If these features are "intangible," how can we know that they exist? How
in fact does the clinician take them into account? Can they not be described in
words? If not, are they more than the prejudices of the observer? (Wilkins, 1978,
As we will see in Chapter 9, verbal descriptions of phenomena are not much different than
numerical measures of some entity; the latter simply force the analyst to be more precise and
rigorous, while thinking through the concept under study in a more disciplined manner. Thus,
common sense and experience certainly serve important functions in sensitizing us to a subject.
However, our separate experiences might better be viewed as limited case studies of a subject
matter that may not be entirely generalizable to the universe of such subjects or as observations
that may be limited by time, place, and the subjective biases of the observer.
In addition to common sense as a rival explanation to science, there are other competing
sources of knowledge. Authorities or experts are sources, but sometimes they are simply
wrong. Tradition (past authorities) may also mislead us. Currently, the media represent a very
powerful source of information on crime, but their portrayals may not reflect reality and
often aim to entertain or gain revenue rather than to inform. Related to the commonsense
approach to research, but of vital importance in criminal justice, are the questions "So what?"
and "Of what practical use are these findings?" These subjects will be discussed in detail in
WHY STUDY RESEARCH METHODS IN CRIMINAL JUSTICE?
Rather than viewing certain elementary research concepts and procedures in scientific methodology
as foreign elements, the criminal justice professional may, once he or she has mastered them,
discover very valuable tools for assessing current and future directions in the field.
Once familiar with these tools, much of the anxious sanctimony bestowed upon technical
reports, academic concepts, and research findings can be dispensed with. Many readers of this text
have a healthy cynicism or critical and suspicious approach to research findings and probably
know and employ more about research methodology than they are aware. In most cases, they
simply lack conceptual frameworks, scientifically acceptable tags, or sufficient knowledge of the
language of research methods to defend their views in an appropriate manner. Research methods
provide the tools necessary to approach issues in criminal justice from a more rigorous standpoint
and enable a venture beyond opinions based solely on nonscientific observations and experiences
(see Black, 1993).
Although many readers may never undertake their own research, all will be consumers or
recipients of findings and policies based upon research. It is not unusual to find students as well as
professionals in criminal justice who are unable to fully understand reports and journal articles in
their own field. Other fields may have this same problem: however, one might certainly be wary of
a surgeon who is performing an operation without an understanding of the latest article on the
procedure in the Journal of the American Medical Association. Similarly, in striving for
professional status, it is imperative that criminologists and criminal justice professionals
comprehend and critically evaluate new developments in their field (Hagan, 1975). Mastery of this
material will assist in this endeavor. A very interesting outcome upon completion of the material is
that many find themselves carefully reading and interpreting the tables presented in studies and
skipping much of the prose. This procedure, which reverses the usual pattern at the beginning of the
course, results in a great economy of time and effort.
An analogy can be drawn between learning research methods and studying to become
a movie director or critic (Tontodonato and Hagan, 1998). A movie director or critic cannot
simply view a film and report, "You just have to see it." They must be more concerned with
technique. What is the plot? Who are the heroes/heroines/villains? From what point of view is
the story told: first person or third person? What was the denouement? Research methodologists
are also interested in ransacking studies and breaking them down into essentials. What are the
research design and hypothesis? What data-gathering procedures were employed? What
were the independent and dependent variables? What type of data analysis and conclusions were
made regarding the null hypotheses? Upon completing this book, it is hoped that the student will
become prolific in being an active consumer of research.
Much of what appears in this text as research methods in criminology and criminal justice
is not unique to criminal justice but is borrowed from the other social sciences and applied to
criminal justice topics and examples. The techniques are applicable to a wide variety of areas and
in that sense are excellent broad-based skills.
THE EMERGENCE OF SCIENCE AND CRIMINAL JUSTICE
Humankind's long stride toward understanding and explaining the universe might be viewed as
a marathon that has accelerated into a 100-yard dash in the past two centuries. In an attempt
to provide a stage for our purposes and avoid philosophical discourses which could occupy far
too much time for our purposes, this development can be described succinctly. Human beings,
6 Chapter 1 • Introduction to Criminal Justice Research Methods
through the creation of symbols or abstractions such as language, are able to develop knowledge.
Knowledge is what people create symbolically to represent reality.
Knowledge might also be viewed as our presently accepted level of ignorance. There is no
guarantee that what one historical generation considers to be a wise policy or procedure will not
be considered ignorant by the next generation. Less than 100 years ago, "feebleminded" persons
were locked up permanently for the protection of society because "feeblemindedness causes
crime" (McCaghy, 1999, p. 10). Not long before that, the physically ill were bled to "cure" their
illness. At present, these methods are considered ignorant and even detrimental. It is important to
recognize the changes in methods between the present day and 100 years ago.
The early French sociologist Auguste Comte described the "progression of knowledge as
being one from predominantly theological or supernatural explanations of reality to metaphysical
or philosophical ones and finally to scientific approaches" (Comte, 1877). Rather than looking to
otherworldly explanations, philosophy sought explanations in worldly events through a new spirit
of inquiry-rationality and logical explanation. Science combined this spirit of rational explanation
with method-empiricism, experimentation, or what has come to be called the scientific method.
The scientific orientation emphasized observation, measurement, replication (repetition of observation),
and verification (checking on the validity of observations). Science subjects ideas or theories
to tests through observation, quantification, and empirical analysis. Unlike the philosopher's often
sole reliance on logic and argumentative reasoning, all scientists can be said to be from Missouri,
the "Show Me" state. Replication is the repetition of experiments or studies utilizing the same
methodology. Verification is confirmation of the accuracy of findings or attainment of greater
certitude in conclusions through additional observations.
Systematic application of the scientific method to research problems provided major breakthroughs
in the development of knowledge. While the scientific era has enabled mankind to take giant
strides in explaining and harnessing physical reality, more recently the social sciences, of which
criminal justice is a progeny, have attempted to apply these same procedures to their subject matters-
society, human behavior, politics, or, of concern in our enterprise, crime and criminal justice.
Because we have gained mastery over more simple and more controllable physical reality,
why, as intelligent humans, is it not possible to gain the same explanation, prediction, and control
over such recurring social phenomena as crime, violence, and their many ramifications?
Criminology and criminal justice as interdisciplinary disciplines draw upon many fields,
both academic and applied, and, in terms of appropriate research methodology, have not only
borrowed and adapted but have made many of their own contributions as will be presented
throughout this book. Whether criminal justice is a science is argued even by practitioners within
the field. Criminal justice researchers, unlike physical scientists, find their subject matter a topic
of popular discussion in which the layperson's experience is viewed as just as good a guide to
policy as that of the researcher. The same people who would not dream of arguing about
molecules, atomic weights, or quasars feel quite qualified to address issues of crime and
punishment. In this writer's view, the appropriate employment of scientific methodology and
procedure qualifies a discipline to claim scientific status. Although this topic is certainly worthy
of debate elsewhere, the approach throughout this text is a very pragmatic one-we are employing
scientific procedures and therefore contributing to the development of a young science.
THE PROBABILISTIC NATURE OF SCIENCE
As social scientists, criminologists and criminal justice researchers assume that the subject matter
they study is probabilistic-that is, they believe that effects will most often occur when certain
causes are present, but not in every single case. In predicting general patterns, trends, and
Chapter 1 • Introduction to Criminal Justice Research Methods 7
relationships among groups, social scientists do not expect these patterns to hold in each individual
case or do not expect absolute determinism. If researchers show a relationship between rising
unemployment rates and suicide rates, they are not assumed or considered to be wrong if the newly
unemployed do not commit suicide. Researchers do attempt to estimate the probability of their
predictions being accurate. For example, there is an 85 percent chance that a savings and loan thief
such as Charles Keating will be white, upper-middle class, and a college graduate.
PROPER CONDUCT OF CRITICAL INQUIRY
Bayley (1978) offers three suggestions for improving criminal justice research:
1. Research requires interdisciplinary efforts as well as the tackling of field-oriented, practical
2. Researchers should cease giving speeches to practitioners about the value of research and
attack their practical concerns with a realistic appraisal of error proneness of any research
3. It is time to be done with "methodological narcissism," methods for methods' sake.
The latter is aptly illustrated by the well-known Martinson Report, which was published
in the early 1970s. Martinson (1974) raised quite a storm in the field of corrections when, based
on his evaluation of rehabilitation programs throughout the United States, he concluded that
none of these programs reduced recidivism or rehabilitated clients. Later, in retracting his own
previously devastating "almost nothing works" critique of corrections research, Martinson
(1978, p. 4) advised that it was time to avoid "methodological fanaticism" (what Bayley called
"methodological narcissism"), in which substance is overlooked in the name of method.
Preferred rigor in research design is seldom realized in criminal justice field studies. This does
not justify throwing out "the baby with the bath water." Instead, such problems represent
challenges to the criminal justice researcher, rather than a justification for self-defeating
pessimism and methodological capitulation.
Hirschi and Selvin give sound advice to those either doing research or criticizing the research
of others. The proper conduct of critical inquiry requires that "those concerned with good research
should be objective and vigilant as well as sympathetic" (Hirschi and Selvin, 1973, pp. 273-274).
Objectivity entails value neutrality or a dispassionate approach to the subject matter that holds
constant personal bias (Weber, 1949). Vigilance involves a concern for accuracy and efforts to
eliminate error. Error, however, is ever present in research. The only perfect research is no research,
which suggests the last point: empathy. In critiquing the research of others, empathy or a willingness
making errors in research, then he or she probably should do none, because error is omnipresent.
In criticizing other research, one often need not go far to find some error. The question is not
whether errors are present, but, rather, whether reasonable attempts were made to acknowledge
and/or eliminate the most obvious errors. Only when such errors so grossly compromise the
accuracy of findings and conclusions thereof, should one scathingly attack other research.
APPROACHES TO THEORY AND METHOD IN CRIMINAL JUSTICE
Theory in criminal justice represents an attempt to develop plausible explanations of reality,
which in this case is crime and the criminal justice system. Theory attempts to classify and
organize events, to explain the causes of events, to predict the direction of future events, and
to understand why and how these events occur (Turner, 1997, p. 2). It represents a reasonable
8 Chapter 1 • Introduction to Criminal Justice Research Methods
and informed guess as to why things are as they appear and to explain their underlying nature
and meaning. Kraska (2006, pp.167-168) explains:
It [theory] defines the parameters for how we think about our objects of study, and
provides us with the lenses through which we filter our subject matter in order to
make sense of complex phenomena. It gives us our organizing concepts, frames our
research questions, guides our scholarly interpretations, and is an unavoidable
presence in crime control policy, practice and decision-making.
Much criminological theory possesses a global or sensitizing quality that alerts us to critical
issues but often lacks the quality of formally testable, scientifically verifiable propositions.
Without the generation of useful theoretical explanations, a field is intellectually bankrupt; it
becomes merely a collection of war stories and carefully documented encyclopedic accounts. It
fails to explain, summarize, or capture the essential nature of its subject matter. Theory asks: What
is the point of all of this? What does it mean? Why are things this way? Willis (1983), in a review
of twenty-five criminal justice textbooks, noted almost a "trained incapacity" or unwillingness to
deal with theoretical issues and a tendency to concentrate on what fictional character Joe Friday,
in the old television series Dragnet, called "just the facts."
In The Structure of Scientific Revolutions, Thomas Kuhn (1970) describes how the evolution
of new knowledge, rather than being slow and incremental, is often dependent upon new paradigms
that may stand previous assumptions on their heads. A paradigm is "some implicit body of
intertwined theoretical and methodological belief that permits selection, evaluation, and criticism"
(ibid., pp. 16-17). A new paradigm represents a new model or revolutionary schema with which to
view reality. Paradigms organize reality by giving structure, framework, and perspective from
which to investigate reality. Kuhn viewed scientific knowledge as achieving breakthroughs by
revolutions fueled by the inability to explain anomalies by means of the existing scientific tradition.
The old paradigm (normal science) is made obsolete by the new paradigm through this process of
revolution. In a related phenomenon, sometimes new knowledge takes place or is discovered by
a process of serendipity, a wholly unanticipated and surprising discovery. In the process of doing
research on one topic, a discovery is made that addresses some different topic. Copernicus and
Galileo put astronomy upside down with their discovery that the earth was not the center of the
universe. In archeology, the pre-Clovis paradigm challenged the prevailing estimate of human
habitation in the Americas. Exhibit 1.1 describes the paradigm shift in policing.
Edwin Sutherland's concept of "white-collar crime" (1940) serves as an example of a
paradigm revolution in criminology, a radical reorientation in theoretical views of the nature of
criminality. After Sutherland, crime was no longer viewed solely as an activity of the underclass.
Copernicus' astronomical theory of the universe made totally irrelevant the previous paradigm of
astrology. Walker 6th ed. (2005, p. 76) indicates:
Finally, Kuhn's perspective cautions adherents of the prevailing paradigm to recognize
that there is nothing permanent or timeless about their viewpoint. This paradigm, which
organizes their thinking, research, and policy proposals, was the product of a scientific
revolution that replaced an earlier paradigm. Science, however, like time, marches on.
We cannot predict what kind of scientific revolution lies in the future and will overthrow
the assumptions shared by virtually everyone reading this article.
Methodology (methods), on the other hand, involves the collection of accurate facts
addresses the issue "why," methodology concerns itself with "what is." There usually exists in
any field a certain division between those who are primarily interested in generating theory and
who view their efforts as classical scholarship akin to philosophy and those (methodologists)
who are viewed as technical and scientific in their approach. In speaking to criminology's
The Paradigm Shift in Policing
Blankenship and Brown (1993) claim that the
criminological literature typically refers to competing
theories as paradigms. A paradigm shift of sorts
has occurred in American policing beginning with
the "Kansas City Preventive Patrol" experiment
(Kelling et al., 1998) which will be discussed in detail
later in Chapter 3. Conducted in the early seventies,
police administrators were at first astounded to find
that neither increases nor decreases in police patrol
seemed to affect the crime rate, public fear of crime,
or public satisfaction with the police. Rapid response
to calls for service and preventive patrol had been to
inviolate doctrines until then. Rather than viewing
such findings as negative, police executives
discovered they had a lot more discretion than they
had anticipated in deploying their forces. Under the
"rapid response to service paradigm" police spent
most of their time rushing to the scene of calls for
service despite the fact that most people waited half
an hour after an event was discovered to report the
crime to the police. Preventive patrol seldom came
upon crimes in progress. Being busy with patrol and
answering calls, the police had little time to
investigate trivial matters.
Windows," James Q. Wilson and George Kelling
(1982) proposed a radically different paradigm.
Kelling et al. (1998) indicates:
Just as unrepaired broken windows can
signal to people that nobody cares about a
building and lead to more serious vandalism,
untended property, disorderly persons, drunks,
obstreperous youth, etc.-both create fear in
citizens and attract predators.
Neighborhood disregard, drunks, panhandlers,
youth gangs, and other "trivial" incivilities unsettle
a community and produce fear and disrupt the
life of a neighborhood. The large decrease in
crime in New York City in the nineties was attributed
in part to a new policing paradigm emphasizing
community policing and zero tolerance for previously
ignored aggressive panhandlers, subway turnstile
hoppers, vagrants, and those involved in disorderly
conduct. Enforcing the previously ignored small things
gave police a better handle on crime (Kelling and
Discussion of the "Crime Dip" later in Chapter 4
will examine other reasons for the decline in crime.
Broken Windows Policing (2001) and Taylor in
Breaking Away from Broken Windows (2001) question
the "Broken Windows Paradigm." They point out that
crime had already begun to decline in New York before
Chief William Bratton introduced the broken windows
policy and declined just as drastically in cities such as
San Diego that never used the strategy. Similarly,
Behind the Crime Crash of the 1990s (2001) credited
the end of crack dealer turf wars for much of the
decline (see Miller, 2001).
Revolutions. 2nd ed. Chicago: University of Chicago
Press, 1970. Kelling, George A. "Police and Communities:
The Quiet Revolution." Perspectives in Policing. National
Institute of Justice, June 1998. XXXXX, XXXXX Q. and
George L. Kelling. "Broken Windows: The Police and
Neighborhood Safety." The Atlantic Monthly (March,
10 Chapter 1 • Introduction to Criminal Justice Research Methods
fascination if not obsession with preferred statistical fads and methodological tools and
techniques at the expense of theory, Sampson and Laub (2005, p. 911) indicate:
The bigger problem is that criminology seems obsessed with tools rather than
keeping its eye on the prize. Indeed we worry less about groups being reified than the
methods themselves; when methods rule we focus on the fundamental processes
that explain crime and its persistence and cessation over the life course. For us, the
botXXXXX XXXXXne of sound research design and basic scientific inquiry is that methods
are inextricably linked to, and the servant of theory.
Good criminal justice requires both. Theory devoid of method-explanation without
accurate supportive data-is just as much a ritualistic dead end as method devoid of theory.
The former resembles armchair theorizing, while the latter resembles a fruitless bookkeeping
operation. Both theory and method should be viewed as a means to an end, the end being
sound criminal justice knowledge.
PURE VERSUS APPLIED RESEARCH
Pure (basic) research is concerned with the acquisition of new knowledge for the sake of science
or the development of the field, whereas applied research is practical research concerned with
solving immediate policy problems. Although we addressed the issue of common sense briefly by
means of "myths of crime," there still exists the broader issue: Criminal justice has experienced
conflict between two camps, the applied practitioner and the nonapplied academic. Although this
division is in part stereotypical, as mutual exclusivity is not in fact the case with these groups, for
heuristic purposes we consider these as "ideal types."1 Being on the front lines of the criminal
justice system, practitioners are most interested in applied research, studies, and findings that
research, which may have no immediate applicability but contributes to the knowledge base and
scientific development of the discipline. Although the practitioner may view the pure scientific
researcher as off in a closeted ivory tower or a likely candidate for former Senator Proxmire's
Golden Fleece Award for irrelevant research,2 the pure scientific researcher may view many of the
policy recommendations of applied research as shamanism, or quackery, an attempt to give advice
or guide policy without adequate theoretical or methodological support. In speaking to the issue of
premature application, Friedman (1980) puts it succinctly: "If you eat the cookies before they are
ready you can get sick."
In reality, neither pure nor applied research fits these neat stereotypical views (Rabow,
1964). Some of the most apparently obscure and abstract research projects may produce the
critical discoveries that in the long run produce more applied payoffs than hundreds of premature
applied projects. On the other hand, many existing projects require informed decisions, which,
although not perfect or entirely supported by research findings, represent the best we have to
offer at the time. A study of state correctional agency practitioners (Light and Newman, 1992)
1Weber (1949) viewed "ideal types" as useful analytic devices that extract pure or overgeneralized elements of a reality, but
seldom exist in pure form. Wilson's (1968) "Watchman," "Service," and "Legalistic" styles of policing are an illustration.
2 In the late 1970s, the late Senator William Proxmire attracted considerable publicity by presenting awards to governmentsponsored
research projects with esoteric titles that he viewed as irrelevant. Proxmire was the subject of an $8 million libel
and slander suit by experimental psychologist Dr. Ronald R. Hutchinson as a result of his being identified as a recipient. See
"Golden Fleece Suit Reaches Supreme Court," Footnotes, American Sociological Association 7 (May 1979): 5.
Chapter 1 • Introduction to Criminal Justice Research Methods 11
found that while such practitioners strongly supported social science research, they reported
using it very little in comparison with other types of information and knowledge.
Many of the major and important scientific discoveries of ancient and modern times were
made not by the kings' wizards commissioned to immediately perform alchemy or other applied
magic, but by abstract "tinkers"-Galileo, Copernicus, Einstein, and Pasteur. What is regarded as
renegade, pure, and ivory-tower research of one epoch often finds itself the basis for important
applied breakthroughs in the next. As mentioned previously, in the late 1970s, in an attempt to point
out wasteful government funding of "irrelevant" research projects, Senator Proxmire periodically
would publicize and give his infamous Golden Fleece Award. To the layperson, obscure-sounding
studies such as "The Sex Life and Mating Habits of Bees in the Upper Amazon Basin" appear to be
projects deserving of derision and attack until at a later date scientists speculate that "killer bees" that
are resistant to existing chemical insecticides may invade North America and result in human deaths.
An excellent example of a basic research project is the Project on Human Development in
Chicago Neighborhoods (Visher, 1994) (see Exhibit 1.2).
This study represents the efforts of over 100 scientists representing the fields of
pediatrics, biology, psychology, sociology, and criminology (NIJ, 1997). Begun in 1989, the
The Project on Human Development: An Accelerated Longitudinal Design Using
Nine Spaced-Age Cohorts
researchers headed by Albert Reiss, Jr. (Yale) and Felton
Earls (Harvard) conducted the study and examined
everything from gestation, infancy, and childhood to
adulthood to age thirty-two. Such a prospective design
would usually have taken thirty-two years; but the
unique feature of the research design, an accelerated
Being the largest longitudinal study ever undertaken in
criminal justice/criminology, NIJ's Project of Human
Development in Chicago Neighborhoods was unprecedented
in scope. It examined a broad range of factors
at the community, family, and individual level that were
believed predictors of crime and deviance. A team of
Subject Age Birth
Age at Enrollment in Study
500 Males Retrospective Data
5 10 15
20 25 30
FIGURE 1 Project on Human Development, Accelerated Longitudinal Design 2002-Completion
of Study. (Source: Visher, 1994, p. 13.)
Research Methods in Criminal Justice and Criminology, Eighth Edition, by Frank E. Hagan. Published by Prentice Hall. Copyright © 2010 by Pearson Education, Inc.
12 Chapter 1 • Introduction to Criminal Justice Research Methods
FIGURE 2 Project on Human Development: The Contents and Factors to Be Studied. (Source: Visher,
1994, p. 14.)
longitudinal design depicted in Figure 1, sped up the
Between 1994 and 2002, the investigators
gathered detailed data about 11,000 individuals and
their communities. A preliminary five-year planning
phase and the accelerated longitudinal design
provided data much more quickly than the standard
longitudinal design. Overlapping age cohorts were
studied, each with five-year overlap periods, thus
mimicking a thirty-two year longitudinal study. Figure 2
describes the key variable examined in the research.
Data collection began in Chicago in August
1994. During the past three years, exploratory studies
of topics and methods took place and research
protocols were pretested. Subjects and their families
were interviewed across the nine age cohorts.
Support from a variety of community organizations
was solicited particularly in investigating mental
health, child development, and substance abuse. It
was hoped that the multidisciplinary approach
integrating community, family, school, peer, and
individual characteristics would do much to advance
our understanding of crime causation and in
developing future crime control policy.
Despite the fact that crime is considered one of
the major U.S. social problems, federal funding for
criminal justice research has been meager. For example,
a 1988 study by the National Science Foundation (NSF)
found that per capita expenditures for research
in health ($432), energy ($11), space ($19), transportation
($4), environment ($4), agriculture ($3),
education ($2), and commerce ($0.53) greatly
exceeded the criminal justice research funding of $0.13
(NSF, 1988). Furthermore, the domestic research
commitment of $21 million for NIJ in 1989 ($100.6
million in 1997) was substantially below the
$229 million for the National Eye Institute, the
$167 million for the National Endowment for the Arts,
the $140 million for the National Endowment for the
Humanities, and the $127 million for the National
Strengths and weaknesses
of the school environment
Composition and size
of social network
Substance abuse and
delinquency by peers
Deviant and prosocial
attitudes of peers
Location of peer networks
(school or community)
Changes in peer relationships
Family mental health
Family history of criminal
behavior and substance
Physical and mental health
Impulse control and
Cognitive and language
Ethnic identity and
Social, economic, and
Informal social control
Level of involvement
in drug and gang
School policies regarding
EXHIBIT 1.2 (Continued)
Chapter 1 • Introduction to Criminal Justice Research Methods 13
project involved an ongoing, overlapping, longitudinal (studies over time) analysis of 8,000
residents of 343 Chicago neighborhoods from birth to age thirty-two. The subjects were
tracked for several years while examining their development in an effort to determine
how family and neighborhood factors influence criminal behavior. Exhibit 1.3 provides an
example of applied research-crime analysis. Examples of applied research will be featured in
Crime Analysis: Applied Criminal Justice Research
Institute for Dental Research (Petersilia, 1991).
Despite the limited funding provided for criminological
research, the potential of the research to aid in public
policy decisions is clearly shown by the work that has
been completed (Sherman et al., 1997).
Sources: Visher, Christy A. "Understanding the Roots of
Crime: The Project on Human Development in Chicago
Neighborhoods." National Institute of Justice Journal.
November 1994: 9-15; Interim reports can be
obtained from: The Project on Human Development
in Chicago Neighborhoods, Harvard School of Public
Health, Department of Maternal and Child Health,
677 Huntington Avenue, Boston, MA 02115. The
project has its own updated Web site: http://phdcn/
Steven Gottlieb, Sheldon Arenberg, and Raj Singh's
Crime Analysis (1994) is the classic work in the
burgeoning applied research field of crime analysis.
They make the distinction between four types of
law enforcement analysis (ibid., pp. 11-12):
Crime analysis allows the analyst to
determine who's doing what to whom by its
focus on crimes against persons and property
(homicide, rape, robbery, burglary, theft, etc.).
Crime analysis can be defined as systematic,
analytic processes aimed at providing practical
information related to crime patterns. It often
involves providing trend correlations to assist
in management and in the solving and prevention
of crime. The functions of crime analysis
include (ibid., pp. 15-16):
Identification of crime patterns
Target profile analysis
Provision of investigative leads
Provision of support data to community
policing and crime prevention programs
Assistance in case clearance
Support for departmental planning activities
Analysis of operational data for departmental
The growing use of crime analysis by police
departments and federal agencies will be featured
in later exhibits in this book.
Intelligence analysis aids the determination
of who's doing what with whom by its
focus on the relationships between persons and
organizations involved in illegal-and usually
conspiratorial-activities (narcotics trafficking,
prostitution rings, organized crime, gangs,
Operations analysis enables the analyst to
ascertain how the agency is using its internal
resources by its focus on the examination of
personnel deployment and workload distribution
Investigative analysis is an exceedingly
specialized type of analysis that is frequently used in
the investigation of unusual or serial homicide cases.
This form of analysis uses crime scene evidence
and information regarding the background of
victims to develop physical, behavioral, and psychological
profiles of the suspect(s) responsible for the
Source: Gottlieb, Steven, Sheldon Arenberg, and Raj Singh.
Crime Analysis: From First Report to Final Arrest. Montclair,
CA: Alpha Publishing, 1994.
EXHIBIT 1.2 (Continued )
14 Chapter 1 • Introduction to Criminal Justice Research Methods
subsequent chapters in this book, including crime profiling, crime analysis, crime mapping,
and statement analysis, as well as many other studies that are specifically undertaken in order
to address immediate policy needs.
Evaluation research, which will be discussed in detail in Chapter 11, is a branch of applied
research that examines public programs and policies. Do programs work? How well do they
work? How can they be made to work better? These are some of the questions asked in evaluation
The 1967 report of the President's Commission on Law Enforcement and Administration
of Justice called for creating a research program in criminal justice at the federal level. In 1968,
the National Institute of Law Enforcement and Criminal Justice, now called National Institute
of Justice (NIJ), was created. Prior to this, the U.S. Department of Justice had no
research and development program. Doris MacKenzie (1998, p. 3) explains the inadequacy of
When this nation wants to win wars, excel in the race to conquer space, or improve
conditions for our rural populations, we put money into research. Perhaps one of the
largest reasons criminal justice research has not reached the achievements of other
fields is the limited funding given to research activities.
QUALITATIVE AND QUANTITATIVE RESEARCH
research, concepts are viewed as sensitizing ideas or terms that enhance our understanding.
Research methods in the social sciences, of which criminal justice is heir, have followed two
basic philosophical traditions. The first legacy reflects a historical, intuitive, or observational
approach and suggests that the physical and social sciences are distinct entities.3 It emphasizes a
qualitative approach to understanding the reality under investigation. Classic sociologist Weber
described it as a Verstehen (in German, understanding or empathy), in which researchers hope
to immerse themselves in the subject matter and develop "sensitizing concepts" that enhance
their understanding and explanation of reality (Weber, 1949). Many field studies and participant
observation studies, in which the researcher lives with and experiences a group's way of life from
the group's perspective, serve as examples. This grounded theory approach enables a shifting of
gears to focus upon issues that were not previously assumed to be of importance at the time of
the beginning of the project.
Positivism, a natural science approach, is often used to describe the second legacy. This
empirical orientation suggests that the same approach applicable to studying and explaining
approach and is concerned with measuring social or, in our case, criminal justice reality. While
qualitative research emphasizes a verstehen approach in which researchers immerse themselves
in the subject matter and develop empathetic understanding, the quantitative approach favors
studying "phenomena that can be measured, observed, ‘objectified' and examined empirically"
(Worrall, 2000, p. 359). Although both legacies as pure ideal types may represent dead ends
in criminal justice research, moderate expressions of either of these strategies have, as we will
discover in Chapter 3, a role in enhancing our understanding of criminal justice. An extreme
3 The qualitative approach is illustrated by such writers as Weber (1949), Garfinkel (1967), and Blumer (1969) and by
groups that advocate "symbolic interactionism" or "ethnomethodology."
Chapter 1 • Introduction to Criminal Justice Research Methods 15
qualitative approach would provide historicism-seeing all social events as a distinct chronicle
of unique happenings. This would involve a denial that any scientific generalities could be drawn
from the world of human events. Such a stance is antiscientific. On the other polar extreme is
scientism or extreme positivism, in which the researcher takes the stance that "if you cannot
measure it, it is not worth studying or commenting on." Although historicism may jade those
from the traditional humanities in their view of criminal justice as a discipline, scientism is
most often the orientation of physical scientists, who may view fields such as criminal justice as
a pretender to the scientific throne or one of a score of "Johnny-come-lately" pseudosciences.
Such physical scientists may feel that unless criminal justice or social science researchers can
attain the same rigorous control over the conditions of study as in the physical sciences, they
are somehow involved in an inferior enterprise, an amateur imitation of real science. Obviously,
social phenomena cannot be put in a test tube or maze.
Criminal justice as an emergent, interdisciplinary, applied scientific field requires for its
mature development a full array of qualitative and quantitative approaches, pure and applied
research efforts, and theoretically incisive as well as methodologically sound studies and evaluations
to gain the academic respectability it both aspires to and deserves.
RESEARCHESE: THE LANGUAGE OF RESEARCH
Sprechen Sie Researchese? To the uninitiated, the language of research is almost like being
exposed to a foreign language. How often have we heard a frustrated reader of a report say, in
despair and disgust: "Why don't these people write in English?" This common reaction might
even be described, to coin a phrase, as research shock-a sense of disorientation experienced by
a person when suddenly confronted with an unfamiliar style of presentation and research
language. What one is reacting to is not complexity or the unlearnable, but merely the unfamiliar.
Those of you who are in-service, criminal justice professionals or who have taken only a few
courses in criminology or criminal justice, soon discover that you have accumulated much of the
specialized language of the field. When you use this argot around others not in the field you may
be surprised that others are unaware of these terms. This, incidentally, explains why many
occupational groups cling together socially because others cannot fully appreciate their jargon or
ideology. It also explains why many spouses at parties request in despair, "Please stop talking
shop!" So, even though at first much of the terminology seems clumsy, stick with it, and by the
completion of this text, you too will be able to read, write, and think researchese (the language
of research). Researchese is a valuable international language and a useful tool for negotiating
and understanding the latest literature in your field.
The notion of causality, a complex subject in philosophy, is the very essence of scientific
inquiry. Science assumes that elements of reality can be isolated, defined, explained, and predicted-
that science holds the key to unlocking the mysteries of the ages. Scientific investigation assumes
that causal principles and laws underlie reality and that by discovering these elements, science can
predict and control reality (Wilkins, 1976). This process begins by naming things.
Concepts are abstract tags put on reality and are the beginning point in all scientific endeavors. Not
to be confused with reality itself, concepts are symbolic human creations or constructs that attempt
to capture the essence of reality. In deciding on a name for some phenomenon, we are attempting to
describe, understand, classify, or become more sensitized to some element of reality. Examples
of concepts that are used in criminological and criminal justice studies include crime, recidivism,
16 Chapter 1 • Introduction to Criminal Justice Research Methods
cynicism, intelligence, risk on parole, defendant's appearance, and police patrol. Age, sex, race,
religion, and social class are other concepts with which we are quite familiar. Concepts may be
viewed as qualitative, sensitizing/global notions, or they can be converted into variables through
Operationalization defines concepts by describing how they will be measured. Working definition
or operational definition are other terms used to refer to this process. The notion of operationalization
can be defined in response to the statement: "I measured it by." Completion of this sentence
constitutes the operationalization of the concept. This process of operationalization has now
quantified (assigned numerical values to) a concept and converted it from an abstract, verbal entity
to a measurable quantity or variable.
Variables are concepts that have been operationalized or concepts that can vary or take on
different values of a quantitative nature. They are the mortar and brick of scientific investigation.
Theoretically, variables can be of a qualitative nature. For example, qualitative distinctions could
be made regarding a person's age (old or young), but the measurement of actual chronological
age would be considered more exact. Crime may be operationalized in a study as having been
measured by official police statistics, surveys of victims, or self-admission reports. Different
measures may yield slightly different pictures and therefore should be defined. Similarly, recidivism
may be defined by means of rearrest rates, reincarceration (imprisonment or jail) rates, or
other measures that could produce quite different assessments of the success or failure of
programs. Table 1.1 provides some illustrations of the conversion of various concepts into
variables by means of operationalization.
The dependent (outcome) variable is the variable one is attempting to predict and by convention is
denoted by the letter Y. Common outcome variables in criminal justice are concepts such as crime or
recidivism. Table 1.1 illustrates that ordinarily the dependent variable is some behavior or attitude
that is usually the subject of one's study. The independent (or predictor) variable is the variable that
causes, determines, or precedes in time the dependent variable and is usually denoted by the letter X
(or any letter other than Y). An independent variable in one study may become a dependent
variable in another. For example, a study of the impact of poverty (X) upon crime (Y) finds poverty
as a predictor (independent) variable, whereas a study that looks at race (X) as a predictor of
poverty (Y) finds poverty as an outcome (dependent) variable. The treatment variable is always an
independent variable, as are demographic variables such as age, sex, and race.
Theories were described previously as attempts to develop plausible explanations of reality. They
are specific statements regarding the relationship between (usually two) variables and are derived
difference and is the one actually tested statistically. For example: Poverty is not related to crime.
Chapter 1 • Introduction to Criminal Justice Research Methods 17
One approach to research involves formulation of hypotheses, the operationalization or measurement
of the variables, and the testing or bringing of evidence to bear upon these. Figure 1.1 outlines
a model of the research process.
TABLE 1.1 Researchese: Basic Terms
Concept Operationalization Variable
Cynicism A Cynicism Scale consisting
of 20 questions ranging from
1 (low cynicism) to 3 (high
cynicism) for each
Cynicism Score 20 (low)
to 60 (high)
Intelligence Administration of an
intelligence test that compares
mental age (scores on a test)
with chronological age
IQ (Intelligence Quotient),
for example, range of below
55 to 145+
Risk on parole A Parole Risk Prediction Scale
called "Salient Factor Score"
Parole Risk Score ranging from
0 (poor risk) to 10 (very good risk)
Raters used a scale and rated
defendants from 0 (poor) to
Appearance Rating Score
0 (poor) to 10 (excellent)
Police patrol Precincts were assigned to be
either proactive (increased
patrol), reactive (decreased
patrol), or controlling (same
as usual patrol)
Police Patrol Strategy:
Independent (Predictor)* Variable (X)
(usually demographic variable or a treatment)
e.g., Appearance, Police Patrol, Age, Sex,
Race, Social Class
Dependent (Outcome)* Variable (Y)
e.g., Crime, Recidivism, Cynicism, Intelligence,
Risk on Patrol
* Identification of independent and dependent variables has been oversimplified for heuristic purposes.
FIGURE 1.1 A Model of the Research Process.
18 Chapter 1 • Introduction to Criminal Justice Research Methods
EXAMPLES OF THE RESEARCH PROCESS
A brief description of this process is shown in Durkheim's (1951) classic study of suicide,
which was originally conducted in 1897 and was one of the first empirical studies in deviant
behavior. At the time Durkheim performed his study-in the late nineteenth century-little in
the way of empirical analysis had been undertaken in the social sciences. Contrary to popular
views of the time, Durkheim proposed that group membership affects suicide. From this he
deduced the specific hypothesis that religious denomination, marital status, and the like
would affect suicide rates. He operationalized his key variables or indicators of group membership
by assuming that married people have greater group ties than singles, or that Judaism
and Catholicism required greater group religious orientation than Protestantism, which was
more individualistic. Through analyzing the available official suicide records in European
countries at the time, he simply compared rates for each variable subcategory. Drawing the
general conclusion that singles and Protestants had higher suicide rates than married people,
Jews, and Catholics, he inferred from these findings a now modified theory: Group membership
does affect suicide.
In examining Figure 1.1 as well as our example, note that reasoning may proceed by means
of an a priori assumption (before-the-fact reasoning), wherein a theoretical idea precedes any
attempt to collect facts or use as a posteriori assumption (after-the-fact reasoning) in which
theory is developed after the data has been collected. The former is an example of a deductive
process of reasoning, with reasoning based on hypothesis or theory, and the latter illustrates the
inductive process, with reasoning based on inference of facts or particulars to general principles
or theory. Thus, theory to fact is deduction and fact to theory is induction. Deduction involves
moving from a level of theory to a specific hypothesis, whereas induction entails inferring about
a whole group on the basis of knowing about a case or a few cases. Sherlock Holmes' famous
compliment to Dr. Watson, "Brilliant deduction, my dear Watson," should probably have read
"induction" because Watson, in helping Holmes solve a case, was proceeding a posteriori from
specific facts or evidence to a conclusion or theory.
In order to further illustrate the research process, let us examine a hypothetical example. In an
experimental correction program called Salvation House, half of those scheduled for incarceration
are sent instead to this new community-based treatment program on the basis of a very general
theory that offenders better adjust, or are more likely to become rehabilitated, in a community
rather than in prison. A specific hypothesis derived from this theory (using, as we will describe in
Chapter 3, an interrupted time-series design) is that the Salvation House experimental group will
experience lower recidivism than the control group of incarcerateds. The dependent variable (Y)
is recidivism, and the predictor or independent variable (X) is assignment to jail or Salvation
House. The concept "recidivism" is defined as a reduction in both the quantity and quality
(seriousness) of crime commission over a one-year period after release compared with a similar
period prior to assignment to either correctional program.
gathering involved the simple examination of official statistics kept by police and
probation/parole agents on violations, as well as the weighting of the seriousness of these
offenses. Let us suppose the analysis demonstrated a "suppression effect"-a decline in both
the quantity and seriousness of offenses of the Salvation House assignees. We might now
Chapter 1 • Introduction to Criminal Justice Research Methods 19
draw the empirical generalization that less recidivism is demonstrated as a result of
community-based corrections, at least in our limited case study. We thus lend support to our
original theory that correction in the community appears to be more successful than isolated
incarceration, at least in bringing about some decline in the seriousness and quantity of
offenses. The theory is not set up for reanalysis and investigation.
An actual evaluation of juvenile intensive aftercare probation for serious offenders in
Philadelphia (Sontheimer and Goodstein, 1993) found it had a major impact on reducing the
frequency of subsequent offenses, but not the incidence of recidivism. Had only the latter been
used to operationalize recidivism, then a successful project would have been evaluated as
showing no difference.
GENERAL STEPS IN EMPIRICAL RESEARCH IN CRIMINAL JUSTICE
Although research varies considerably in scope, style, and procedure, most studies-particularly
those of a more quantitative nature-follow these general steps in research:
1. Problem formulation. Review, selection, and specification of the area to be investigated.
2. Research design. Type of experimental or nonexperimental approach, studies of a group
(or groups) at one time or over a period of time, and use of control groups.
3. Data collection methods. Choice of a variety of methods such as observation, reanalysis of
existing data, questionnaires, and interviews.
4. Analysis and presentation of findings. Summarizing, reporting, and statistically analyzing
where appropriate and presenting findings.
5. Conclusions, interpretations, and limitations. What the researcher believes the study has
Despite the neat, logical appearance of the research reported in journal articles, it is the
rare project that follows these steps in a straightforward fashion. Some research, for instance,
is exploratory and hypothesis generating rather than hypothesis testing in nature. The steps
remain, however, a useful heuristic device for explaining the process at this stage in a simple
manner. Although we will discuss problem formulation next, the organization of the text is
designed to follow these steps. Chapter 2 examines ethics in research, Chapters 3-8 look at
different data-gathering methods, and subsequent chapters discuss the presentation of
findings and their interpretation. By combining these steps with the model of the research
process described in Figure 1.1, we find that the investigator is often faced first with the issue
of finding a research problem.
PROBLEM FORMULATION: SELECTION OF RESEARCH PROBLEM
Problem formulation may be guided by many considerations including personal experiences. Take
your gut feelings seriously and pursue them. Chances are that each of us has unique experiences or
sensitivities that give us an edge in terms of interest or feel for a subject. This is an advantage that
should be recognized and acted on. Practical concerns may govern one's decision-data are
available on the job, our agency needs to have a similar study done, or the subject is manageable and
likely to be completed in the time allotted. The latter is certainly an important consideration in
academic theses and dissertations.
20 Chapter 1 • Introduction to Criminal Justice Research Methods
In selecting a research problem, one should look for gaps in theory or the current state of
the art, feasibility of doing the research, ambiguous and conflicting findings in the current literature,
as well as the potential timeliness of policy implications. Additionally, the availability of
funding and sponsorship is an important consideration, particularly in large-scale projects. The
NIJ has an active research agenda and specified the following areas for funding for fiscal year
2007 (NIJ, 2008):
Research on sex offenders
Analysis of existing data
Trafficking in human beings
Retail drug markets
Electronic monitoring for moderate to high risk offenders
Abuse, neglect, and exploitation of elderly individuals
Research on policing and public safety interventions
Rape in correctional settings
Intimate partner violence and stalking
Other sources that may influence one's choice of a research problem include administrative
decision-making needs, scientific or intellectual interests, and/or attempts to ameliorate
crime or injustice. One should kick around topics and ideas with fellow students, advisors, and
professors and search and browse the literature. Peruse the library stacks and consider replication
studies of previous research. Minority and feminist scholars have charged that much research
and scholarship in criminology and criminal justice has ignored minorities and females and has
originated primarily from a white, androcentric (male-centered) bias (Chesney-Lind, 1989;
Russell, 1992; and Mann, 1993). Exhibit 1.4 examines some of these issues.
Feminist Perspectives and Research Methods
positivism to rest in the human disciplines" (Denzin,
1989, pp. 66-67; Denzin, 1984). In calling for
nonsexist research methods, Eichler (1988) speaks of
"gender insensitivity" in social science research. This
involves an ignoring of gender as an important social
Others attack "malestream" approaches to
empirical criminal justice and argue that feminist
writings and voices should be incorporated into
the mainstream of criminal justice education
(McDermott, 1992; Renzetti, 1993). Such inclusion
Feminist theory has emerged as a major force in
criminology and criminal justice. While it has many
expressions, it draws on Marxist, interactionist, and
critical theory and advocates a methodology that
differs from the dominant empirical positivism.
Some feminist writers view the latter as failing to
include gender as a central force and as being blind
to its own ideological bias, an androcentrism that
ignores females as a central part of crime and justice
issues. "Observers are gendered beings, and the
research act is a gendered production. Feminism lays
PROBLEM FORMULATION: SPECIFICATION OF RESEARCH PROBLEM
The mere selection of a subject for investigation is only the beginning. One must now formulate
hypotheses, define key concepts, indicate appropriate operationalization, or decide upon a qualitative
sensitizing approach, decide upon research strategies, and finally relate one's research
problem to broader issues in criminal justice.
One key way to search for research problems is, of course, through a literature review.
Such a search will more likely than not hone what may have begun as a simple, vague hunch.
Many novice researchers, anxious to get on with the task, view the problem formulation and
literature review stages of a research project as a waste of precious time or a painful process
preceding the actual research. In reality, this is the most important period of a study; it refines
that which is to be examined and relates it to current and past inquiries, thus preventing the
reinvention of the wheel or rediscovery of a dead end. Table 1.2 lists selected journals and
abstracts that are useful in a literature review in criminal justice and criminology. Sources of
existing data are treated in more detail in the section on uses of available data in Chapter 8.
Students have available to them instant access to thousands of academic journals and
periodicals from any computer with an Internet connection. In addition to abstracts, many of
these online data sources contain full documents that may be viewed or downloaded. As
of feminist methods in social research (Reinharz,
1992) can be illustrated by a study by Elizabeth
Men Experience Sexual and Physical Dangers, in
which she used in-depth interviews to tap women's
experiences that do not turn up in standard surveys.
Feeling that much nonfeminist research is sexist due
to cultural beliefs and a preponderance of male
researchers, feminists question perspectives that
assume traditional gender roles. This bias
particularly has expressed itself in the past in
writings on topics such as rape and domestic
In a content analysis of twenty-two introductory
criminal justice texts in print in 1989, Dorworth and
Henry (1992) found that women and blacks were
underrepresented in photographs as authorities. While
women were overrepresented as victims, blacks were
overrepresented as offenders. Criminology has
omitted black females by equating "woman" with
white woman and "black" with black male.
McDermott (1992, pp. 247-248) enjoins:
The newer feminist perspectives suggest
that reality isn't clean and tidy, and that
experiences don't come in little boxes that
are ready to be labeled and counted. . . . It
is frightening but necessary to begin to
understand knowledge as situated and as
socially constructed, and to view our methods
of obtaining knowledge as potentially biased.
We should encourage our students to consider
the potential contributions of newer
perspectives and to acknowledge that the
issues surrounding methodology and
epistemology [how we come to know] exist
and are real. There is no other way to move
Sources: Denzin, Norman. The Research Act.
Englewood Cliffs, N.J.: Prentice-Hall, 1989; 66-67;
Denzin, Norman. "An Interpretation of Recent Feminist
Theory: A Review Essay." Sociology and Social Research
68 (1984): 712-718; McDermott, Joan M. "The
Personal is Empirical; Feminism, Research Methods, and
Education 3 (Fall, 1992): 237-249; Eichler, Magrit.
Nonsexist Research Methods. Boston: Allen and Unwin,
1988. Renzetti, Claire M. "On the Margins of the
Malestream (Or They Still Don't Get It. Do They?)."
Journal of Criminal Justice Education 4 (Fall, 1993):
219-234; Stanko, Elizabeth. Everyday Violence. London:
Pandora, 1990; and Dorworth, Vicky, and Marie Henry.
Education 3 (Fall, 1992): 251-260.
EXHIBIT 1.4 (Continued )
22 Chapter 1 • Introduction to Criminal Justice Research Methods
TABLE 1.2 Literature Review Sources
Selected Journals Relevant to Criminal Justice and Criminologya
American Criminal Law Review Journal of Family Violence
American Journal of Criminal Justice Journal of Justice Issues
American Journal of Police Journal of Law and Society
American Journal of Sociology Journal of Legal Studies
American Sociological Review Journal of Police Science and Administration
British Journal of Criminology Journal of Quantitative Criminology
British Journal of Sociology Journal of Research in Crime and Delinquency
Canadian Journal of Criminology Journal of Society Administration
Corrections Digest Judicature
CJ International Justice Quarterly
Crime and Delinquency Law and Society Review
Crime and Social Justice NIJ (National Institute of Justice) Reports
Criminal Justice and Behavior Police Chief
Criminal Justice Ethics Police Studies
Criminal Justice Newsletter The Public Interest
Criminal Justice Policy Review Public Opinion Quarterly
Criminal Justice Review Social Forces
Criminology Social Problems
Federal Probation Social Science Quarterly
International Journal of Criminology and Penology Sociological Inquiry
Journal of Contemporary Criminal Justice Sociology and Social Research
Journal of Crime and Justice Victimology
Journal of Criminal Justice Violence and Victims
Journal of Criminal Law and Criminology
Abstracts on Police Science Police Science Abstracts
Crime and Delinquency Abstracts Psychological Abstracts
Criminal Justice Abstracts
Criminal Justice Periodical Index
Reader's Guide to Periodical Literature
(for popular sources only)
Criminology and Penology Abstracts Social Science Index
Document Retrieval Index Social Sciences Citation Index
Encyclopedia of Crime and Justice Sociological Abstracts
New York Times Index
a This list is only a selection of journals and not an exhaustive list.
b These are among the many abstracts that are available in the reference section of the library. For more details on sources of
information in criminal justice and criminology, see Chapter 8.
illustrated in Figure 1.2, one simply chooses articles that are available in full text. Exhibit 1.5
provides an illustration of the growing online services available on computer networks and
the electronic highway. Conducting Internet searches using search engines such as Yahoo,
Infoseek, Lycos, or Excite is a useful means of becoming more familiar with one's topic
while also conducting a literature review.
Table 1.3 presents an abbreviated version of the APA (American Psychological Association)
style manual, which is the most widely used reference style in the social sciences.
FIGURE 1.2 Online Research. (Source: National Criminal Justice Reference Service. www.ncjrs.gov)
The World Wide Web (WWW)
engines will list several Web addresses which
are called URLs (Uniform Resource Locators). For
example, Cecil Greek's homepage at the University
of South Florida is www.criminology.fsu.edu/
cjlinks/. Each of these sites contains clickable listings
which will connect you with other Web pages.
Some useful criminal justice sites on the World Wide
The World Wide Web (WWW) is the fastest growing
component of the Internet. Through the use of Web
browsers such as Netscape and/or Internet Explorer,
scholars are able to access a variety of information.
Search engines such as Yahoo, Lycos, Excite, and
Infoseek will search by topic/author and provide
listings of sites which can be found on the Web.
Once the listing of sites is provided, the search
Cecil Greek www.stpt.usf.edu/~greek/cj.html
National Institute of Justice www.ojp.usdoj.gov/nij/
Bureau of Justice Statistics www.ojp.usdoj.gov/bjs/
National Criminal Justice Reference Service www.ncjrs.org
24 Chapter 1 • Introduction to Criminal Justice Research Methods
Financial Crimes Enforcement Network http://www.fincen.gov
Drug Enforcement Agency www.usdoj.gov/dea/
Federal Bureau of Investigation www.fbi.gov/
Federal Bureau of Prisons www.usdoj.gov/bop/bop.html
Library of Congress www.loc.gov/
National Archives www.nara.gov/
United Nations Criminal Justice and Crime
National Archive of Criminal Justice Data
(Interuniversity Consortium on Political and
Social Research-University of Michigan)
U.S. Department of Justice www.usdoj.gov/
U.S. Federal Judiciary www.uscourts.gov/
The White House www.whitehouse.gov/WH/Welcome.
National Institute on Drug Abuse www.nida.nih.gov/
Office of International Criminal Justice www.oicj.org
Pavnet Online (Partnership Against Violence) www.pavnet.org/
Amnesty International www.amnesty.org
U.S. Census www.census.gov/
Central Intelligence Agency www.cia.gov
U.S. Secret Service www.ustreas.gov/treasury/bureaus/usss/
Bureau of Alcohol, Tobacco and Firearms www.atf.treas.gov/
Drug Enforcement Agency www.usdoj.gov/otj/otj.html
Office of Juvenile Justice and Delinquency www.ojjdp.gov/
National Clearinghouse for Alcohol and Drug
U.S. Government Printing Office www.access.gpo.gov/
National Security Agency www.nsa.gov
Britannica Online www.britannica.com
Department of Justice Career Opportunities www.usdoj.gov/careers.html
National Employment Listing Service
General Accounting Office www.gao.gov/
National Institute of Justice www.ojp.usdoj.gov/nij
Justinfo Online www.ncjrs.gov
New York Times Navigator www.nytimes.com/library/cyber/
EXHIBIT 1.5 (Continued )
TABLE 1.3 APA (American Psychological Association) Reference Style
The following is an abbreviated version of the most widely used style reference manual in the
Association (APA). Consult the APA Manual for more details (www.apastyle.org).
Reference citations are incorporated directly into the paper.
Hagan (2009) indicates that the APA style incorporates citations directly in the text.
Direct quotes should provide a page number and "p" or "pp" either after the year or at the end
of the quote.
Hagan (2009, p. 10) indicates that "the APA style incorporates citations at
the end of the quote."
For two or more authors, use "and" in the text and "&" in the citation.
Hagan and Rooney (2009) claim that the Steelers are the best team in the NFL.
Writers have claimed (Hagan & Rooney, 2009) that the Steelers are the
best team in the NFL.
Internet sources are cited in the same manner as print sources.
Provide the name of the author(s) followed by the year of publication: (Hagan, 2009).
Alphabetize all references at the end of your paper.
Hagan, F. E. (2009). Research methods in criminal justice and criminology (8th ed.).
Upper Saddle River, NJ: Prentice-Hall.
Two or more authors
Hagan, F. E., & Tontodonato, P. (2004). Classical and sociological theories of delinquency. In
P. Benekos & A. Merlo (Eds.), Corrections: dilemmas and directions. Cincinnati: Anderson.
List author, year, and title of the article without quotation marks, with the first
word and any proper nouns capitalized, name of the journal italicized, volume
number italicized, and inclusive page numbers not preceded by "p" or "pp."
Hagan, F. E. (1986). Sub rosa criminals. Clandestine Tactics and Technology, 11, 298-320.
In addition to consulting the actual APA publication manual (www.apastyle.com),
the reader might find it more convenient to consult books or periodicals that use the
APA style for examples.
Source: Adapted by the author from the Publication Manual of the American Psychological Association
26 Chapter 1 • Introduction to Criminal Justice Research Methods
Methodological narcissism 7
"Broken Windows" 9
Pure research 10
Applied research 10
Crime analysis 13
Quantitative research 14
Qualitative research 14
Research shock 15
Dependent Variable 16
Independent Variable 16
Steps in research 19
Problem formulation 19
In this introductory chapter, we discussed why those
who attack criminological or criminal justice research
as being common sense often deceive themselves with
nonsense, substituting their own personal bias or ignorance
for objective, scientific information. The study
of research methods was described as an invaluable
tool for understanding the latest developments in
criminal justice as well as in society. Knowledge is
what people create symbolically to represent reality
and was described by Comte as "progressing through
three stages": theological (supernatural), metaphysical
(philosophical, rational), and scientific (rational plus
scientific method or proof). Criminal justice strives for
scientific status. Scientists rely upon probabilistic
knowledge, that is, predicting general trends, not
In conducting or critiquing studies, researchers
are advised to avoid methodological narcissism
(fanaticism for one method or method for method's
sake). Those concerned with good research should
be objective, vigilant (for error), and sympathetic
(because all research is infested with error). Theory is
an attempt to develop plausible explanations of
reality, whereas methodology is an attempt to collect
accurate facts or data. Both are indispensable in
providing sound criminal justice knowledge. New
knowledge often takes place due to paradigm shifts.
A paradigm is a model or scheme with which reality
is viewed. Pure research is directed at the acquisition
research is interested in knowledge for the practical
research (positivism) is concerned with measuring
social reality by using the scientific method.
Researchese, the language of research, includes
concepts (abstract tags put on reality), operationalization
(defining concepts by describing how they will
be measured), variables (operationalized concepts
or concepts that vary), theories (general statements
regarding relationships between variables), and
hypotheses (specific statements regarding the relationship
between variables). Variables may be dependent
(outcome) (the variable one is attempting to predict,
denoted by the letter Y) and independent (the predictor
variable, denoted by the letter X or any letter
other than Y).
The research process was illustrated as a
circular process from theory to hypothesis to
research design to data gathering to findings and
then back to theory.
justice are: problem formulation, research design,
data-collection methods, analysis/presentation of
findings, and conclusions/interpretations/limitations.
The first step-problem formulation-was discussed;
the remaining steps will be the subject of subsequent
The feminist perspective on research methods
offers an alternative to androcentric bias in
1. Name a myth or inaccurate commonsense view of
crime or criminal justice other than one presented in
this chapter. Indicate how research has clarified this
2. Are criminology and criminal justice sciences?
Discuss some developments that support their claim
to scientific status.
3. What is the role of theory in criminological/criminal
justice research, and why has there been such a shortage
of new theory since the 1960s?
4. Choose a recent journal article and identify: (a) the
research problem, (b) research design, (c) data-gathering
strategy, (d) dependent variable/s, (e) independent
variable/s, and (f) operationalization of the key dependent
and independent variables.
5. What is the feminist perspective on research
methods? Why is there a need for such a perspective?
How does it differ from the "malestream"?
Research Navigator www.researchnavigator.com
Academy of Criminal Justice Sciences www.acjs.org
United Nations Crime and Justice Information
National Criminal Justice Reference Service (NCJRS)
THE END OF CHAPTER 1
C H A P T E R
2 Ethics in Criminal Justice Research
Ethical Horror Stories
Social Science Examples
Exhibit 2.1 AIDS Research in Africa
and Asia: Is It Ethical?
Exhibit 2.2 The Minerva Consortium
and the Human Terrain System
Researcher Fraud and Plagiarism
Exhibit 2.3 Legendary Research Scams
The Researcher's Role
Research Targets in Criminal Justice
Ethics and Professionalism
Ethics in Criminal Justice Research
History of Federal Regulation of Research
The Belmont Report
Institutional Review Boards (IRBs)
Research Activities Exempt from HHS Review
National Institute of Justice's Human Subject
Confidentiality of Criminal Justice Research
Exhibit 2.4 Codes of Research Ethics of the
Academy of Criminal Justice Sciences (ACJS)
Ethical Issues in Criminology/Criminal
Avoid Research That May Harm Respondents
Honor Commitments to Respondents
and Respect Reciprocity
Exercise Objectivity and Professional Integrity
in Performing and Reporting Research
Protect Confidentiality and Privacy of
The Brajuha Case (Weinstein Decision)
The Ofshe Case
The Hutchinson Case
The Scarce Case
Additional Ethical Concerns
Avoiding Ethical Problems
ETHICAL HORROR STORIES
Nazi doctors tortured, maimed, and murdered innocent captive subjects. Some scientists have
purposely allowed subjects to suffer and even die of a disease while withholding a known cure.
As part of other experiments, researchers have deceived people into believing that they were
electrocuting people; they have created an artificial prison in which participants become hostile
and aggressive; or they have spied on secret sexual activity and later showed up at the subjects'
homes and invaded their privacy as part of a scientific survey. Intelligence agencies have
employed social scientists to gather data on dissidents in Third World countries and, along with
the military, have employed researchers and scientists (including former Nazis) to conduct often
bizarre and dangerous experiments on unknowing subjects. If all of these situations resemble
plots for a XXXXX XXXXX gothic novel, they are not. Each is an actual example of a project that
has raised ethical controversies.
Major ethical concerns about the use of human subjects in research originally arose as a
result of the outrageous examples of inhuman Nazi experiments during World War II. Dr. Josef
Mengele, the "Angel of Death," performed horrifying human experiments in which captive
subjects were tortured and killed in the name of scientific research. These experiments were coldblooded
and inhumane. In the name of medical research, people were infected with diseases, used
as guinea pigs to test new drugs, administered poisons, and exposed to extreme temperatures and
decompression to test reactions to high altitudes (Katz, 1972). After the war, the Nuremberg trials
defined such behavior as war crimes and crimes against humanity. The Nuremberg Code set forth
principles governing the use of human subjects in research, including the requirement that such
subjects "voluntarily consent" to participate in a study (Wexler, 1990, p. 81).
In the infamous Tuskegee Syphilis Study (Brandt, 1978), the U.S. Public Health Service
withheld penicillin, a known cure for syphilis, from 425 uneducated black male sharecroppers
who suffered from and eventually died of untreated syphilis. This study, which began in 1932
before a cure for syphilis was available, was completed in the 1970s, well after such medicine
was developed. Such inhumane biomedical research on unsuspecting subjects did not end with
the Tuskegee Study (Jones, 1982). In the 1960s, live cancer cells were injected into elderly
patients at a Brooklyn hospital without their knowledge. The U.S. military services, during
and after World War II, exposed their own soldiers to mustard gas and nuclear radiation,
resulting in cases of chronic ailments and premature death. During the post-World War II Cold
War era, American intelligence agencies, with the cooperation of the scientific community,
performed bizarre and dangerous experiments on unknowing subjects. In the early years of the
Cold War, American intelligence agencies had become convinced that the Communists had
developed secret mind control and brainwashing techniques and that it was necessary in this
battle for human minds and world domination to pull out all the stops (Scheflin and Opton,
1978; Hagan, 2006).
This explains, but does not condone, the following abuses:
employee, who then committed suicide. Later, the government refused to tell his grief- and
guilt-stricken family what really happened.
U.S. military experimented with behavior-control devices and interrogation techniques,
including ESP, drugs, polygraphs, hypnosis, shock therapy, surgery, and radiation. These
experiments involved secret testing on unsuspecting citizens and, if death or injury occurred,
a cover-up (Cousins, 1979; Simon and Eitzen, 1996).
of experiments at a Montreal mental hospital (Witt, 1988, p. 2A). As an example, a
Canadian teenager seeking medical treatment for an arthritic leg was first given LSD,
and then he was subjected to electroshock therapy and forced to listen to taped messages
saying, "You killed your mother." Such studies were conducted on more than 100 Canadians
30 Chapter 2 • Ethics in Criminal Justice Research
by an American doctor who had been the former president of the American Psychiatric
Association, and the studies were financed by the CIA.
thirty-year period beginning in the mid-1940s, federal agencies had conducted exposure
experiments on American citizens, including injecting them with plutonium, radium, and
uranium. These studies included feeding radium or thorium to elderly patients during an
experiment at the Massachusetts Institute of Technology, irradiating inmates' testes with
x-rays, and exposing people to open-air fallout tests and feeding them real fallout from a
Nevada test site (Lawrence, 1988). It was later revealed that the U.S. military even
employed former Nazi doctors and scientists, using them to conduct chemical experiments
on U.S. military personnel at Fort Dietrick (Aberdeen, Maryland).
governmental experiment of open-air nuclear testing in the post-World War II period.
Between 1951 and 1963, the U.S. government detonated 126 nuclear bombs in the sky above
Nevada. Each detonation sent a cloud of radiation aloft equal to that released during the
Chernobyl nuclear disaster in the former Soviet Union. Such testing spoiled soil in Virginia,
milk in New England, wheat in South Dakota, and fish in the Great Lakes. During the same
period, U.S. Army infantry were ordered to observe nuclear tests from unprotected trenches.
and graduate student Mary Tudor and her professor Wendell Johnson taught children at an
Iowa orphanage a "lesson they would never forget"-how to stutter (Lessons, 2001).
While the experiment helped thousands of children overcome speech difficulties, this took
place at the expense of some of the children unnecessarily being subject to lives as outcasts
and misfits. Thirteen of her subjects who are still alive learned of the experiment in 2001
when reporters from the San Jose Mercury News contacted them. The now stuttering
children had been divided into two groups of eleven, one labeled normal speakers and
given positive speech therapy and the other group induced to stutter. Eight members of the
treatment group became permanent stutterers. While Tudor had guilt feelings and returned
to the orphanage a number of times to attempt to reverse the damage, Johnson did nothing
and became very prominent in his field of speech pathology due to the findings. Tudor
describes how during the experiment the trusting orphans greeted her, running to her car
and helping to carry in materials for the experiment. In 2007, the state of Iowa agreed to
pay $925,000 to six of the subjects of the study who had been harmed by the University of
Iowa researchers. The 1939 experiment became known as the "Monster Study" because of
the methods used by the researchers. The San Jose Mercury News broke the story in 2001
based on statements made by Mary Tudor (Associated Press, 2007, p. 8A).
federal system for protecting human research subjects was breaking down. The report
claimed that Institutional Review Boards (IRBs), which judge the ethical merits of research,
were overburdened, insufficiently staffed, and subject to conflict of interest (Weiss, 1998).
Although the details of one study were more complicated than described, newspapers reported
that physicians in New York offered to give Toys R Us gift certificates to thirty-six healthy
black and Hispanic elementary school pupils if the children agreed to enroll in a medical
study that required them to take a potentially life-threatening drug (actually low doses of
fenfluramine). The drug was used to measure levels of brain hormone implicated in antisocial
behavior. The older brothers of the young males were delinquent and the study was
designed to examine whether this increased the latter's delinquency potential.
Chapter 2 • Ethics in Criminal Justice Research 31
to discontinue their medication. This caused some to be confused, violent, and to lose their
ability to concentrate. One committed suicide. Of particular concern in these studies was the
cavalier manner in which vulnerable populations (young, mentally disturbed, and minorities)
were put at risk.
it was reported in 2008 that the drug manufacturer Merck wrote research reports praising
their drugs and then paid doctors to attach their names to those reports. The ghostwritten
medical reports were published in professional journals. Merck had manipulated dozens of
such reports in order to promote Vioxx, a drug that was later recalled (Guterman, 2008;
Exhibit 2.1 considers a controversial experiment with an AIDS vaccine in Third World
Most of the foregoing examples have been biomedical in nature; but because social and behavioral
research likewise puts subjects at risk, its activities have led to similar ethical concerns.
Many social scientific studies related to crime and deviance have come under scrutiny. The three
AIDS Research in Africa and Asia: Is It Ethical?
In September 1997, an international debate began
when it was revealed that American-sponsored
field experiments with AZT (azidothymidine)
vaccine (a believed treatment for AIDS) were being
conducted in Asia and Africa on pregnant women
who had been diagnosed with the AIDS virus.
While some of the women (the experimental
group) were given AZT, others (the control group)
were given placebos. Critics charged that by not
giving AZT to all, the newborns of the control
group were knowingly being given AIDS.
Defenders of the experiment pointed out that
without the experiment there would have been no
AZT treatment for anyone in such poor countries.
They argued that such treatment was ethical since
half of the women who had received some
treatment would have otherwise received none.
Such treatments are routine in wealthier countries
such as the United States as soon as AIDS is
Critics also drew parallels with the Tuskegee
Syphilis Experiment and questioned the morality of
withholding a likely cure, while proponents noted
that the findings promised to discover a shorter,
cheaper course of AZT treatment that could protect
newborns. Marjorie Speers is in charge of human
subject protection at the Center for Disease Control
and Prevention (CDC).
Consent is another minefield, Speers said.
U.S. regulations require individual informed
consent. In other nations, husbands consent
for wives and tribal leaders consent for villagers.
To ask for more is insulting. Speers said the
very notion of informed consent is alien in parts
of the former Soviet Union after decades of
totalitarianism. And the CDC was forced to
cancel research in two Arab countries because
the governments refused to allow women to
serve on ethical review panels (Long, 1998).
Source: Adapted from XXXXX, XXXXX. "AIDS Research
Dealer, 20 May 1998, A-1.
32 Chapter 2 • Ethics in Criminal Justice Research
social scientific studies that seem to be cited in the literature most often are: Stanley Milgram's
Obedience to Authority (1974), Philip Zimbardo's simulated prison experiment (1972, 1973,
1974), and Laud Humphreys' Tearoom Trade (1970).
In his classic Obedience to Authority study, Stanley Milgram had one very important
objective: He wanted to discover the causes of the Holocaust. During the Nazi era in Germany,
some people performed their gruesome duties as if it was "just another day at the office." How
and why do average, "normal" people commit the most monstrous acts? In Milgram's study,
which was designed to answer this question, volunteer subjects were recruited and paid to act as
"teachers" while "confederates" (fake subjects who were really in the study) acted as "pupils."
The subjects (teachers) were then deceived into believing that each time they threw a lever on a
shock apparatus, they were administering gradually more painful electric shocks to pupils,
whom they could hear but not see. When the pupils failed to answer a question correctly, the
teachers were to administer shocks. Despite protests from the pupils, teachers were willing to
administer levels of voltage which they believed to be dangerous, particularly when assured by
lab assistants, who appeared as scientific authorities, that such behavior was necessary. Subjects
experienced personal turmoil both during and immediately after the experiment, although
debriefing (explaining the purpose of the study after-the-fact) seemed to have resulted in no
long-term harm. Do experimenters have the ethical right to deceive and put subjects in a position
of emotional stress in the name of science?
In Philip Zimbardo's simulated prison study, male, undergraduate, paid volunteers
assumed the roles of either guard or prisoner. A mock prison was constructed in the basement of
a Stanford University building; and prisoners and guards assumed their respective roles, complete
with uniforms, nightsticks, and mirrored sunglasses for the guards, and numbers and prison
garb for the prisoners. While the experiment was to have lasted at least two weeks, individuals
became so carried away with the roles-passivity and hostility by prisoners and aggressive and
dehumanizing behavior by guards-that Zimbardo cancelled the study after six days rather than
risk harm to the participants. The "Lucifer effect" (www.lucifereffect.org) is a term coined by
Philip Zimbardo to refer to a transformation of human character that causes good people to
engage in evil actions. This could include sexual degradation and torture such as at the Abu
Ghraib prison in Iraq. In The Lucifer Effect: Understanding How Good People Turn Evil,
Zimbardo (2007a) describes one humiliating exercise in the Stanford experiment called the
After a brief consultation, our toughest guard (nicknamed "John Wayne" by the
prisoners) and his sidekick devised a new sexual game. "OK pay attention. You three
are going to be female camels. Get over there and bend over, touching your hands to
the floor." When they did, their naked butts were exposed because they had no underwear
beneath their smocks. John Wayne continues with obvious glee, "Now you
two, you're male camels and hump them."
Although their bodies never touched, the zombie-like prisoners began to simulate the
sodomy. One of Zimbardo's associates, after observing such exercises, berated him for contributing
to the suffering of human beings. This snapped Zimbardo back to his senses and led to him
canceling the experiment (Zimbardo, 2007b).
In 2002, the British Broadcasting Corporation (BBC) cancelled plans to replicate for television
the "notorious" Stanford experiment, featuring fifteen participants, for fear for their emotional
Chapter 2 • Ethics in Criminal Justice Research 33
and physical well-being. The volunteers had been "incarcerated" in a simulated "prison" at Elstree
studios in Hertfordshire. The BBC had advertised for volunteers, warning participants that those
chosen would be exposed to "exercise, tasks, hardship, hunger, solitude and anger" (Wells, 2002).
Overseen by two psychologists, Alex Haslam (Exeter University) and Stephen Reicher
(St. Andrews University), as well as an independent "ethical committee," the experiment was terminated
only a day or two short of the planned ten days. Zimbardo himself commented, "that kind
of research is now considered to be unethical and should not be redone just for sensational TV and
Survivor-type glamour. I am amazed a British university psychology department would be
The possibility of such potential harm is even worse if the subjects do not consent to
participate in a study; in such experiments, observation usually takes place in private settings, and
the behavior involves activity which society may regard as immoral or illegal. All of these factors
were present in Laud Humphreys' controversial study, Tearoom Trade (1970), one involving
secret male homosexual behavior in public restrooms. Pretending to be a "watchqueen" (voyeur),
Humphreys served both as a lookout and as a hidden observer of such behavior. He copied license
numbers and traced them to the owners' homes. Changing his appearance, Humphreys showed up
at their homes under the guise of a mental health researcher.
Although Humphreys' research was important to the criminal justice system, which gained
important insights into the nature of such participants who engage in impersonal homosexual
liaisons in public places, was the obtaining of such knowledge justified given the risk for potential
harm to subjects if their secret sexual behavior were to become known to legal authorities,
family, or employers? Even though Humphreys claimed to have taken great precautions to protect
the anonymity of the subjects, did he have the right to put them in harm's way without their
permission? Is there any way of studying such behavior without using deceit and deception?
Should criminological researchers study only volunteers?
An amusing example is provided by a disguised observational study of the "Church of
Satan." The researcher (Alfred, 1976) felt guilty that he had deceived members of a satanic cult,
until he revealed his misrepresentation and was applauded by the group, since lying is considered
an appropriate satanic act.
What if the researchers are themselves the subject of deception? Through various fronts
during the 1960s and 1970s, the CIA, apparently without the knowledge of the recipients,
funded social-psychological research by such names as Sherifs, Orne, Rogers, Osgood,
and Goffman (Marks, 1979, p. 121), and it financed the publication of more than one
thousand books, pretending that they were the products of independent scholarship (Cook,
1984, p. 287).
In Project Camelot (Horowitz, 1965), for example, U.S. researchers studied student and
peasant insurgency movements in Chile. Because this was an area of the world where government
opponents routinely "disappeared," many subjects were justifiably suspicious of this effort, and
they feared the data it would generate. Believing that the "Yanqui" researchers were gathering
this information for intelligence purposes, they thought the study would have a chilling effect on
dissent. Although the researchers denied such CIA involvement at the time, some of the
researchers later discovered that their data were in fact being gathered for intelligence purposes.
The government of Chile, which had been unaware of the project, expelled the researchers and
brought to an end the infamous Project Camelot. Among the many questions raised by this study
are: Should researchers do the bidding of intelligence agencies, thus acting as spies? Whose side
is social research on? Should researchers refuse certain sponsorship or specify the conditions
under which sponsorship will be accepted?
34 Chapter 2 • Ethics in Criminal Justice Research
In 1998, over 100 boys ages six to eleven were given fenfluramine, half of a later
banned diet drug combination known as fen-phen, as part of a study of brain activity and
aggression. Some of the boys had been chosen because apparently they had older siblings
who were juvenile delinquents. The project was conducted by the New York State Psychiatric
Institute, Queen's College Psychology Department, and the Mount Sinai School of Medicine.
The boys were given a single dose of up to 10 milligrams of fenfluramine to see if it raised
serotonin levels in the brain. Such an increase is believed to reduce aggression (Associated
While the researchers claimed the boys' parents were informed of the risks and that the
level of dosages was not harmful, federal investigators such as at the Office of Protection from
Research Risk are reviewing the study. The Food and Drug Administration (FDA) estimated that
one-third of all adults who took the diet pill regularly may have experienced significant heart
valve damage. Fenfluramine and a related drug, Redux, were recalled by the FDA in the summer
There has been no shortage of ethical horror stories including those involving correctional
Punishment (1973) documents abuse of inmates by medical researchers, as does Allen Hornblum
in his book Acres of Skin (1998). Hornblum relates his first experience with such experiments on
entering a Pennsylvania prison in 1971 and noticing bizarre patchworks of gauze on many prisoners
and being told that the prisoners were testing perfumes. They were human guinea pigs not
only for perfumes, but also for soap, cosmetics, and even dioxin (radioactive isotopes and
psychological warfare agents). Hornblum's research revealed that, by 1969, 85 percent of all new
drugs were being tested on prisoners.
Hornblum used the Freedom of Information Act in order to obtain old records and found
that most of the subjects were never checked for long-term effects. Claiming that one researcher,
University of Pennsylvania professor Albert Kligman, ran a virtual human research factory using
prisoners as guinea pigs, Hornblum documents:
subjects to experience hallucinations and confusion for up to three weeks. The inmates
referred to EA 3167 and other mind-altering drugs that they received as LSD.
their instructions by subjecting several inmates to 7,500 micrograms or 468 times the
required dosage. Two inmates later sued, settling for only a few thousand dollars
By the late 1970s, a National Commission for the Protection of Subjects of Biomedical and
Behavioral Research suggested banning such a practice. The U.S. Food and Drug Administration
agreed, and by the early 1980s, most prison testing was outlawed. Former prisoners who had
participated in such experiments related negative psychological and physiological harm long after
the experiment. Most participants were poor and uneducated (Kinney, 1998).
In Darkness in El Dorado: How Scientists and Journalists Devastated the Amazon,
Tierney (2001) alleges major ethical violations by anthropologists who violated the human rights
of the Yanomami, a native people of Brazil and Venezuela. The Yanomami were injected with a
measles vaccine known to be virulent in isolated populations. A subsequent epidemic ensued,
and the research team supposedly was instructed not to assist the dying Yanomami. No informed
consent was gained from the vulnerable tribe (Fleur-Lobban, 2000). These accusations threw the
Chapter 2 • Ethics in Criminal Justice Research 35
American Anthropological Association into turmoil, with a committee charging that Tierney's
accusations were largely false but that many of his other points were well founded (Glenn, 2006).
To illustrate the seriousness with which federal agencies view the protection of human subjects,
on July 19, 2001, the U.S. federal government cancelled all federal funding for research
projects at Johns Hopkins University due to the death of a subject in a university-run experiment.
In 2000, Hopkins had led the nation in federally funded research. The dead volunteer subject had
participated in an asthma experiment in which the informed consent form failed to adequately
advise her of the experimental nature of the drug that killed her when she inhaled it.
Julianne Basinger in an article entitled "Research at What Cost?" (2001) examines the case
of Shawn Wight, a twenty six-year-old graduate student from Ohio State, whose participation in
research cost him his life. Accompanied by his professor, Shawn and other graduate students
were on a high altitude expedition in Western China to examine and drill ice cores from glaciers.
He developed severe reactions to the altitude that ultimately resulted in his death. While emergency
measures for dealing with such problems appeared inadequate, the State of Ohio ruled in
favor of the university in a $24 million liability suit brought by Wight's parents. The case represented
a wake-up call to universities and research directors, and Ohio State has since improved
its procedures and oversight functions. There is a growing awareness of the risks involved in
research for not only the subjects, but also the researchers and their employees (Craig, Corden,
and Thornton, 2000). Exhibit 2.2 describes the controversial Minerva Consortium and the
Human Terrain System that recruits academics to assist in the war on terrorism.
The Minerva Consortium And The Human Terrain System
Academia has a long history of aiding military and
intelligence agencies going back to the Office of
Strategic Services in World War II, Vietnam, the Cold
War, and before. Franz Boas, the father of American
anthropology, once complained that his colleagues
were spying on the German navy during World War
I while pretending to be doing research. Critics of
such activity argue that it violates the neutral
observer status of the professionals in the discipline
(Embedded Anthropologists, 2007, p. B4). In
America, the discipline of psychology has a long
relationship with the U.S. military. Controversy has
been raised regarding the role of the military's
behavioral science consultation teams (known as
biscuits). These are made up of psychologists and
others, who assist in interrogations. In light of
charges of torture of prisoners at Camp Delta at
Guantanamo Bay by the U.S. military, social
scientists must be careful in their involvement. The
American Psychological Association's (APA) code of
ethics permits consultative roles in such interrogations
as long as psychologists do not participate in
certain coercive practices such as waterboarding, in
which captives are placed in fear of drowning (Carey,
2008). Recent APA debates have questioned psychology's
role in the coercive practices used during the
George W. Bush Administration's antiterrorism
During World War II, the Manhattan Project
enlisted the talents of scientists and academics to
develop the atom bomb. Also at this time, the
Army's Psychological Warfare unit recruited social
scientists to develop wartime propaganda tools and
means of promoting allied military morale and
undermining enemy morale. In 2006, the Pentagon
launched a program entitled the "Human Terrain
System," in which academics were recruited in order
to conduct field research in Afghanistan and Iraq to
assist the U.S. military in waging a smarter
counterinsurgency war (Ephron and Spring, 2008).
36 Chapter 2 • Ethics in Criminal Justice Research
EXHIBIT 2.2 (Continued )
The specialists were charged with mapping the
population of communities, identifying clans and
conflicts between them and advising the military on
obtaining local support. Some possible participants
in the Human Terrain System program were
discouraged when the American Anthropological
Association indicated that such researchers would
most likely be violating the ethics of their profession
because they would be contributing data that might
be useful in military operations.
The "Minerva Consortium" is named for the
Roman goddess of wisdom. In 2008, it was
introduced by Robert Gates, secretary of defense, in
a speech before the Association of American
Universities (Goldstein, 2008). During the speech,
Gates quoted historian Arthur M. Schlesinger, Jr.'s
1957 speech after the Russian Sputnik beat the
United States into space, when he said the United
States "must return to acceptance of eggheads and
ideas if it is to meet the Russian challenge"
(Goldstein, 2008, p. B4). The Minerva Consortium
would establish a project that finances social science
research that is relevant to national security. Such a
proposal has been met with mixed reactions. Various
charges for and against such a project opined that it
would be a justification of imperialism, be needed to
counteract jihadist ideology, be used to justify
human rights abuses, provide the opportunity to
break new ground in the social sciences, contribute
to a further militarization of universities, and
produce a brain drain from other areas of research.
At least two graduate students who had volunteered
to serve in the Human Terrain System were killed in
Afghanistan (Glenn, 2008).
Sources: Adapted from: Carey, Benedict. "Psychologists
Clash on Aiding Interrogations." New York Times, 16
August 2008, A1; "Embedded Anthropologists."
Chronicle of Higher Education, 19 October 2007, B4;
Ephron, Dan, and Silvia Spring. "A Gun in One Hand,
A Pen in the Other." Newsweek, 21 April 2008, 34-35;
Glenn, David. "Peacekeeper and Scholar is Killed in War
Zone." Chronicle of Higher Education, 4 July 2008, A1,
A6, A9; Goldstein, Evan. "Enlisting Social Scientists."
Chronicle of Higher Education, 4 July 2008, B4, B5.
A fundamental expectation of any piece of scientific research is that it be accurate, honest, and
properly referenced. This, incidentally, is the reason why professors harp upon these themes when
correcting student papers; those who ignore these lessons may be haunted by them later in life.
Research fraud occurs when researchers purposely fabricate or misrepresent their findings.
Despite the pressure on researchers to "publish or perish"-tenure and grants are often
dependent on their success in getting published in research publications-the actual number of
cases of research fraud are relatively rare. In 1989, however, the HHS started an Office of
Scientific Integrity to address this issue (Neuman, 1991, p. 438; "Fraud in Research," 1994).
Perhaps the most celebrated case of a researcher's dishonesty was that of Sir Cyril Burt, a
famous British psychologist, whose studies on twins had demonstrated the inherited nature of
intelligence. After his death, researchers discovered that he had faked his data and that he had even
created nonexistent coauthors (Wade, 1976). It should be noted, however, that more recent investigations
of the Burt affair have drawn differing conclusions (Hearnshaw, 1979), with some
reviewers indicating that Burt was innocent of outright fraud and that some of his detractors may
have been guilty of character assassination (Joynson, 1989; Fletcher, 1991). In 1995, Dr. Gerald L.
Gerson, after examining Louis Pasteur's 102 laboratory notebooks, charged that Pasteur had
misled, lied, secretly stolen a rival's techniques, and otherwise deceived the scientific world in
order to receive grants, patents, and awards (Altman, 1995). Exhibit 2.3 depicts two legendary
Plagiarism, as most college students know, is a type of fraud in which a writer presents the
ideas or work of someone else as his or her own. Prominent figures such as Senator Joseph
Chapter 2 • Ethics in Criminal Justice Research 37
Biden, Alex Haley, John Hersey, Martin Luther King, Jr., Doris Kearns Goodwin, and Stephen
Ambrose have been accused of plagiarism. Researchers should be very careful to properly
acknowledge-and therefore not take credit for or steal-the ideas of others (Broad and Wade,
1983; LaFollette, 1992). Some students mistakenly believe that since they can copy material with
the click of a mouse, they do not have to cite or acknowledge that the material was taken from
another source. Simply changing the order of wording in a text still requires citation. Figure 2.1
presents some additional issues regarding plagiarism.
Publishers of academic journals have begun to use anti-plagiarism software such as
iParadigms, Turnitin, and CrossCheck to detect plagiarism (Rampell, 2008, p. A17). The Office
of Research Integrity (ORI) is a federal agency supported by the U.S. Public Health Service that
is concerned with detecting and preventing scientific misconduct. It evaluates all of the investigative
records submitted by institutions and helps determine whether there has been any misconduct
at institutions that receive support from the HHS. According to the U.S. federal definition of
research misconduct, it includes fabrication, falsification, or plagiarism in proposing, performing,
or reviewing research or in reporting research results (Titus, Wells, and Rhodes, 2008). In
2006, 2,212 research scientists, a 51 percent response rate, responded to a survey of research
misconduct. Of these, 192 scientists (8.7 percent) indicated having observed or of having direct
evidence of researchers in their own department committing one or more incidents of misconduct
over the past three academic years. They described a total of 265 incidents (Titus, Wells,
and Rhodes, 2008).
Scientific misconduct may involve any number of other offenses including negligence,
deception, cover-ups of misconduct, reprisals against whistleblowers, malicious allegations of
misconduct in science, and even violations of due process in the handling of misconduct
("On Being a Scientist," 1998). Other categories of misbehavior may include sexual or other forms
of harassment, misuse of funds, gross negligence in professional activity, and/or tampering with the
experiments of others or with instrumentation and violations of government research regulations.
Legendary Research Scams
Most famous cases of research fraud have taken
place in medicine and the physical sciences, but two
cases in the social sciences have raised considerable
controversy: the "Piltdown Hoax" and the "Tasaday
Hoax." The Piltdown Hoax was perpetrated in
England in 1911 with the claim that the fossil
remains of the evolutionary "missing link" between
apes and humans had been discovered. The
unearthed bones had some features of humans and
some features of apes. In the 1950s, researchers
using carbon dating were able to document that the
remains were of recent origin, that the human skull
had simply been combined with portions of the jaw
of an orangutan, and then the entire remains were
treated to appear to be very old in origin (Broad and
Wade, 1983; Weiner, 1955; Spencer, 1990).
A similar outright fraud took place in the
1970s in the Philippines. Television documentaries
were made on, and large-scale media exposure
was given to, the discovery of a lost tribe, the
Tasaday, a peaceful, "Stone Age" community that
had no previous contact with the modern world.
Later investigations revealed that no humans had
probably ever lived on the supposed Tasaday
Island and that the entire hoax had involved
government officials encouraging local peasants
to pretend to be primitives in order to attract
tourists or other publicity (Marshall, 1989).
38 Chapter 2 • Ethics in Criminal Justice Research
Justice, 23-27. Upper Saddle River, N.J.: Pearson/Prentice Hall, 2004. www.researchnavigator.com.)
Plagiarism involves the theft of another person's words or ideas and presenting them as your own. The
advent of the Internet in particular raises problems in that some students mistakenly believe that, because
they can easily copy such material with the click of a mouse, the information does not have to be referenced,
cited, or acknowledged in any other way. Others mistakenly assume that their professors will be unable to
track down the actual sources of information; this is no longer the case since professors now have access to
Web sites that can search the Internet for plagiarized material.
The most blatant forms of plagiarism involve the use of another's work, the purchase of research
papers, or knowingly copying whole sections of another work into a paper without acknowledgment. Simply
changing sentence order or a few words in a passage is not enough-it is still plagiarism.
While you should credit it sources for ideas or words that are not your own, it is not necessary to document
the obvious or common knowledge. Paraphrasing involves restating "in your own words"materials written
or spoken by another person; you reflect the original material in your own words. Merely changing a few
words and the order of phrases is plagiarism, not paraphrasing. In taking notes and collecting sources for
your paper, be sure to keep track of your sources so that you can later properly acknowledge them. Your
professor may require different documentation styles, although most prefer the APA (American Psychological
Association) style (see www.researchnavigator.com). It is crucial that sources be cited and acknowledged
within the text itself. A paper featuring no cited sources followed by a list of works used is not acceptable.
Directly quoted material should be identified by means of quotation marks
THE RESEARCHER'S ROLE
Ethical concerns in criminal justice research raise potential problems for the researcher with
respect to the various roles she or he must often play. The role of researcher as scientist may
intersect with, and sometimes conflict with, the roles of criminal justice practitioner, citizen, and
humanitarian. Rabow (1980), for instance, sees the conflict of scientific and treatment roles in
corrections as hindering the effort to improve and apply treatment successfully.
The role of researcher requires that one be objective and "value free" in approaching and
reporting on the subject matter. As was indicated, particularly in our previous treatment of participant
observational studies of criminals, such a stance often impinges on one's concept of the
proper role of a criminal justice practitioner. The practitioner is involved in programmatic efforts
to prevent, rehabilitate, and otherwise process criminals and/or crime. Such a role obviously
conflicts with the role of neutral observer and scientist. Similar conflicts may take place with
one's role as citizen or humanitarian, wherein one is concerned with cooperating with public
officials or expressing concern and supporting efforts for eliminating inequitable social conditions
or human maladies. In a classic statement, Polsky directs himself to the moral issues of
field studies of criminals:
If one is effectively to study adult criminals in their natural settings, he must
make the moral decision that in some ways he will break the law himself. He need
not be a "participant" observer and commit the criminal acts under study, yet he
has to witness such acts or be taken into confidence about them and not blow the
Research Methods in Criminal Justice and Criminology, Eighth Edition, by Frank E. Hagan. Published by Prentice Hall. Copyright © 2010 by Pearson Education,Inc.
Chapter 2 • Ethics in Criminal Justice Research 39
whistle. . . . According to Yablonsky, nonmoralizing on the part of the researcher,
when coupled with intense interest in the criminal's life, really constitutes a
romantic encouragement of the criminal. . . . [Polsky feels] the burden of proof
rests upon those who claim that abstention from moralizing by the field investigator
has any significantly encouraging effect on criminal's lifestyles, and they
have not supplied one bit of such proof. Finally, our society at present seems
plentifully supplied with moral uplifters in any case, so one needn't worry if a
few sociological students of crime fail to join the chorus. (Polsky, 1967, p. 139)
The National Advisory Committee on Criminal Justice Standards and Goals addresses the
intersection of researcher as scientist, researcher as criminal justice practitioner, and researcher
Criminal justice researchers who are funded by, work closely with, or are employees
of agencies whose functions include law enforcement can encounter ethical problems
when they appear to assist in law enforcement activities. Although most researchers
would support the objective of enhancing the effectiveness of the criminal justice
system and recognize their duties as citizens to do so, the progress of research may
nonetheless be undermined by failure to distinguish between their roles as researcher
and the roles of other criminal justice personnel. The burden of maintaining this
distinction falls on both researchers and agencies, but researchers who study any type
of organization should guard against having to assume any nonresearch roles or even
appearing to do so. (National Advisory Committee, 1976, p. 131)
In self-mediating the potential conflicting roles of the criminal justice researcher, it is
incumbent upon the investigator to enter the setting with eyes wide open. A decision must be
made beforehand on the level of commitment to the research endeavor and the analyst's ability to
negotiate the likely role conflicts. Although there are no hard and fast rules and each research
enterprise is in many ways a unique reality, the researcher's primary role is that of scientist. That
is not to say that this role should in all cases take total precedence over other agenda; however,
the researcher should determine limits, priorities, and subject accountability as soon as practicable
in embarking on a study (Reynolds, 1982; Punch, 1986).
RESEARCH TARGETS IN CRIMINAL JUSTICE
Criminal justice research focuses on a variety of targets or subject matter: the criminal, the
victim, the criminal justice system, and practitioners, as well as the general public. Each of
these topics raises unique ethical problems or concerns for the investigator. One group that
has been a traditional source of research subjects and of increasing controversy has been the
A national prison research commission report questioned the legitimacy and ethicality of
prison research in the United States. They doubted whether prisoners' voluntary decisions to
participate in studies reflect volunteerism or fear that not to "volunteer" would bring reprisals.
Instead of banning all such investigations, the committee called for a review by outside boards
that would be made up of various constituencies including prisoners, prisoner advocates, and
representatives of racial and cultural minorities (Branson, 1977).
40 Chapter 2 • Ethics in Criminal Justice Research
ETHICS AND PROFESSIONALISM
Criminologists and criminal justice researchers as social scientists strive for acceptance as
professionals. The regulation of ethically acceptable research conduct may take one of three forms:
Codes of ethics and institutional review boards, which are adopted by professional associations
or institutions doing research (such as universities).
Procedures imposed by the federal government (primarily to regulate biomedical research,
but with bearing on social research).
Legal regulation in the courts (Reynolds, 1982, p. 100).
Following what has been identified as the "classic professionalism model," occupations
and occupational incumbents attempt to convince the public, lawmakers, and other professionals
that they are deserving of high respect, prestige, autonomy, privilege, and remuneration on the
basis of two key elements (Hagan, 1975). The first is that the occupation begins to generate its
own esoteric and useful knowledge. Many of the methodological issues addressed in this text are
illustrative of major steps that have been taken in this direction. Unless criminologists and criminal
justice researchers view themselves as simply efficient, bureaucratic technocrats or social
accountants, it is essential that its incumbents ascribe to a code of ethics, a dedication to service
and service (ethics) dimensions, occupations may claim or be granted autonomy or high professional
regard. Public askance of claims by car sales personnel, insurance agents, morticians,
florists, and the like basically questions the relevance or applicability of this model to all who
aspire. Criminologists and criminal justice investigators, unless they wish to be regarded in the
same league as used car salespeople with a gimmick, must encourage the highest of ethical ideals
not only in dealing with clients, but also in conducting research. Furthermore, such regulation of
conduct must be mandated from within the profession, rather than solely being imposed by
outside government funding agencies (Hagan, 1975).
Some take strong issue with this view and see professional ethics as "a deceit and a snare"-
a means by which the establishment within an occupation can control and hide its activity from the
public and thereby create a monopoly. Even if the occupation takes its self-policing seriously, it is
used as a club to control deviance of the more creative nonestablishment members whose new ideas
are vital, particularly for young professions (Douglas, 1979, p. 13). Douglas feels that ethical rules
are created for outside public consumption, so that an occupation can gain a monopoly, and have
little impact within the group. On the other hand, this view may be overly cynical because, if a
group refuses to set its own standards, whether rigid or flexible, it is solely at the mercy of outside
regulatory groups (such as government agencies) to set standards for it. Douglas seems to view
codes of ethics as a pincer attack on field studies by quantitative researchers within and government
bureaucrats without. Revisions of the HHS guidelines, to be discussed shortly, may, in part, have
calmed these fears.
ETHICS IN CRIMINAL JUSTICE RESEARCH
In 1998, the Academy of Criminal Justice Sciences began drafting a code of ethics which included
a research code of ethics. The American Society of Criminology developed a similar draft that
year. This presentation is given very much in the same spirit used by the National Advisory
Criminal Justice Research.
Chapter 2 • Ethics in Criminal Justice Research 41
The intent . . . is not to propose a rigid set of guidelines for each researcher to follow.
Rather, the principles and recommendations call attention to contemporary issues that
neither policymakers nor researchers may have considered in a systematic manner.
The application of these principles and recommendations must be tailored to the
needs of each individual research project according to the unique conditions that
surround it. (National Advisory Committee, 1976, p. 38)
The last statement cannot be emphasized enough. Much of the academic guerrilla warfare
taking place regarding ethical codes is a reaction to the Orwellian "big brother" approach in
which dictating and controlling the research enterprise is common. Rigid commandments will
only invite subterfuge and hypocrisy. Reynolds (1982, p. 103) indicates that "the development
of federal procedures for prior review of research with human participants is dramatic evidence
of the failure of associations to convince the public that their members are to be trusted as
individuals or that the associations are to be trusted to control them."
Historically, the most important source of guidance for ethical research in the United States
Policy on Protection of Human Subjects (1971), which requires that any grant recipients abide
by its stipulations. Since 1980, the most important source has been that of the HHS. Both the
codes of ethics of professional associations (internal controls) and federal requirements (external
controls) are constantly changing.
Until revision of the HHS guidelines in 1981, bitter debate took place between the social science
research community and federal officials with respect to the applicability of informed consent
requirements (at that time, HEW guidelines) to much of social science research.
Initiated in the 1960s and eventually extended to all federally funded research, the guidelines
required the informed consent of research participants as well as prior review by IRBs weighing
the costs to participants versus the benefits to science and society. Although each agency, including
the Department of Justice, had its own separate requirements, the HEW (now HHS) procedures
were the most developed and tended to be adopted by the other agencies (Reynolds, 1982, p. 104).
Institutional Review Boards (IRBs) are research screening committees set up in colleges and
universities to oversee the ethical propriety of research.
In a sense, the original HEW guidelines were comparable to a researcher's "Miranda
warning"-basic information that must be assured with respect to the consent of research
subjects. The original 1971 HEW guidelines contained six elements for obtaining informed
consent (Code of Federal Regulations, 1975, pp. 11854-11858):
1. A fair explanation of the procedures to be followed, and their purposes, including identifications
of any experimental procedures.
2. A description of any attendant discomforts and risks that can be expected.
3. A description of any benefits reasonably to be expected.
4. A disclosure of any appropriate alternative procedures that might be advantageous for the
5. An offer to answer any inquiries concerning the procedures.
6. Instruction that the person is free to withdraw consent and discontinue participation at any
time without prejudice to him or her.
42 Chapter 2 • Ethics in Criminal Justice Research
Because HEW supported both physical (particularly biomedical) and social science
research, these rules originally applied equally. Some critics suggested that this marriage produced
unworkable requirements, particularly when applied to field research.
The requirement that sponsored research include provisions for making the subjects aware
of the intentions of the study and sign informed consent forms obviously represented problems in
field research of deviants. Weppner saw two problems that this raised for Street Ethnography
(1977, p. 41): The first is the possible desire of subjects, if engaged in deviant or illegal activity, to
remain anonymous; and second, the difficulty of gaining informed consent from others who turn
up on-the-scene but are not the primary subjects of observation. He felt that a strict interpretation
of HEW guidelines would make street ethnography impossible and that the ability of subjects to
withdraw might also destroy random samples in other types of research. O'Connor, in reviewing
revised recommendations of HEW guidelines, also felt that, "If the recommendations of the
Commission are accepted as they stand, the covert observation debate, long an issue of contention
within social research, will have been settled" (1979, p. 253). The settlement would be a ban
against such research. Talarico (1980, p. 207) views the issue as a two-edged sword in that important
concerns for privacy may protect influential system officials from needed investigation.
Should criminal justice become a subject that studies solely volunteers?
Assuming that the specter of criminal justice researchers as mad Nazi scientists is misplaced
and that professional investigators do not go out of their way to harm or threaten the well-being of
respondents, some privacy invasion is necessary. Sometimes, investigative research strategies
involve deceit and infiltration as necessary approaches to studying hidden behavior. This intrusion
is tolerable if potential harm to respondents is avoided and if the researcher takes adequate steps to
assure that the respondents' identity is protected in any publications (Douglas, 1978). Such protection
may include, as Denzin suggests, the decision to delete or not report certain portions. Publication is
delayed until the subjects have left the scene, and thus the information is no longer threatening to
their status (Denzin, 1989, p. 336).
Informed consent was viewed as applicable in controlled biomedical and psychological
experimentation where researchers have definite plans for their subjects, but in fieldwork, the issue
was seen as less straightforward. Wax (1980, pp. 275-276) identifies six paradoxes of consent as
applied to the practice of fieldwork:
1. Many people studied may be semiliterate and not accustomed to the legal argot of forms.
2. Many will distrust a situation requiring their endorsement of a piece of paper.
3. Consent of subjects is a continual process dependent on mutual learning and evolution.
4. Knowing nothing of ethnography, they have no basis upon which to decide to give or not to
5. Ethnography involves observation and discussion and not a rationalistic a priori analysis.
6. Fieldwork is an evolving process; thus the subjects of investigation are likely to shift
during the course of study.
A related question is "Are all subjects equally deserving of informed consent?" Public figures
and institutions can be observed by anyone and are less vulnerable than private citizens in private
places (Thorne, 1980). The right to privacy may not apply to Hitler, Stalin, the Ku Klux Klan
(KKK), or Murder Incorporated, (Fichter and Kolb, 1953). Galliher very lucidly makes this point:
While all people may be worthy of the same respect as human beings, it does not
necessarily follow that their activities merit the same degree of protection and
respect. As indicated earlier, Lofland questioned possible prohibitions on the
undercover study of fascist groups. It is questionable whether the files of the
Chapter 2 • Ethics in Criminal Justice Research 43
American Nazi Party are deserving of the same respect as any other data source;
must one secure the active cooperation of the Ku Klux Klan, or for that matter of
the Pentagon, before conducting research in their organizations or with their
personnel? While doing research in South Africa, van den Berge concluded
"From the outset, I decided that I should have no scruples in deceiving the
government. . . ." The question is, how much honor is proper for the sociologist
in studying the membership and organization of what he considers an essentially
dishonorable, morally outrageous, and destructive enterprise? Is not the failure
of sociology to uncover corrupt, illegitimate, covert practices of government or
industry because of the supposed prohibitions of professional ethics tantamount
to supporting such practices. (Galliher, 1973, p. 96)
Application of the same rule intended to protect the powerless from powerful institutions is
misdirected (Galliher, 1980, p. 305). Applications of HEW standards to Woodward and
Bernstein's Watergate investigations would have changed the course of American history
(O'Connor, 1979, p. 264).
Reaction to concerns regarding written informed consent led to the National Research Act of
1974, which created the National Commission for the Protection of Human Subjects (NCPHS).
Ethical Principles and Guidelines for the Protection of Human Subjects of Research, in which it
proposed altering the role of IRBs so as not to interfere with the investigator's freedom of
research and recommended alteration of the informed consent in the case of field research
(HEW, 1978a). Although distinctions were made between biomedical and social science
research, the same restrictions on medical research in the Belmont Report were still used in overseeing
social science research.
The Belmont Report (National Commission for the Protection of Human Subjects, 1978a)
called for the recognition of three basic principles:
the principle of respect for persons
the principle of beneficence
the principle of justice
According to the principle of respect for persons, individuals are to be treated as autonomous
agents, and if autonomy is diminished, they are entitled to protection. The principle of respect for
persons is realized through informed consent. The principle of beneficence requires that research
not harm subjects and that possible benefits be maximized and potential harm minimized. This is
implemented through risk-benefit assessment. Finally, the principle of justice asks that both the
benefits and burdens of research be distributed equitably through the selection of subjects
(O'Connor, 1979, p. 229).
Perhaps the final resolution of the application of HEW regulations to social and behavioral
research took place in January 1981 with the publication of new regulations by the HHS. Taking
into account most of the criticisms by social scientists of the original regulations, it virtually
excludes most social science research from the regulations. The types of research excluded from
the review requirements are survey or interview procedures, observations of public behavior, and
the use of existing data if these are not linked to identifiers ("Regulations on the Protection of
Human Subjects," 1981).
44 Chapter 2 • Ethics in Criminal Justice Research
Since the Belmont Report, most of the decisions regarding protection of human subjects in
research are under the purview of IRBs. Any research conducted within the jurisdiction of the
university, its faculty, administration, or students is subject to its oversight. These committees
consist of professors as well as professionals from the community who check research proposals
in order to assure that no harmful procedures are employed, that participants are not in harm's
way and are availed informed consent, and that subjects' privacy and confidentiality are protected.
Studies should not begin until they have been approved by the IRB. Even though much social
science research is exempt, it must still be reviewed. In such circumstances, the proposal is
usually given an expedited review.
Berg (2007, p. 65) points out some of the controversy related to IRBs' expansion of their
authority and the potentially challenging academic freedom in research. Initially created to review
consent agreements to protect human subjects in federal government-funded research, over time
IRBs expanded their mandate to include all research in institutions funded or unfunded. The gauntlet
of obtaining IRB approval before undertaking a research project may have forced some to throw
in the towel and not pursue some research projects. Perhaps, IRBs became too big for their britches.
In response to the Belmont Report as well as criticism by social scientists of strict federal
regulations designed to regulate biomedical research, major changes in the HHS guidelines and
federal regulation of research took place on January 26, 1981 (Federal Register 46, no. 16,
8366-8392). These changes dramatically reduced HHS review over most social science research
and placed the actual decisions involving studies in the hands of IRBs-committees at the
researcher's home institution. Principal changes included the following:
The regulations apply only to research with human subjects conducted with HHS or
supported fully or in part by HHS funds.
Most areas of social science and criminal justice research, for example, most field studies,
are exempt from the regulations.
Many projects (particularly routine biomedical research) are now qualified for "expedited"
review, usually approval by only one member of the IRB.
The new federal regulations were viewed as minimal standards, and local IRBs could require
higher standards if they wished. The new guidelines acknowledge that HHS has no jurisdiction
over research receiving no federal funding.
Figure 2.2 illustrates the procedures one must address in order to have one's research
proposal approved by an IRB.
The actual procedures will vary, of course, by institution. Most are posted on college or
university Web sites. While admitting that IRBs have done much to protect human research subjects,
some feel that we may have unnecessarily handicapped researchers who pose little harm to
subjects and that we should exempt such low-risk research from federal regulations (Shamoo,
2007, p. B16). Current practice may accomplish this in part through "expedited review," in which
a member of the IRB may give quick approval of such low-risk research. The American
Association of University Professors' Committee on Academic Freedom and Tenure issued a
report in 2006 indicating that the IRBs may have overextended their reach into areas such as oral
history, cultural anthropology, and journalism, which involve little risk. Such regulation of
studies may violate the First Amendment by restraining a faculty member's free speech
(Shamoo, 2007, p. B16). The committee also indicated that regulation of studies involving public
data and undergraduate class projects should also be exempt.
Date Submitted: Advisor's Name (if applicable):
Investigator(s): Advisor's Signature of Approval:
Check here if advisor has approved research
Address: Title of Research Project:
E-mail: Date of Initial Data Collection:
Please describe the proposed research and its purpose, in narrative form:
1. Do you have external funding for this research (money coming from outside the College)?
Funding Source (if applicable): _________________________________________________________
2. Will the participants in your study come from a population requiring special protection; in other
words are your subjects someone other than Mercyhurst College students (i.e., children 17 years old
or younger, elderly, criminals, welfare recipients, persons with disabilities, NCAA athletes)?
If your participants include a population requiring special protection, describe how you will obtain consent
from their legal guardians and/or from them directly to insure their full and free consent to participate.
Indicate the approximate number of participants, the source of the participant pool, and recruitment
procedures for your research:
Will participants receive any payment or compensation for their participation in your research
(this includes money, gifts, extra credit, etc.)? Yes No
If yes, please explain:
3. Will the participants in your study be at any physical or psychological risk (risk is defined as any
procedure that is invasive to the body, such as injections or drawing blood; any procedure that
may cause undue fatigue; any procedure that may be of a sensitive nature, such as asking
FIGURE 2.2 Mercyhurst College, Institutional Review Board Research Proposal. (continued)
46 Chapter 2 • Ethics in Criminal Justice Research
questions about sexual behaviors or practices) such that participants could be emotionally or
mentally upset? Yes No
Describe any harmful effects and/or risks to the participants' health, safety, and emotional or social
well-being, incurred as a result of participating in this research, and how you will insure that these risks
will be mitigated:
4. Will the participants in your study be deceived in any way while participating in this research?
If your research makes use of any deception of the respondents, state what other alternative (e.g.,
non-deceptive) procedures were considered and why they weren't chosen:
5. Will you have a written informed consent form for participants to sign, and will you have appropriate
debriefing arrangements in place? Yes No
Describe how participants will be dearly and completely informed of the true nature and purpose of the research,
Please include the following statement at the bottom of your informed consent form: "Research at
Mercyhurst College which involves human participants is overseen by the Institutional Review
Board. Questions or problems regarding your rights as a participant should be addressed to Dr. Terry
F. Pettijohn; Institutional Review Board Chair; Mercyhurst College; 501 East 38th Street, Erie,
Pennsylvania(NNN) NNN-NNNN Telephone (XXX) XXX-XXXX."
6. Describe the nature of the data you will collect and your procedures for insuring that confidentiality
is maintained, both in the record keeping and presentation of this data:
7. Identify the potential benefits of this research on research participants and humankind in general.
Please submit this file and accompanying materials to the IRB Chair, Terry Pettijohn, via electronic mail
Source: Reproduced with permission of the Mercyhurst College Institutional Review Board.
Research activities exempt from HHS review include research in educational settings related
to normal educational practices, such as curriculum strategies and instructional techniques.
Also exempt is research using educational tests, as long as confidentiality is maintained. Most
important to social science researchers is the general exemption for research involving survey
or interview procedures, except where all of the following conditions exist:
Responses are recorded in such a manner that the human subjects can be identified, directly
or through identifiers linked to subjects.
The subject's responses, if they became known outside the research, could reasonably place
the subject at risk of criminal or civil liability or be damaging to the subject's financial
standing or employability.
The research deals with sensitive aspects of the subject's own behavior such as illegal
conduct, drug use, sexual behavior, or use of alcohol. All research involving survey or
interview procedures is exempt, without exception, when the respondents are elected
or appointed public officials or candidates for public office.
Similarly, research involving the observation (including observation by participants) of
public behavior is exempt, except where all of the following conditions exist:
Observations are recorded in such a manner that the human subjects can be identified,
directly or through identifiers linked to the subjects.
The observations recorded about the individual, if they became known outside the
research, could reasonably place the subject at risk of criminal or civil liability or be
damaging to the subject's financial standing or employability.
conduct, drug use, sexual behavior, or use of alcohol.
Research involving the study of existing data, documents, records, pathological specimens,
or diagnostic specimens is exempt, if these sources are publicly available or if the information is
recorded by the investigator in such a manner that subjects cannot be identified, directly or
through identifiers linked to the subjects (Federal Register, 1981, pp. 386-388).
The specific areas identified for expedited review had relevance primarily to biochemical
rather than social science research, for example, the use of medical and dental diagnostic
equipment. The one area that had some relevance was studies of perception, cognition, game
theory, or test development in which there is little stress placed on, or manipulation of, subjects
(Federal Register, 1981, p. 8392).
Projects that are not exempt still require full review, in which case the IRB must weigh the
cost and benefits of such studies.
[T]he IRB is expected to consider the extent to which risks to the participants are
minimized, the relationships between risks and anticipated benefits of research, and
the importance of the knowledge to be developed. The IRB will also consider the
equitable selection of subjects and the acquisition and documentation of informed
consent, it will monitor participants when necessary to ensure their safety and make
provisions for protecting their privacy. If participants are vulnerable to coercion
(such as mental patients and children), additional safeguards should be considered.
(Reynolds, 1982, p. 106)
48 Chapter 2 • Ethics in Criminal Justice Research
The "Elements of Informed Consent" applicable to nonexempt programs detail specific
guarantees to subjects required of all projects, additional requirements for more hazardous
projects, exceptions for evaluations of existing public programs, and exceptions for "deceptive
research" in which subjects are informed afterward so as not to destroy the scientific purposes of
the project (Federal Register, 1981, pp. 8389-8390). These are expanded requirements of the
original HEW guidelines discussed previously. There are fewer objections by social researchers
now, because most of their projects are exempt.
In reviewing federal regulation of research, the National Research Act of 1974 had created
IRBs to monitor research, but ambiguous guidelines from the HEW and the utilization of
biomedical guideposts created a storm of protests from social scientists. The revised regulations
represented a victory of sorts for the social sciences.
Initially, the withdrawal of some government regulation over many areas of social research
was a recognition that professional groups will regulate their own ethical conduct. However,
giving much of the regulatory power to IRBs may have opened yet again a "Pandora's box."
Ferrell and Hamm (1998, p. xiv) indicate:
Beginning in the late 1970s, but not fully taking hold until the 1990s, Institutional
Review Boards (IRBs) at most colleges and universities have made ethnographic
work on criminal and deviant groups almost impossible to conduct. Even the new
Code of Ethics of the American Sociological Association yields to the decisions of
these boards, claiming that if projects are disapproved by these agencies, the
research, in the association's eyes, is unethical. Potentially gone, then, is any
ethnographic research involving a covert role for the investigator (thus removing
hidden populations further from view), any ethnographic research on minors that
does not obtain parental consent (obviously problematic for youth involved in
deviance and crime or who are victims of parental abuse), and any ethnographic
research on vulnerable populations or sensitive (including criminal) issues without
signed consent forms that explicitly indicate the researcher's inability to protect
Ferrell and Hamm (1998, p. xv) conclude that government forces do not believe that
researchers can police themselves.
National Institute of Justice's Human Subject Protection
NIJ's Regulations on Confidentiality protect individuals by forbidding the use of any
research or statistical information that might identify them. In addition, the Institute has
adopted the HHS Model Policy on Human Research Subjects. This policy requires that each
institution engaged in NIJ research provide written assurance that it will comply with these
regulations as codified at 45 Code of Federal Regulations (CFR) 46. Pursuant to that policy,
each research project falling within the guidelines established by HHS must be approved by
the recipient's IRB prior to initiation of the project. Approval by the IRB need not precede the
submission of a proposal to NIJ but it must be obtained prior to the beginning of any research
activity (NIJ, 1994). Applicants should file their plans to protect sensitive information as part
of their proposal. Necessary safeguards are detailed in 28 Code of Federal Regulations
Chapter 2 • Ethics in Criminal Justice Research 49
paragraph 22. A short "how-to" guideline for developing a privacy and confidentiality plan can
be obtained from NIJ program managers.
In addition, the U.S. Department of Justice has adopted Human Subjects policies similar
to those established by the U.S. Department of Health and Human Services. In general, these
policies exempt most NIJ-supported research from IRB review. However, the Institute may
find in certain instances that subjects or subject matters may require IRB review. These
exemptions will be decided on an individual basis during application review. Researchers are
encouraged to review 28 CFR 46, paragraph 46.101 to determine their individual project
requirements (NIJ, 1994, p. 30).
The federal policy on human subjects that formerly applied to HHS research (45 CFR 46, A)
has now been adopted by seventeen federal agencies including the Justice Department (28 CFR 46,
512, and 22). Subpart A, known as the Common Rule, may be found at www.hhs.gov/ohrp/policy/
under the category "policy guidance." It requires that informed consent include a statement about
how the researcher will maintain confidentiality (Sieber, 2001, p. 2).
CONFIDENTIALITY OF CRIMINAL JUSTICE RESEARCH
Regulations on the confidentiality of research and statistical data were enacted as part of the
1973 amendment to the Omnibus Crime Control and Safe Streets Act. Section 524(a) reads:
Except as provided by Federal law other than this title, no officer or employee of
the Federal Government, nor any recipient of assistance under the provisions of
this title, shall use or reveal any research or statistical information furnished
under this title by any person and identifiable to any specific private person for
any purpose other than the purpose for which it was obtained in accordance with
the title. Copies of such information shall be immune from legal process, and
shall not, without consent of the person furnishing such information, be admitted
as evidence or used for any purpose in any action, suit, or other judicial or administrative
proceedings. (LEAA, 1979, p. 1)
This law basically constitutes a "shield law" for researchers performing federally funded
research. Shield laws constitute a governmental immunity from prosecution-a state-guaranteed
right to confidentiality for researchers if they are subpoenaed. "All identifiable research or
statistical information is with limited exceptions, immune from administrative or judicial
process" (Dahmann and Sasfy, 1982, p. 13). Investigators may be encouraged to probe more
sensitive topics because they are able to protect their data. This law also protects respondents
by ensuring that the data they have provided will not be used to invade their privacy. Guidelines
such as those of the HHS or NIJ are, of course, established, issued, and promulgated by federal
agencies, that is, governmental bodies outside of the occupation. They are intended for data
gathered under their auspices or sponsorship. Of primary concern is the fact that the Freedom of
Information Act of 1976 makes it possible for individuals to obtain access to nonclassified information
that is collected with public funds. This could include field data that could compromise
confidentiality. Trend, for instance, describes a project in which he was involved where, after the
fact, the General Accounting Office (GAO) requested "unrestricted access to certain records"
he and others had gathered as part of a Housing and Urban Development project. They wanted
the case files so that an audit could check family sizes and incomes as part of an eligibility study
50 Chapter 2 • Ethics in Criminal Justice Research
(Trend, 1980, p. 344). Eventually, a compromise was worked out that allowed a GAO audit while
maintaining confidentiality. Trend, as a result of his experience, offers the following advice:
Telling people to read the contract they sign, to know how far they're willing to go if
pressed, and to not make promises they cannot keep all seems pretentious and sappy. . . .
In the end, the best advice I can give is to not be gulled into thinking that your notes are
sacrosanct and nobody can get to them no matter what. If you're sure of your contract,
your client, and your own resolve, then promising confidentiality in writing may increase
the chances that you can maintain it. However, if none of those prerequisites obtain, then
the time to stop and think is before you start the research. (Trend, 1980, p. 348)
In 1972, the U.S. Supreme Court ruled that journalists had no right to refuse to name their
sources. Despite this, as of 2007, thirty-three states and the District of Columbia had enacted shield
laws for journalists (Kiely, 2007, p. 5A). Without guarantees of confidentiality, the likelihood that
whistleblowers and informants will provide important information is less. In many cases, prosecutors
have tried to turn journalists into witnesses and to break their promises of confidentiality.
Confidential information was instrumental in revealing abuses at XXXXX XXXXX Hospital, torture at
Abu Ghraib prison, Watergate, Enron, and drug scandals in professional sports. In 2007, the U.S.
House of Representatives passed legislation prohibiting courts and federal prosecutors from forcing
journalists to violate confidentiality, except in cases vital to national security or in prosecuting a
crime when proving guilt is not possible by any other means (When Reporters Can't Shield, 2007,
p. 12A). As of July 2008, the House of Representatives had passed this federal shield law legislation.
The U.S. Senate and President George W. Bush opposed such legislation even though the House bill
still requires reporters to disclose sources in cases of national security.
In addition to governmental regulations, more fully developed professions attempt to
police themselves and to establish their own guideposts for ethical conduct. In 1998, the
Academy of Criminal Justice Sciences adopted its code of ethics for research. See Exhibit 2.4.
Codes of Research Ethics of the Academy of Criminal Justice Sciences (ACJS)
Objectivity and Integrity in the Conduct
of Criminal Justice Research
1. Members of the Academy should adhere to
the highest possible technical standards in
2. Since individual members of the Academy vary
in their research modes, skills, and experience,
they should acknowledge the limitations that
may affect the validity of their findings.
3. In presenting their work, members of the
Academy are obliged to fully report their findings.
They should not misrepresent the findings
of their research or omit significant data.
Details of their theories, methods, and research
designs that might bear upon interpretations of
research findings should be reported.
4. Members of the Academy should fully report
all sources of financial support and other
sponsorship of the research.
5. Members of the Academy should not make
any commitments to respondents, individuals,
groups, or organizations unless there is full
intention and ability to honor them.
6. Consistent with the spirit of full disclosure
of method and analysis, members of the
Academy, after they have completed their own
EXHIBIT 2.4 (Continued )
analyses, should cooperate in efforts to make
raw data and pertinent documentation available
to other social scientists, at reasonable
costs, except in cases where confidentiality, the
client's rights to propriety information and
privacy, or the claims of a field worker to the
privacy of personal notes necessarily would be
7. Members of the Academy should provide
adequate information, documentation, and
citations concerning scales and other measures
used in their research.
8. Members of the Academy should not accept
grants, contracts, or research assignments
that appear likely to violate the principles
enunciated in this Code and should disassociate
themselves from research when
they discover a violation and are unable to
9. When financial support for a project has been
accepted, members of the Academy should
make every reasonable effort to complete the
proposed work on schedule.
10. When a member of the Academy is involved
in a project with others, including students,
there should be mutually accepted explicit
agreements at the outset with respect to
division of work, compensation, access to
data, rights of authorship, and other rights
11. Members of the Academy have the right to
disseminate research findings, except those
likely to cause harm to clients, collaborators,
and participants, those which violate formal
or implied promises of confidentiality, or
those which are proprietary under a formal or
Disclosure and Respect of the Rights
of Research Populations by Members
of the Academy
12. Members of the Academy should not misuse
their positions as professionals for fraudulent
purposes or as a pretext for gathering intelligence
for any individual, group, organization,
13. Human subjects have the right to full disclosure
of the purposes of the research as early as
it is appropriate to the research process, and
they have the right to an opportunity to have
their questions answered about the purpose
and usage of the research.
14. Subjects of research are entitled to rights of
personal confidentiality unless they are waived.
15. Information about subjects obtained from
records that are open to public scrutiny
cannot be protected by guarantees of privacy
16. The process of conducting criminal justice
research should not expose respondents to
more than minimal risk of personal harm,
and members of the Academy should make
every effort to ensure the safety and security
of respondents and project staff. Informed
consent should be obtained when the risks
of research are greater than the risks of
17. Members of the Academy should take culturally
appropriate steps to secure informed
consent and to avoid invasions of privacy. In
addition, special actions will be necessary
where the individuals studied are illiterate,
under correctional supervision, minors, have
low social status, are under judicial supervision,
have diminished capacity, are unfamiliar
with social research, or otherwise occupy
a position of unequal power with the
18. Members of the Academy should seek to
anticipate potential threats to confidentiality.
Techniques such as the removal of direct
identifiers, the use of randomized responses,
and other statistical solutions to problems of
privacy should be used where appropriate.
Care should be taken to ensure secure storage,
maintenance, and/or destruction of
19. Confidential information provided by research
participants should be treated as such by
members of the Academy, even when this
information enjoys no legal protection or
privilege and legal force is applied. The obligation
to respect confidentiality also applies to
52 Chapter 2 • Ethics in Criminal Justice Research
ETHICAL ISSUES IN CRIMINOLOGY/CRIMINAL JUSTICE RESEARCH
From what has been thus far indicated, there is no limit to ethical issues confronting criminal
justice researchers (Adamitis and Haghighi, 1989). Criminal justice researchers adhere to most
of these principles as a matter of professionalism, although as we will see, rigid adherence to
a checklist is simplistic and does a disservice to the complexity of the research experience. These
guidelines are often difficult to interpret in individual cases. The codes of ethics include that the
criminal justice researcher take personal responsibility to:
Avoid procedures that may harm respondents.
Honor commitments to respondents and respect reciprocity.
Exercise objectivity and professional integrity in performing and reporting research.
Protect confidentiality and privacy of respondents.
harmful to subjects, particularly if the potential harm has not been explained to the subjects and
their informed consent elicited. The National Advisory Committee on Criminal Justice
Standards and Goals suggests that participants give their formal consent to serve as subjects,
based on full knowledge of the experiment (National Advisory Committee, 1976, p. 38).
It is the researcher's duty to assume personal responsibility for all phases of the project as
it may potentially impinge on the well-being of subjects. If, after considering the ethicality of
research, problematic areas still exist, the investigator should seek out advice from university,
professional, or governmental committees to ensure adequate safeguards. Failure of a researcher
to obtain informed consent or give full disclosure of the study to respondents increases the need
to safeguard confidentiality. Thus, when deception is necessary in a study, it becomes more
incumbent on the people conducting the study to prevent harm and where appropriate debrief,
reassure, and explain the project afterward to subjects (Lee, 1993).
The informed consent issue is a complex one, particularly in correctional research. As
experimentation entails unequal or different treatment of experimental and control groups, is the
unequal treatment justified for research or scientific purposes? There is a major issue of
informed consent in prison research. Communication beforehand as to who is in the treatment
and who is not creates "reactivity" or "Hawthorne effects." There are studies in which the aim is
to hide the treatment in order not to arouse anxiety.
members of research organizations (interviewers,
coders, clerical staff, etc.) who have access
to the information.
20. While generally adhering to the norm of
acknowledging the contributions of all collaborators,
members of the Academy should be
sensitive to harm that may arise from disclosure
and respect a collaborator's need for
21. All research should meet the human subjects'
requirements imposed by educational institutions
and funding sources.
22. Members of the Academy should comply
with appropriate federal and institutional
requirements pertaining to the conduct of
Source: The full codes of ethics are available on the
Association's Web site: http://www.acjs.org/
Association's Web site: www.ACJS.ORG ISBN 0-558-58864-6
Chapter 2 • Ethics in Criminal Justice Research 53
The protection of human subject guidelines is also viewed as proposing a risk-benefit ratio,
wherein the potential benefits must outweigh the possible hazards to respondents. That is, some
risks are justified as long as the scientific knowledge gained exceeds the potential harm. Douglas
(1979, p. 30) perhaps correctly suggests that in the final analysis the attempt to rationalize a
cost-benefit analysis of research borders on the simplistic in that the conclusion is almost always
going to be ruled in favor of science. Such a decision-making process converts the researcher into a
"moral administrator" who assumes the right to inflict harm on the subjects in the name of
science-once again the mad scientist hangup (Reiman, 1979, p. 45). Although an interesting
philosophical debate, the fact of the matter is that most criminal justice researchers simply have no
interest in being Dr. Frankenstein. Obviously, any research that is likely to impose long-term harm
on participants is anathema to ethical concepts of investigation. In the name of research, one should
not espouse behavior that would not be considered acceptable in normal interpersonal conduct. The
exclusion of most social science research from review by HHS guidelines was a frank recognition
that there has been little documented harm associated with such studies.
Honor Commitments to Respondents and Respect Reciprocity
Researchers have an ethical responsibility to keep any promises or agreements made with
subjects during and after the course of study. The notion of reciprocity involves a mutual trust
and obligation between researcher and subject. The researcher would have been unable to obtain
information without the cooperation of participants who were willing to share of themselves in
the belief that the investigator is obliged not to betray this trust by using the information in an
inappropriate manner or one that may prove harmful or embarrassing to the subject. Klockars
feels that simple researcher-subject models such as in biomedical research do not begin to
capture the complexity of reciprocal obligations in field research.
Vincent was not only my subject but also my teacher, student, fence, friend and
guide. Likewise to Vincent, I was not only researcher but biographer, confidant,
customer, friend, and student. These roles, most of which involve multiple obligations
and responsibilities and expectations, are potentially in conflict not only in
the researcher-subject dimensions but in other dimensions as well. To speak of the
working relationship between the life historian and his subject as a researchersubject
relationship simply misconstrues what happens in the context of life
history work. The researcher who treats his friends as subjects will soon find that
he has neither. (Klockars, 1977, p. 218)
and Reporting Research
Honesty, integrity, and objectivity are essential expectations of ethical professional conduct. The
researcher should attempt to maintain a value-free, politically indifferent approach to the subject
matter. Personal, subjective feelings should be kept separate from a disinterested scientific study
of things as they are. The researcher first and foremost is an investigator, not a hustler, huckster,
salesperson, or politician. Researchers should bar themselves from studying subjects or subject
matter for which they feel they cannot properly control potential subjectivity such as a strong
aversion or affinity toward the object of investigation.
It is important that researchers be frank and honest in conducting their affairs. Researchers
should not misrepresent their research abilities or generalize beyond their data. The researcher,
in addition to having a concern for accuracy, should avoid any statistical misrepresentations of
54 Chapter 2 • Ethics in Criminal Justice Research
findings or purposely choose techniques that are most likely to produce positive results.
Unethical practices or sponsors who attempt to control the outcome should be avoided. Finally,
the researcher has a responsibility to communicate results to professional audiences that are in
the best position to judge the findings. Of course, the investigator should always give proper
acknowledgment to others who assisted in the research.
An example of a fake survey that clearly violated the rules of integrity was conducted by
Francis Flynn of Columbia University's Business School (Foster, 2001). In 2001, Flynn wrote
a letter on department stationery to 250 restaurant owners in New York City indicating that he
had been stricken with food poisoning after eating at their establishments. Flynn claimed to
be conducting a study of how owners respond to polite versus enraged-sounding customer
complaints. When one owner contacted the Dean of the Business School, the project was
canceled and letters of apology sent to the restaurant owners.
Protect Confidentiality and Privacy of Respondents
Erikson takes a very rigid stand in suggesting that the criminal justice researcher should avoid
deliberate misrepresentations of his or her identity when entering the private realm of subjects'
lives which otherwise would be barred to the researcher. He feels that it is generally unethical to
misrepresent the purpose of research (Erikson, 1978, p. 244). Although such an inflexible view is
not always appropriate, its general applicability to much research is warranted. To give a personal
example of naiveté regarding this matter, the author, while doing field research for a graduate
thesis, ineptly handled this privacy matter and almost sabotaged a research project.
In researching life in the early development of the new town of Columbia,
Maryland, the author began attending many community meetings to get a feel of
the temper of the pioneers in the young community. Having spotted an announcement
in the local paper of a meeting of the "Informal Discussion Group," whose
topic was to be the "Economics of New Towns," the author, after consulting with
his research mentor, attended the meeting. As perhaps should have been surmised
by the group's name, the meeting was very "informal" with less than ten people in
attendance and conversation never got around to the scheduled topic. The main
point of discussion revolved around griping by the residents about feeling like they
were living in a glass bowl. Sightseers on weekends created traffic jams outside
their homes; national news magazines had interviewed, at one time or another,
nearly everyone in the room or members of their family. Worse yet, one person
complained that he heard that the developer was bringing in his behavioral scientists
to see if "us rats are running the maze properly." At this point the author began
to feel very uncomfortable as I had failed to announce my purposes beforehand or
ask permission to attend the session. Finally, since everyone but the author had
participated in the discussion, one person indicated that he had not heard anything
from me. I then proceeded to explain that I was a graduate student doing research,
and he (a local councilman) took down my name and affiliation. Not surprisingly,
the meeting ended almost immediately afterwards even though I tried to explain
that I was a harmless researcher and not a spy for the developer. Needless to say
I regretted my naiveté in not making known my presence from the beginning, and
the feeling of having invaded one of the last privileged sanctuaries of the harassed
residents was very much impressed upon me. The very next day a representative
Chapter 2 • Ethics in Criminal Justice Research 55
of the developer (who had obviously heard from the councilman) requested and
had a meeting with the author and his mentor, and to his credit offered to assist in
my research on the community as long as the data gathering would avoid direct
participant observation. A study of the planning process was viewed as far less
obtrusive under the circumstances. (Hagan, 1968, p. 128)
The folly of naively approaching the research setting without making one's presence
known is certainly illustrated by this example. But what about secret observation? Erikson's
stance (1978) would exclude many subjects from inquiry. Roth (1962) suggests that in a sense,
most research design in the social sciences involves a level of deceit, because the exact nature
of the study is often hidden. The revelation of the true purpose would either bias the outcome or
eliminate the possibility of research (Henslin, 1972, p. 48). To answer a question we asked
earlier in this chapter, criminal justice should not restrict its research targets to volunteers.
Confidentiality. All social science researchers, including criminal justice investigators,
have a special obligation to protect the confidentiality of such information.
As indicated previously, confidentiality of government-sponsored research is guaranteed in
Section 524(a) of the Omnibus Crime Control Act of 1973 as amended. In gaining access to confidential
government data, the criminal justice researcher is generally interested in identification
of subjects for sampling and follow-up purposes.
Researchers rarely require records containing the direct identification of individuals
or organizations except during the initial stages of a project when possible personal
identifiers should be deleted or destroyed or if needed for follow-up link files (which
match coder identifiers to personal identifiers) should be stored in a remote and
secure location. (National Advisory Committee, 1976, pp. 42-43)
Government-sponsored research appears to provide safeguards to ensure confidentiality of
data. What of the private researcher?
Unlike priests, doctors, or other client-oriented practitioners, the independent criminal justice
researcher has no legally recognized privilege of confidentiality. Such researchers are then
potentially vulnerable to subpoena. In that sense, social science researchers find themselves in a
situation akin to the journalist in which they must decide whether they would be willing to go to
jail rather than violate confidentiality. Soloway and Walters (1977), in examining the
Pennsylvania Penal Code and decisions on this matter, clarify some of the issues regarding
researcher complicity or culpability. In most instances, the researcher would actually have had to
assist, aid, or abet the actual commission of a specific criminal act to be liable under statutes in
that state. Soloway and Walters proclaimed this warning:
Let us not bask too very long in this unaccustomed legal comfort, for, while such laws
seem to relieve us of responsibility prior to a governmental investigation, they pertain
very little after such an investigation has begun. Once a criminal investigation and/or
prosecution has commenced, we are still absolved of a legal responsibility to come
forward, of our own accord, with our information. Once summoned to so testify,
however, we have no legal recourse but to divulge our information and its source
56 Chapter 2 • Ethics in Criminal Justice Research
under the threat and consequence of a contempt citation. This was made amply clear
in the U.S. Supreme Court's landmark decision in 1972 against Paul Branzburg,
a journalist investigating junkies in Kentucky, and Earl Caldwell, a journalist
investigating the Black Panthers and other militant black organizations. (Soloway and
Walters, 1977, p. 175)
These decisions basically held that a journalist or investigator had no privilege to refuse
appearance before a grand jury or to decline answering questions regarding sources or information
received in confidence. James (1972, p. 139) is the only case Soloway and Walters
could find that involved a scientific investigator being subpoenaed. In this case, the subpoena
was dropped when the investigator produced a previously signed agreement to safeguard the
confidentiality of sources.
Wolfgang (1981, 1982) reports that in his cohort study (Wolfgang et al., 1972) four
respondents admitted involvement in criminal homicide and that seventy-five were involved in
forcible rape for which none of them had been arrested. Members of the research staff could
have theoretically been prosecuted for "misprision of a felony" (being accessories after the
fact), which, although outdated in most states, is a federal offense.
Two anthologies explore this "no man's land" of field research. Ferrell and Hamm's (1998)
Ethnography at the Edge and Miller and Tewksbury's Extreme Methods (2001) examine ethical
dilemmas raised in studying active criminals in their natural settings. Other researchers who have
faced hazards in conducting their research have included: James Inciardi (1993), who was
arrested while conducting research in a crack house and also had a price put on his head by
a crack dealer; Bourgois (1989) was heavily questioned by police during his research on the
crack economy; while Armstrong (1993) was arrested and threatened during his field research on
British soccer fans.
Longmire (1983) in a survey of a sample of members of the American Society of
Criminology found that 63 percent indicated having experienced one or more ethical dilemmas.
With respect to ethical problems impacting on participants, the biggest problem was confidentiality
problems, with 9 percent so indicating. Professional ethical issues were more of a problem,
with 27 percent indicating that they had experienced pressure to engage in undesired research.
This involves primarily academic institutions "twisting the arm" of researchers to pursue research
in areas for which there is grant money.
Although vulnerability to subpoena is an ever-present threat to social science researchers,
Reynolds (1979) was able to find fewer than a dozen such cases and most incidents seemed to
In discussing the run-in with the law of Yablonsky (1968a), author of The Hippie Trip,
Irwin states that:
There has been some concern expressed over the danger of arrest while studying
criminals, mainly because the researcher will have firsthand knowledge of felonies
and misdemeanors. . . . To my knowledge the closest anybody ever came to having
legal sanctions imposed on him because of his research was Lewis Yablonsky. . . .
The judge asked him nine times if he had witnessed Gridley [one of his informants]
smoking marijuana. Yablonsky refused to answer because of the rights guaranteed
him in the Fifth Amendment of the United States Constitution. Although he did not
actually receive any legal sanctioning, he stated that the incident was humiliating
and suggested that researchers should have guarantees of immunity. Despite this
Chapter 2 • Ethics in Criminal Justice Research 57
case, I feel there is small risk of being prosecuted. If we keep our "heads straight"
and avoid being sucked in or bowled over by the criminal world, and thereby do not
slip or plunge into greater complicity than knowledge of crimes, we actually do
have immunity. (Irwin, 1972, pp. 128-129; see also Yablonsky, 1968b)
In 1979, political scientist Samuel Popkin became the first American professor incarcerated
for defending his right to maintain confidentiality regarding The Pentagon Papers. Popkin had
refused to reveal his sources to a federal grand jury and subsequently spent one week in jail until
the jury was dismissed (Wolfgang, 1982, p. 395).
In April of 1984, Judge Jack B. Weinstein of the U.S. Eastern District Court of New York ruled that:
Serious scholars [Mario Brajuha] cannot be required to turn over their fieldnotes in
a grand jury investigation when the government fails to establish a "substantial
need" for them to do so. Weinstein's ruling establishes a "qualified privilege not to
reveal documents or confidential sources" for social science researchers, akin to the
privileges enjoyed by journalists. . . . According to Weinstein, "Serious scholars are
entitled to no less protection than journalists." (Erikson, 1984)
The Weinstein decision cited as support the American Sociological Association's code of
ethics, which indicates that "Confidential information provided by research participants must be
treated as such by sociologists, even when this information enjoys no legal protection or privilege
and legal force is applied" (Erikson, 1984).
The researcher in this case was Mario Brajuha, a sociology graduate student at
SUNY-Stony Brook, who had been subpoenaed by a grand jury investigating a restaurant fire.
He had been doing a participant observation study of the restaurant as a waiter for ten months
prior to the fire and had collected data for his dissertation.
The Weinstein decision was appealed by the prosecutors and in a directive issued by the
court of appeals judge, a compromise of sorts was struck. Brajuha's defense attorney prepared an
edited version of Brajuha's field notes, which was accepted as sufficient to satisfy the original
subpoena and dismissed the case against him. As Brajuha's attorney explains:
From the beginning, Mr. Brajuha was prepared to testify as to his observations and
nonprivileged communications. . . . In the context of this case, a claim of privilege
with respect to (his) observations was unnecessary. The issues in this litigation centered
around the portions of the research journal which contained communications
with privileged sources and matters of personal privacy, for example, opinions.
(Thaler, 1985, p. 1)
A standard procedure for attempting to protect the identity of subjects, organizations, or
communities is the use of pseudonyms (false names) in publications. "Doc," "Chic," "The Lupollo
Family," "Deep Throat," "Wincanton," "Cornerville," "Slumtown," and "Middletown" are but
a few of the many aliases given such subjects, often to little avail. For instance, even the Trobriand
islanders were aware of Malinowski's books and one even indicated that he did not understand their
system of clans and chiefs (Barnes, 1970). Gans in The Urban Villagers (1962) noted that during an
58 Chapter 2 • Ethics in Criminal Justice Research
election in the same West End Boston area Whyte had studied, one of the candidates had received
negative voter reaction based in part on Whyte's description of the individual years ago in
Streetcorner Society (1955).
Despite Klockars' efforts to protect his fence's identity, vanity won the day:
I also told Vincent that I would not reveal his identity unless it meant that I was going
to jail if I did not, and he told me that he really could not expect me to do more.
These contingencies notwithstanding, Vincent just could not resist a little advance
He told everybody-judges, lawyers, politicians, prosecutors, thieves, hustlers,
and most of his good customers. He started this word-of-mouth publicity campaign
a full year before the book appeared in the New York Times Book Review; anyone
who did not know who Vincent Swaggi really was simply confessing that they were
outsiders to the Philadelphia scene. (Klockars, 1977, p. 214)
Vincent even sold and gave away autographed copies in his store, although he turned down
offers to appear on nationally televised talk shows. Threats to researchers may come from many
sources. In the Ofshe case, the threat was from an organization he studied, whereas in the
Hutchinson case it was from a member of Congress.
Organizations or individuals may file lawsuits against researchers. In the 1960s, Synanon
(a drug rehabilitation program headquarters in California) had received widespread, positive,
national publicity. In the early 1970s, Richard Ofshe, professor of sociology at the University
of California at Berkeley, began an investigation of the organization and, along with journalist
colleagues, began to uncover and expose patterns of extreme violence and intimidation
employed by Synanon. In addition to scholarly works and an investigative newspaper series,
Ofshe and colleagues published a book which was later the basis of a CBS television movie
Since 1979, the Synanon foundation has filed three lawsuits charging Ofshe with libel and
slander. Although the cases were eventually dismissed or charges were dropped by the Synanon
litigants, Ofshe was fortunately assisted by the University of California at Berkeley with legal
costs, since the basis of the suits rested on his academic research. In 1987, Ofshe was awarded
over $500,000 for costs of litigation by a Marin County judge in connection with his fight against
the Synanon suits. "Based on a preliminary review, the amount is believed to be one of the most
substantial costs of litigation ever granted to an academic in a case involving pursuit of research
and academic freedom" (Maldonado, 1987, p. 3). Ofshe is continuing to press his legal case,
charging Synanon with malicious prosecution and hoping for a ruling that the suits were a form
of harassment and attack on academic freedom.
During the 1950s, the United States went through a period of "anti-Communist" hysteria led by
Senator Joseph McCarthy of Wisconsin. Many people had their careers and lives ruined by innuendos
that they were, or had been, associated with Communists. During this period, intellectual
debate was intimidated or closed as researchers and others retreated from controversial
topics and the onslaught of political demagoguery. Beginning in the 1970s, a new brand of
Chapter 2 • Ethics in Criminal Justice Research 59
McCarthyism presented itself, although this time the device was public ridicule and the charge
against its targets was that federally funded researchers were "ripping off" the taxpayers' money
on useless studies of "ridiculous" topics.
Ironically, one of the chief engineers of the new McCarthyism was the very man who took
the senatorial seat of the late Joseph McCarthy, the late Senator William Proxmire. Through his
infamous "Golden Fleece of the Month Awards," Proxmire intended to call public attention to
the waste of public tax money on research works whose titles sounded absurd. This ridicule of
specific research impugned the integrity and motivation of the researchers themselves, in particular,
Ronald Hutchinson. Proxmire ridiculed Hutchinson's work as making a monkey out of the
taxpayer, and he incorrectly charged that Hutchinson had personally realized a half-million
dollars in research funds. Even though representatives of the agencies that funded the research
praised it as well-executed and important, Proxmire continued his anti-intellectual attack,
ridiculing Hutchinson's work in the media and pressuring agencies to terminate funding.
Being singled out for the "Golden Fleece award" was a disaster for Hutchinson's professional
and personal life. His children were ridiculed at school and his family was subjected to hate mail and
phone calls. In a law suit brought before the U.S. Supreme Court, a settlement was agreed to in
which Proxmire issued a public apology on the Senate floor and paid Hutchinson $10,000 from his
personal funds. Even though Hutchinson experienced high legal fees, Proxmire's bill of $130,000
was paid by the U.S. Senate (Fund to Protect Scholars from Defamation, 1980).
Important in the Supreme Court's willingness to hear this case was its decision that
research scientists could sue for defamation of character even if they were receiving federal
research funds and that senatorial privilege did not extend to remarks made outside the Senate
Although the Hutchinson case represents the most glaring example of what has been
characterized as the new McCarthyism, more subtle attacks certainly continue. A related concern
is the possibility that criminal justice and other social science researchers will concentrate
inquiry into fundable, safe, societally directed areas and thus avoid controversy or studies that
may uncomfortably focus on "crime in the suites."
Section B(7) of the American Sociological Association (ASA) Code of Ethics states:
"Confidential information provided by research participants must be treated as such by sociologists,
even when this information enjoys no legal protection or privilege and legal force is applied"
(American Sociological Association, 1984, p. 3).
In 1990, Rik Scarce was a Ph.D. sociology student at Washington State University and
published a book entitled Eco-Warriors: Understanding the Radical Environmental Movement
(Scarce, 1990). Using various sources, the book also utilized interviews with members of Earth
First!, the Sea Shepherd Conservation Society, and the Animal Liberation Front (ALF). These
groups had been extensively involved in direct action campaigns and civil disobedience. He had
begun a dissertation on the Radical Environmental Movement. Scarce had been unaware that, as
a student, his research had to be preapproved by the university's IRB. When an ALF raid in
protest of animal experimentation took place on the campus, Scarce was subpoenaed to appear
before a grand jury.
Scarce was interrogated by FBI agents and subsequently jailed for 157 days for contempt
of court after refusing to violate the ASA code of ethics, which forbade him from sharing
confidential information with law enforcement authorities. Scarce warns that, if researchers
60 Chapter 2 • Ethics in Criminal Justice Research
make absolute assurances of confidentiality, they should be fully aware of what those assurances
portend-"a conscious decision to go to jail rather than relent in the face of legal pressure to
violate confidentiality agreements" (Scarce, 2001, p. 271).
When an anthropologist involved in a medical lawsuit refused to turn over her field
notes, the American Anthropological Association issued a statement that said researchers
have "primary ethical obligations to the people, species, and materials they study . . . " but
these "may be affected by the requirements of other codes, laws and ethics of the country"
(Wilson, 2003). Instead of arguing in the courts for greater protection of confidentiality, some
social scientists have backed away from promising confidentiality. Rik Scarce states that this
view says, "I'll try as hard as I can [to protect confidentiality], but if worse comes to worse
I may have to give you up" (ibid.). New York Times reporter Judith Miller was jailed on
July 6, 2005, for contempt of court for not revealing who in the White House leaked the
identity of a CIA agent. She was later released when the informant, Scooter Libby (an aide
to Vice President Cheney), came forward. Scarce maintains that proposed congressional
legislation for shield law protection for such journalists should include social scientists
(Scarce, 2005a, 2005b).
Perhaps as a footnote to the Scarce case is another related to environmental issues. In
September 1998, the University of Denver "withdrew" an article that they had already agreed
to publish. The article, "The Critical Need for Law Reform to Regulate the Abusive Practices
of Transnational Corporations: The Illustrative Case of Boise Cascade Corporation in
International Law and Policy, September 1998 edition. The Summer 1999 edition printed an
"errata" notice that "this article has been retracted for its lack of scholarship and false content"
(Monaghan, 2000, p. A14). The authors, Professors Mark Buchanon, William Wines, and
XXXXX XXXXX, received the letter from the Boise Cascade Corporation, the target of criticism
in the article. Despite vigorous denials by the university, the authors claim that the university
caved in to the intimidation and did not even contact the authors before retracting their article.
University officials denied that threats of a lawsuit were instrumental in their highly unusual
decision. In August 2000, the authors filed a lawsuit in federal court in Idaho, arguing that
their reputations have been damaged and that university was guilty of breach of contract and
destroyed two years' worth of scholarship.
In yet another twist related to censorship (in this case sponsorship) of research, it was
reported in 1999 that Wyeth-Ayers Laboratories (producers of fen-phen, a diet drug that was later
found to be dangerous) hired ghostwriters for articles promoting the drug. It then used unknowing
prominent researchers to publish works under their names. Only two of the ten articles paid for by
Wyeth were published in medical journals before the drug was pulled from the market. Plans
to publish the others were canceled. The company claimed, "This is common practice in the
industry. It's not particular to us" (Diet drug, 1999).
Emphasis in large universities on obtaining research funding may force scholars to
abandon basic research and scholarship in favor of applied research and entrepreneurship, the
"corporatization" of higher education. The concern is with knowledge being viewed as a private
commodity rather than a public good and overemphasizing scholars as entrepreneurs rather than
scholars (Desruisseaux, 1999).
The issue of ethics in criminal justice research draws greater attention now than in the past.
Ethical concerns in research gain complexity with revisions in federal laws regulating privacy,
confidentiality, and freedom of information; debates continue in the various professional
Chapter 2 • Ethics in Criminal Justice Research 61
disciplines involved in studying criminology and criminal justice. The desire to address these
issues is reflected in the establishment of a specialized journal, Criminal Justice Ethics.1
The parent disciplines to criminology/criminal justice, such as sociology, psychology, political
science, and anthropology have both unique and common ethical concerns, as reflected in
their codes of ethics. Because professionals pride themselves on autonomy and self-regulation,
these requirements are usually beyond any minimal standards set by government agencies.
Some specific ethical considerations are incorporated by these codes. Although every effort may
be made to preserve their anonymity, subjects should be made aware that they may unintentionally
be compromised. A researcher's pseudonyms should not be indiscriminately revealed.
Anthropologists indicate that, if secretive research is performed for a sponsor, such reports should
also be released to the public and subjects. Researchers should be honest with sponsors regarding
qualifications, capabilities, and harm, and open concerning the acknowledgment of sponsorship of
research. Any government-supported research should be unclassified, and researchers should not
use their research as a cover for government intelligence work.
If research has potential policy implications, it is even more imperative that investigators
state the limitations of their findings. Relationships that may compromise objectivity or create
a conflict of interest should be avoided. Psychologists insist on a high standard of competence,
including use of the latest rules regarding validity and reliability of tests and measurements
employed in research or practice. Political scientists have specific guidelines on involvement in the
political arena, often a necessary part of the discipline. Sociologists mirror most of these same
themes, but add that, regardless of work setting, sociologists are obligated to report findings fully
and without omission of significant data. Also as an ethical matter, researchers are obligated
to make their data available to other qualified social scientists at reasonable cost once they have
completed their study.
AVOIDING ETHICAL PROBLEMS
Because of the nature of the subject matter, ethical problems are likely to present themselves in
consider alternate means of data gathering that may not entail ethical problems. Some study
designs pose less hazard than others. For example, perhaps existing data or some other unobtrusive
method would make it unnecessary to collect new data. Rather than set up an experimental
group and a control group in a prison, one could locate two similar prisons that already constitute
natural experimental and control groups. Simulations, either human or computer, may enable one
to address the same issue without as much of an ethics problem. One could seek conditions under
which negative effects have already occurred (Bailey, 1987, pp. 407-410). Similarly, the use of
samples or only low levels of the treatment may reduce the potential harm.
The reporting of aggregative rather than individual data with proper prior destruction of
identifiers may ensure protection. Such simulations as "mock jury studies" may also avoid
ethical problems with real populations.
1 Published by Institute for Criminal Justice Ethics, John Jay College of Criminal Justice, 444 West 56th Street, New York,
62 Chapter 2 • Ethics in Criminal Justice Research
In an earlier discussion of sample surveys it was pointed out that there is a potential growing
"respondent revolt" against the increased requests to participate in studies. All research is to some
extent an imposition on the lives or time of subjects. By the same token, research also poses, as
we have seen in this chapter, potential unethical impacts on those involved in a study. Although the
points set forth in this section can alert the investigator to broad guideposts, the actual research path
is in the last analysis the sole responsibility of the researcher.
Ethical horror stories were reviewed in order to document
some examples of unethical behavior in
research. Accounts of biomedical research by Nazis,
the Tuskegee Syphilis Study, experiments on prisoners,
as well as secretive research sponsored by intelligence
agencies preceded discussion of social science
examples. The latter included Milgram's "obedience
to authority" study, Zimbardo's "simulated prison
study," and Humphreys' Tearoom Trade. Zimbardo's
"Lucifer effect" refers to situations in which good
people do evil things. Plagiarism and researcher
fraud such as the Piltdown and Tasaday hoaxes were
also discussed as examples of scams and researcher
As one of the social sciences, criminal justice
must be concerned with many of the same issues of
ethical behavior in conducting research on human
activity. Potential role conflict exists for the investigator
who has to balance the roles of researcher, criminal
justice practitioner, citizen, and humanitarian. In
mediating these roles it is essential that the researcher
anticipate many of the possible points of friction
primary role is that of scientist.
Research targets in criminology and criminal
justice include the criminal, the victim, the criminal
justice system, practitioners, and the general public.
Each research situation presents its own unique set
of ethical problems, just as ethics itself is relative to
the setting. Although science itself is ethically neutral,
individual scientists cannot entirely avoid moral
concerns in studying human groups or individuals.
In aspiring for acceptance as professionals, criminal
justice researchers attempt to demonstrate that on
the basis of knowledge and ethical conduct they
should be granted autonomy and high professional
regard. Although this striving and establishment of
ethical codes may be viewed as a smokescreen to
gain monopolistic power, if a field fails to set its own
standards, the government will set the standards for
them. A code of ethical conduct in research that
exists in criminal justice was still in draft stage in
late 1998. The purpose of this chapter has been to
call attention to key features. Rather than a rigid set
of requirements, they might best be viewed as general
principles that must be tailored to the unique features
of individual research projects.
The most influential guidelines governing
funded research in the United States are those of the
HHS, Model Policy on Human Research Subjects.
Researchers involved in ethnography or field studies
were critical of the attempted application of these
informed consent guidelines to participant observation
studies and felt that they would make such
studies impossible. Fieldwork was viewed as different
in kind than biomedical and experimental studies
of controlled subjects. The National Commission for
the Protection of Human Subjects (1978a) in the
Belmont Report partially revised these guidelines,
putting their interpretation in the hands of review
boards and suggesting an alteration of informed
consent in the case of field research. Revision of
HHS guidelines in 1981 virtually excluded most
social science and criminal justice/criminological
research from required review.
In addition to government agency regulations
regarding funded research, professional associations,
for example, the American Sociological and
Psychological associations, have their own codes of
ethics governing research. Taking both governmental
Principles in Criminal Justice Research are presented
as suggestive elements of consideration in investigations:
(1) researchers should avoid procedures
Chapter 2 • Ethics in Criminal Justice Research 63
which may harm respondents; (2) one should honor
commitments to respondents and respect reciprocity;
(3) investigators should maintain objectivity and
exercise professional integrity in performing and
reporting research; and (4) investigators should
protect the confidentiality and privacy of respondents.
In addition, discussion was presented that
criminal justice researchers should be alert to the fact
that they enjoy no legally recognized privilege of
confidentiality (except, of course, in some cases of
funded research that are restrictive in coverage).
Some ethical problems experienced by
researchers such as Brajuha, Ofshe, and Hutchinson
were discussed. Finally, some means of avoiding
ethical problems are the use of alternate methods that
possess fewer ethical problems, study groups that
possess characteristics or natural treatments, the use of
samples instead of larger populations, the reporting of
only aggregative data, the use of simulations, as well
as exposure of groups to only low levels of treatment.
In the last analysis, researchers must assume personal
responsibility for the morality of their research.
Lucifer effect 32
Research fraud 36
Role of researcher 38
Research targets in criminal
Professional ethics 40
Ethical principles for criminal
justice research 40
Institutional Guide to DHEW
Policy on Protection of
Human Subjects 41
Institutional review boards
Informed consent 42
NIJ's Regulations on
Shield laws 49
Risk-benefit ratio 53
1. Research in criminal justice and criminology faces
many ethical hazards. How concerned should
researchers be with ethical conduct in research? What
is the researcher's role and what constitutes appropriate
2. The regulation of ethically acceptable research conduct
may take one of three forms. Discuss these and include
your opinion as to which of these is most effective.
3. How did the new HHS guidelines resolve the principal
objections of field researchers to the informed consent
4. What are some elements of a code of ethics for criminology/
criminal justice research?
5. Discuss some examples of researcher misconduct.
Why does this occur and how can it be prevented?
Ethics and Justice www.ethics-justice.org/
Office of Research Integrity http://ori.dhhs.gov/
Association of Internet Researchers (Ethics Working
Office of Human Research Protections (IRB
National Academy of Sciences (Responsible Conduct
in Research) www.nap.edu/htm/obas/
Institute for Criminal Justice Ethics (John Jay
THE END OF CHAPTER 2
3 Research Design
The Experimental Model and Its Variations
The Experimental Model
Research Design in a Nutshell
Resolution of the Causality Problem
Rival Causal Factors
Internal Factors: Variables Related
to Internal Validity
External Factors: Variables Related
to External Validity
Reactivity or Awareness of Being Studied
Related Rival Causal Factors
Post Hoc Error
The Classic Experimental Design
Some Criminal Justice Examples of the
Classic Experimental Design
Exhibit 3.1 The Kansas City Gun Experiment
Other Experimental Designs
Posttest-Only Control Group Design
Solomon Four-Group Design
One-Group Ex Post Facto Design
One-Group Before-After Design
Two-Group Ex Post Facto Design
Cross-Sectional and Longitudinal Designs
Exhibit 3.2 The Cycle of Violence and
Victims of Child Abuse
Multiple Interrupted Time-Series Designs
Some Other Criminal Justice Examples of
Variations of the Experimental Model
The Provo and Silverlake Experiments
Exhibit 3.3 Evaluations of Shock Incarceration
The Kansas City Preventive Patrol Experiment
Chapter 3 • Research Design 65
The Minneapolis Domestic Violence
The Experiment as a Data-Gathering
Advantages of Experiments
Disadvantages of Experiments
Research design is the plan or blueprint for a study and includes the who, what, where,
when, why, and how of an investigation. The research design should flow from the
problem formulation and critical issues that were identified for observation. Who is to
be investigated-an individual, one group, many groups, organizations, or communities? What
is to be investigated-attitudes, behavior, or records? Where is the study to be conducted? Do
we wish to look at the past (after-the-fact, post hoc, or a posteriori studies) or the present or to
predict the future? Do we want to look at a group once or over time? Why do we wish to do the
investigation-to describe, explain, or predict? Finally, how do we design the study so that
upon completion we are able to address the hypotheses and present findings that resolve in
some manner the research problem?
THE EXPERIMENTAL MODEL
Some people view experiments as involving white-coated scientists with an impressive assortment
of equipment tediously studying obscure phenomena in some isolated laboratory. Although
such a picture may indeed be accurate in some instances, the experimental model contains many
variations and should not be restricted to this stereotypical view. The experimental model will be
treated in this chapter as the benchmark for comparison of all other research designs and methods.
Most studies of an empirical nature in criminology and criminal justice can be viewed as
variations of the experimental model (Weubben, Straits, and Shulman, 1974; Campbell, 1977; and
Cook and Campbell, 1979). The research design notation (X's and O's) used in this chapter may at
first appear intimidating, but if you give it a chance, you will find it excellent shorthand for
dissecting any research study. This chapter contains more "researchese" than any other chapter,
but if you bear with it, you will be rewarded by becoming fluent in the language of research.
RESEARCH DESIGN IN A NUTSHELL
In learning something new such as swimming or driving, the initial lessons seem the most difficult;
once the foreignness of a new experience is overcome, the rest is relatively easy. Unfortunately,
learning research methods is similar. The language of research design in this chapter by its very
nature has this same foreignness at first. One philosophy is to teach something by simply doing it,
such as throwing the student in the deep end of the pool or on the fast lane on the Capitol Beltway
(which, incidentally, is where this writer learned to drive). Rather than following this practice
and throwing you in the deep end, let us begin with a short lesson. Read Figure 3.1, even if you
do not fully understand it. It provides the "guts" of the entire chapter. As you are reading or after
completing the chapter, you may wish to reread it because, if you understand Figure 3.1, you have
the underlying logic of research design.
66 Chapter 3 • Research Design
The ultimate purpose of all scientific investigation is to isolate, define, and explain the relationship
between key variables in order to predict and understand the underlying nature of reality.
The problem of causality has been a subject of continuing philosophical discussion, but scientific
investigation is based on the a priori assumption that the fundamental nature of reality can be
known-that causation lies at the basis of reality.
FIGURE 3.1 Research Design in a Nutshell.
X treatment (independent variable), e.g., Foot Patrol
Y outcome (dependent variable), e.g., Crime Rate
Z any rival causal factor (other variables besides X that could really be causing a change in Y),
e.g., history, selection bias, testing, etc.
O observation (some measurement or assessment of dependent variable)
E equivalence (randomization or matching)
1, 2 time
Design 1: O1XO2 (One Group Before-After Design)
rate of 500 (O2).
(X), e.g., History, Selection Bias, Testing Effects.
(or other variables).
Design 2: EO1XO2
EO1O2 (Classic Experimental Design)
Both are observed (O1) and have a crime rate of 1,000.
One precinct receives foot patrol (X) and is the experimental group.
The other precinct receives no treatment (control group).
After one year, the experimental precinct has a crime rate of 500 while the control group has a crime
rate of 1,000 (or no change). The decrease is attributed to foot patrol.
The change could not have been due to some historical event if we assume that both groups were
similar and exposed to the same history.
The change could not be due to selection bias because we purposely chose two similar or matched
precincts (E) and their crime rates were the same from the beginning (O1).
If there were testing effects (citizens were not surveyed; we used police reports), both groups should
have reacted the same to an awareness of being studied.
Thus, the relationship between X and Y is not due to Z [we have gained internal validity (accuracy)].
country? This is the problem of external validity.
Chapter 3 • Research Design 67
Resolving the Causality Problem. The first step involves the demonstration of a relationship or
covariance between variables. That is, one variable is related, increases or decreases in value, in
some predictable manner along with increases or decreases in the value of another variable.
The second step consists of specifying or indicating the time sequence of the relationship. Which
variable is the independent or predictor variable X, and which is the outcome or dependent
variable Y? Generally, logic or knowledge of which variable comes first gives one the direction
of causation. For instance, it would make more sense to assume that criminality of parents (X)
would precede in time and possibly predict criminality in offspring (Y), rather than vice versa.
In most instances, one has little difficulty in identifying the outcome (Y) one is interested in
predicting. It is usually the subject of the study. Although this process of causality resolution in
research has been greatly oversimplified for presentation purposes, most studies of an empirical
and predictive nature in criminal justice can be found to undergo essentially the first two steps
that have been outlined. The third step is the stage where many studies bog down and where
factors, or the elimination of other variables that could conceivably explain away the original
relationships the researcher had claimed. Other variables or rival causal factors may be responsible
for the variations discovered (Hirschi and Selvin, 1966).
In excluding rival causal factors, researchers are attempting to demonstrate that the
relationship between X and Y is nonspurious. A spurious relationship is a false relationship;
that is, one that is not caused by the believed variables but can be explained by other variables.
The presumed relationship between foot size (X) and intelligence (Y) may disappear
(be demonstrated to be spurious) when controlled for age (Z). That is, among thirty-year-olds
there is no relationship between foot size and intelligence.
causality problem are:
1. Demonstrate that a relationship exists between the key variables.
2. Specify the time order of the relationship.
3. Eliminate rival causal factors.
Suppose a researcher wanted to prove that a relationship exists between the increase in foot
patrols in a precinct and a decline in crime. Assuming that foot patrols have been increased in the
target precinct, the researcher looks at some measurement, such as precinct records of reported
crime. If no relationship between increased foot patrols and crime is discovered, the entire
process stops with step 1. There is little need to proceed if no relationship exists at all. If a
decrease or, for some reason, an increase in crime is discovered, however, the researcher
goes on to the next step. For our purposes, we will assume that an increase in foot patrols is the
predictor variable, X, and a decrease in reported crime is the outcome, Y. We assume that foot
patrols affect crime rates, rather than vice versa. One could see, however, where a researcher
might be interested in studying the latter; that is, high crime areas may more likely precipitate the
deployment of foot patrols.
Finally, suppose a relationship was discovered and specified: namely, there was an increase
in foot patrols and a decline in reported crime within the precinct. Does this then prove that
increases in foot patrols will cause a decrease in crime as measured by crime reported to police?
The answer to this question is no. The most obvious reason, which will become clear in Chapter 13
68 Chapter 3 • Research Design
causation. Such a finding merely brings us to stage 1 or 2 of our steps in resolving the causality
problem. If the prudent investigator has not already guessed, one's critics will very quickly point
out that other variables could have accounted for this relationship.
RIVAL CAUSAL FACTORS
Rival causal factors are any variables other than X (the treatment) that may be responsible for the
relationship. It is traditional in the social sciences, following the lead of Campbell and Stanley
(1963), to discuss these other variables, or rival causal factors, as being of two general types:
internal factors or other variables within the study itself that may tend to invalidate one's
findings and conclusions, and external factors or elements outside of one's immediate study that
may imperil the researcher's attempts to draw generalizations from the study and infer one's
findings to be true of larger populations.
Validity refers to accuracy or correctness in research. Internal factors question the internal
validity of research, whereas external factors impugn the external validity of findings. The former
asks whether the observational process itself produced the findings; the latter is concerned with
whether the results were unique and applicable only to the group or target studied. In checking
internal validity, one is concerned with whether a variable other than X (the treatment) may
have produced a change in Y (the dependent variable). With external validity, one asks what
other variables may limit one's ability to generalize the findings in a study to larger populations or
INTERNAL FACTORS: VARIABLES RELATED TO INTERNAL VALIDITY
for Research (1963), points to the following internal factors as possibly threatening the internal
validity of a study:
History Statistical Regression
Maturation Selection Bias
Testing Experimental Mortality
Instrumentation Selection-Maturation Interaction
All of these are rival causal factors that could have been responsible for producing the
results rather than the treatment assumed to be responsible. That is, although X and Y are related,
the real reason for this relationship is Z, some other variable or rival causal factor.
History refers to other specific events that may have taken place during the course of the study
and may have produced the results. For example, other than increased foot patrols what events
may have occurred in the hypothetical precinct and accounted for the decrease in crimes
reported? Perhaps the area was the target of a "Crime Watch" program that encouraged citizen
vigilance and reporting; or a new employment program was initiated to hire unemployed
Chapter 3 • Research Design 69
youths; or urban renewal changed the nature of the population inhabiting the precinct. Social,
seasonal, and other events may be responsible for changes in a study target. Garwood (1978)
gives an example of a burglary reduction program using "operation identification" in which
belongings are engraved to discourage burglars and fences. A northeastern experimental city
received the treatment, operation identification, and, especially during January 1978, showed a
dramatic decrease in reported burglaries. Can we assume that the X, operation identification,
was responsible for bringing this about? No. On further investigation, it was discovered
that other cities without such a program, notably Buffalo, Detroit, and Boston, demonstrated
equally impressive declines in burglary. What may have occurred in January 1978 to account
for this? It was the winter of incredibly deep, record-breaking, crippling snows in the
Northeast, and thus bad weather, a historical hidden variable, was most likely responsible for
the decline in burglary statistics. Similarly, a tough 1880 anti-horse theft law in New York City
is not responsible for the virtual disappearance of such thieves a hundred years later; rather,
it is the historical change in transportation.
In reviewing shortcomings of sentencing research, Farrington (1978) points out that because
of the lack of premeasures and control groups, it is difficult to assume that any decrease in crime
after a change in legal penalties or sentencing is due to these changes or other unmeasured social
changes that may have taken place at the same time.
Maturation refers to biological or psychological changes in the respondents during the course
of study that are not due to the experimental variable. "Time heals all wounds," according to
the old medical dictum, refers to the phenomenon in medical research wherein a given number
of patients can be expected to reveal improved conditions with or without treatment. Perhaps
the precinct under investigation in our foot patrol example was in the process of change, either
deterioration or upgrading, that brought about the change in crime reporting irrespective of
A hypothetical example may serve to illustrate maturation as a rival causal factor. An
interesting controversy of the 1960s was fluoridation of water in the United States.
Opponents claimed that the addition of such chemicals was potentially harmful. Suppose an
avid supporter of such a view were to state, "In 1850 Erie, Pennsylvania, was the first city to
fluoridate its water and not a single citizen from that time is alive today." Quite obviously,
the demise of this population was due primarily to natural causes, maturation, rather than the
assumed cause, fluoridation.
In a more serious vein, claims as to the long-term effectiveness of rehabilitation programs
must certainly control for the fact that as a given age cohort matures, its crime commission in
general tends to decrease; that is, there are very few eighty-year-old cat burglars. As a more
detailed example will illustrate later in this chapter, all other things being equal, older delinquents
can be expected to show lower crime commission over time than younger delinquents.
Testing (pretest bias) refers to the bias and foreknowledge introduced to respondents as a result
of having been pretested. On a second testing, the respondents are no longer naive regarding the
subject matter and can make use of sensitivities, information, and attitudes garnered from the
first testing. If one wanted to test a bank's reaction to a simulated robbery, the reaction of a bank
that had been held up the previous day would probably not yield valid or typical results.
70 Chapter 3 • Research Design
Instrumentation involves changes in the measuring instrument from the beginning or first period
of evaluation to the second, later, or final evaluation. The measuring instrument may refer to
observers, questionnaires, interviews, analyses of existing records, or any standard method of data
gathering. Suppose, for instance, that in our foot patrol experiment, the method of recording citizen
complaints was dramatically improved from what it had been at the beginning of the project. An
increase in crimes reported to police at the end of the study could very well have resulted from
instrumentation, a rival causal factor, rather than the assumed predictor variable-foot patrol.
A major limitation of comparing crime rates of today with those of yesteryear relates to the continual
improvement in record keeping so that an indeterminate proportion of the increase may simply
be evidence of improved instrumentation. We will discuss this subject in detail in Chapter 4 in the
section on the Uniform Crime Report. The crime rate of a city may show an increase of 100 percent
over the previous year, not necessarily because of increased crime commission, but because of
installation of a computer, a change in the manner in which crime is measured or recorded.
Statistical regression is the tendency of groups that have been selected for study on the basis of
extreme high or low scores to regress or move toward the mean or average on second testing. As in
our example, if a precinct were selected for study on the basis of an extremely high or low volume
of citizen complaints, it is expected, irrespective of the treatment variable, that the second reading
will be closer to the average for all precincts or certainly less at the extreme.
It should not be surprising that extremely high or low scores would move toward more
normal scores upon retest. Extremely tall people as a group are likely to have children shorter
than themselves, just as extremely short people are likely to have children taller than themselves.
Imagine taking the first examination in a class that you detest; after having had a bad night the
night before (illness, an all-nighter), you flunk the test. As a member of the lowest group, you are
then chosen for study. But before the second test you have a normal night's sleep, so we might
expect your performance to improve. The point is that the improvement may not be due to any
increase in intelligence but that your first test performance was atypical.
In critiquing reported positive claims of a program involving diversion alternatives
for youths who would otherwise have been incarcerated, Gordon and associates point out that
regression effects had been overlooked. The juveniles studied were chosen on the basis of
extremely high crime commission that could be expected to decrease upon second observation
even without intervention. Not surprisingly, it was claimed that the most extreme delinquents
demonstrated the greatest drop in recidivism as a result of a wilderness program. The most
likely explanation, however, was an expected regression toward the mean on the basis of the
initial choice of extreme cases (Gordon et al., 1978).
Selection bias occurs when the researcher chooses nonequivalent groups for comparison. Studies
that compare the attitudes or behavior of volunteers and nonvolunteers are often subject to selection
bias. If in our foot patrol study the precinct chosen for the experiment was characterized by high
levels of citizen involvement and reportage of crime, and these data were compared with those
for another nonfoot patrol precinct with historically low levels of reported crime, the rival causal
factor, selection bias, rather than foot patrol, might explain the differences in findings.
Chapter 3 • Research Design 71
Similarly, comparison of an experimental group consisting of all model prisoners and
a control group of incorrigible prisoners would hardly be fair. Many demonstration projects have
been accused of "creaming clients"-taking the cream of the crop-or stacking the deck by
assigning the best clients to the treatment and the dregs to the control group.
In studying the same group over a period of time, an expected loss of subjects can be anticipated.
This loss is referred to as experimental mortality. In our foot patrol experiment, a decline in
residential population as a result of urban renewal would certainly impinge on the number of
crimes reported. In correctional research, long-term recidivism studies have been handicapped
by the inability to follow all or most of the original respondents. Perhaps those who cannot be
found are more likely to be successes or failures than those on which data are available.
One method of assessing possible bias as a result of the loss of respondents is to compare
known characteristics of respondents with those of nonrespondents. Similarity in such demographic
characteristics as sex, age, race, and income may lead one to suspect that nonrespondents
do not differ much from respondents and therefore introduce little bias.
Obviously, factors within the experiment other than the assumed predictor variable may be
responsible for the findings. Interaction or combination impacts of any or all of these variables
may bring about the obtained results, for instance, interaction of selection bias and differential
maturation of groups. Selection-maturation interaction was illustrated by Gordon et al. (1978) in
their critique of a diversion program in which the researchers failed to control for age-selection
bias-or to spot a potential maturation effect when older delinquents were placed in the wilderness
programs that showed the greatest decline in recidivism. The latter can be viewed as a maturation
effect in that as a group ages, its overall crime commission declines.
EXTERNAL FACTORS: VARIABLES RELATED TO EXTERNAL VALIDITY
External factors refer to rival causal factors that negatively affect external validity or the representativeness
or generalizability of study findings to larger populations beyond the group studied
(Campbell and Stanley, 1963). The following are examples of external factors:
Reactivity or awareness of being studied
Although a clever researcher may do much to control for the effects of rival causal factors
within a study, this may not enhance the ability to generalize beyond the group studied. The testing
effects and selection bias previously discussed as affecting internal validity also affect external
validity (reactive or interaction effects of pretesting).
Testing effects point to the tendency of pretests to destroy the naiveté of respondents with respect
to the variable(s) being studied and decrease or more predictably increase the subjects' awareness
72 Chapter 3 • Research Design
or sensitivity, thus complicating the ability to generalize their responses to a larger population that
has not been pretested. For illustrative purposes, let us alter the foot patrol example by adding a
different dimension. Assume that the purpose of introducing walking beats was to enhance community
relations and public attitudes toward the police. At the beginning of the study, residents of
the precinct were questioned regarding these matters, then foot patrols were introduced, and the
residents were questioned again. A more favorable attitude toward police would be assumed to
have been produced by walking patrols; however, perhaps it was induced in part, or primarily, by
the group having been pretested and thus having had time to reflect and consider their views.
Furthermore, attitudes in this precinct could not be generalized to other even similar precincts,
without some hazard, unless a similar pretest-posttest had occurred there also.
Selection bias can have negative impacts on the ability to infer findings beyond the group studied.
Nonrepresentative selection of a study group obviously invalidates any attempt to generalize to
larger populations. For instance, the purposive selection of a precinct with high citizen vigilance
in responding to crime as the setting for an experiment in police deployment would not be a fair
test of how this same program would operate in more typical settings.
Reactivity or respondent awareness of being studied tends to produce atypical or unnatural behavior
on the part of subjects. Most people have had experience with previously announced inspections,
visiting or guest teachers and the like, to realize that behavior observed during that day tends to be at
times quite different from what normally occurs. Similarly, if the sample foot patrol precinct were
announced and continually covered in the media during the course of the experiment, the residents'
behavior, as well as the behavior of the police, would be different than usual.
This phenomenon is variously described as the "Hawthorne effect," "placebo effect," or
"stooge effect" and will be discussed shortly. Thus, an awareness of being studied, rather
than the experimental treatment, may become the major factor bringing about a particular
Multiple-treatment interference occurs when more than one treatment or predictor variable is used
on the same subjects. The outcome may be brought about by a specific sequence or combination of
independent variables that can be uncovered only by more complicated research designs, as will be
examined later in this chapter. If the foot patrol officers also wore blazers and did not carry guns and
gave out free tickets to sports events, a more positive attitude may have been produced by any one or
combination of variables, in addition to or irrespective of foot patrols.
RELATED RIVAL CAUSAL FACTORS
Although not distinct from those already discussed, a number of other Hawthorne-related terms
for sources of invalidity can be identified. The Hawthorne effect serves as an example of
reactivity resulting in atypical behavior or attitudes on the part of research subjects as a result
of their awareness of being studied. This factor gets its name from a pioneering industrial study
Chapter 3 • Research Design 73
Post hoc error
of a group of workers in the Hawthorne plant of the Western Electric Company in Chicago
(Roethlisberger and Dixon, 1939). To the bemusement of the researchers, alterations in treatment
designed to either increase or decrease worker efficiency consistently increased efficiency.
Rather than reacting to the treatment variable, X, workers were reacting to a rival causal factor-
the fact that they had been singled out for an important study. The workers reacted as they
suspected the researchers wanted them to act, rather than as they would under normal circumstances.
These acquiescent, "guinea pig," or "stooge" effects are likely in situations where the
group being studied is aware that they are being studied.
The halo effect was coined by Thorndike (1920, p. 25), who noticed that when supervisors rated
subordinates, the ratings were all "higher than reality." It refers to observer bias in which observers,
perhaps unconsciously, follow an initial tendency to rate certain objects or persons in a particular
manner; this initial orientation carries over into all subsequent ratings. The less specified and
discretionary the variable to be rated, the greater the danger of the halo effect (see Cooper, 1981).
Related in part, but more subtle than the halo effect, is the carryover into research of a
phenomenon first noted by sociologist W. I. Thomas. His basic maxim of "the definition of the
situation," or what others refer to as self-fulfilling prophecy, has a major bearing upon the bias
of the researcher. "If groups or individuals define a situation as real, it is real in its consequences"
(Thomas and Swaine, 1928, pp. 571-572). A researcher's own hidden biases and expectations
may influence his or her perception of events so as to bring about that which had been assumed.
Selective perception may lead to one ignoring anything that does not fit one's preset cognitive
map and thus presents us with an experimenter effect (Rosenthal, 1966).
Post hoc error comes from the Latin phrase "post hoc; ergo, propter hoc," literally, "after this;
therefore because of it." It is a fallacy to argue that one variable is the cause of an outcome because
it precedes that outcome in time. What is considered an effect is often only a subsequent event. An
example would be to argue that because every morning when the rooster crows the sun rises-the
crowing causes the sunrise. Another example can be the common claims made by police chiefs that
crime (reported crime) declined in their city in the 1980s because of new and effective policies.
More likely than not, what one was observing was the predicted "crime dip" of the 1980s, resulting
in part from demographic shifts in what has been described as the post-World War II "baby boom."
Gelles (1977) gives an example of such fallacious reasoning in research on child abusers.
Sometimes psychological conditions that are identified as being present after the abuse incident
are viewed as the cause of the incident; for example, abusers may be found to be paranoid and
depressed, conditions that may be results of the incident, rather than its cause.
Complete enumeration of related tags or descriptions of rival causal factors would be endless and
effect. This involves, similar to the Hawthorne effect, the tendency of subjects to react to a known
stimulus in the predicted manner (Loranger, Prout, and White, 1961). Commonly used in medical
research, the "sugar pill" or a placebo (fake treatment) with no known effects is administered
to the study group to hide the real treatment group and also to control for the placebo effect.
The actual effects of the true experimental pill can then be compared with effects induced in the
Chapter 3 • Research Design 75
Equivalence refers to the attempt on the part of the researcher to select and assign subjects
to comparison groups in such a manner that they can be assumed to be alike in all major respects.
The two methods by which equivalence of groups to be compared is gained are randomization and
matching. Randomization is the random assignment of subjects from a similar population to one
or the other group(s) to be compared in such a way that each individual has an equal probability of
being chosen and an equal probability of being assigned to any of the groups to be compared.
We discuss the process of randomization more thoroughly in Chapter 4, but at this point, it will
suffice to indicate that one of the principal means of accomplishing randomization is by use of
simple random samples or some means of selection in which each case in the population has an
equal probability of appearing.
Matching deals with assuring equivalence by selecting subjects for the second or other
comparison groups on the basis of matching certain key characteristics such as age, sex, and
race, so that the groups are similar or equivalent with respect to these characteristics. Matching
and randomization can be combined. In the Cambridge-Somerville study (McCord and McCord,
1959), 325 pairs of boys were matched on delinquency potential, and one member of each pair
was randomly assigned to treatment (Farrington, 1983, p. 261).
Assuming that the groups are similar, both are exposed to a pretest or observation prior
to exposure to treatment and a posttest or measurement after exposure to treatment. Finally, the
group exposed to treatment is called the experimental group; the group that is not exposed
to the stimulus or predictor variable is the control group. The original meaning of the term
"control" is "check"-the term comes from "counter-roll," a duplicate register or account made
to verify an account (Oakley, 1998).
For heuristic purposes, we adapt the notation developed by Campbell and Stanley (1963) for
schematically depicting the various research designs: X equals treatment, O symbolizes observations
(some researchers use T instead of O). Subscripts for O, such as O1 and O2, represent the first and
second observations, respectively, and E stands for the equivalence of comparison groups. Please
note that purists would be more conservative and insist that randomization (R) and not matching is
necessary for a classic experimental design.
E O1 X O2
E O1 O2
Following our previous presentation of this design with our newly introduced notation,
we find that the classic experimental design involves an equivalent assignment to experimental
and control groups, which are observed both before and after the experimental group receives
Using some of the rival causal factors affecting internal validity, we can now examine
the point that experimental designs are effective in controlling for many of these sources of
error before the fact. Suppose a classic experimental design had been employed for the foot patrol
experiment discussed earlier. Two precincts alike in all possible respects would have been chosen
for study (equivalence). The experimental precinct would have been pretested prior to the treatment
(foot patrol), whereas the control precinct would have received no treatment (retained usual
76 Chapter 3 • Research Design
patrol practices). At the end of the study, both groups would once again have been observed, and
any differences between them would be assumed to have been produced by the deployment of foot
patrols. History and maturation would not be likely rival explanations for these differences, because
both groups were similar and thus exposed to the same historical and maturation conditions. Both
groups were exposed to a pretest; thus, the differences could not result from the pretest, or at least
the extent to which the pretest influences results can be assessed because it will show up in both
groups. If it is assumed that both groups received the same instrument, changes in the instrument
would not account for differences as long as these changes were the same for both groups.
Statistical regression should be the same for both groups. Equivalence assures no selection bias and
experimental mortality should be the same.
This hypothetical foot patrol study is an example of a field experiment, a type of experiment
that is conducted in a natural (field) setting; a laboratory experiment is a type of experiment
that takes place in contrived or researcher-created conditions. Our next examples illustrate field
experiments. Because most of these studies employ matching rather than randomization as the
means of assuring equivalence, purists might argue that they are not true experimental designs but
rather quasi-experimental designs.
SOME CRIMINAL JUSTICE EXAMPLES OF THE CLASSIC
While the use of videotape for security purposes is routine today, at one time the usefulness
of such electronic devices (in this case, photographs) was not commonly understood. In 1975,
the Seattle Police Department installed hidden cameras in stereo speaker boxes in seventy-five
commercial establishments identified as high-risk potential robbery victims. These businesses
constituted the experimental group; a group of similar businesses received no treatment. The
pretest for both groups consisted of gathering statistics on the percentage of robberies cleared by
arrest and conviction rates prior to the study. If held up, a clerk triggered the camera by pulling a
"trip" bill from the cash drawer. A special project director would make prints of the photograph
of the robber available. A posttest comparing the two groups found that 55 percent of the
robberies of experimental companies were cleared by arrest versus 25 percent of the control
firms. Similarly, 48 percent of robbers at hidden camera sites were convicted, compared with
19 percent of the control group robbers ("Hidden Cameras Project," 1978).
In the late 1970s, claims of reduced recidivism among juveniles in trouble with the law in response
Straight. The assumption was that much of the "glamor" attached to criminal life by juveniles
on the road to more trouble could be nipped in the bud by blunt, heart-to-heart dialogue with
specially selected prisoners. Many jurisdictions began to set up what appeared to be a new gimmick
in corrections. Later evaluations, however, suggested that the benefits claimed were premature.
Yarborough conducted an evaluation of the JOLT (Juvenile Offenders Learn Truth) program
at the State Prison of Southern Michigan at Jackson. Unlike some earlier programs, verbal attacks
and obscenities were deemphasized. In 1978, subjects were randomly assigned to experimental and
Chapter 3 • Research Design 77
control groups and then measured at three- and six-month follow-up periods. All subjects were
male and had been arrested or petitioned for an offense that, had they been adults, would have been
criminal. No significant differences were found between those who had attended the JOLT session
and those who had not. There were no differences in the proportion having petitions filed or in the
types of offenses committed (Scared Straight, 1979).
In 1987, WOR TV (Secaucus, New Jersey, June 21, 9:00 P.M. EST) aired a documentary
entitled "Scared Straight: Ten Years Later." This program presented a very positive picture of
the experiment, relying primarily on interviews with people who had attended. Finckenauer
(1982), in Scared Straight! and the Panacea Phenomenon, felt that the original documentary
(1977) had misled the American public into thinking this was a miracle cure for juvenile crime.
Despite many problems and lack of cooperation in attempting to evaluate the "Juvenile
Awareness Program" or "Lifer's Program" (Scared Straight Program) at Rahway, Finckenauer
found that his randomly assigned Scared Straight experimentals actually had higher seriousness
delinquency scores afterward than the controls who did not attend the program. Similarly, many
of the juveniles put through the program were not the hardened "junior criminals" the public
had been led to believe.
A review of nine Scared Straight experiments over its thirty-three-year history indicated
that, despite good intentions, the program not only did not deter future criminal behavior
but actually led to more crime by program participants. The researchers concluded: "Given
the possibility of harmful effects of interventions, government has an ethical responsibility to
rigorously evaluate, on a continuing basis, the policies, practices, and programs it implements"
(Petrosino, Turpin-Petrosino, and Finckenauer, 2000).
Community policing has become a subject of much interest in law enforcement since the 1980s. The
term broadly refers to a variety of strategies that attempt to get the police away from rapid response
to service and closer to the community on a day-to-day basis. Order-maintenance, community crime
prevention, problem solving, neighborhood safety, foot patrol, and a host of police-community
relations strategies are all included under community policing (Mastrofski, 1992).
A variety of field experiments regarding the impact of neighborhood safety programs
such as "neighborhood or block watch," "police storefronts," and "foot patrol" experiments
have been undertaken. The Police Foundation's analysis of a neighborhood watch experiment
in Houston found no noticeable reduction in crime compared with a similar area that had not
received the program. Another Houston program established a police department storefront
(a combination precinct station, social center, and community outreach center). Houston also
experimented with personal-contact patrol in which officers attempted to stop and talk to as
many citizens as possible. Despite problems in maintaining experimental conditions, some
positive findings were obtained. The neighborhood watch and storefront programs had no
noticeable impact upon crime reduction but an enormous impact upon reduction of citizens'
fear of crime. In addition to reducing fear, the personal-contact patrol reduced household
victimizations by one-half and resulted in improving the attitudes of residents on community
issues (Sherman, 1985; Wycoff et al., 1985a, 1985b). Exhibit 3.1 describes a field experiment
in Kansas City that met with success.
A continuing subject of debate has been public dissatisfaction with anonymous, routine
policing in automobiles and requests for foot patrols, even though most police managers until
recently viewed such assignments as inefficient deployment of limited police personnel.
78 Chapter 3 • Research Design
The Kansas City Gun Experiment
The United States has both the highest violent crime
and homicide rate of any developed country, as well
as the largest armed civilian population in the world.
National attempts to significantly control firearms are
effectively blocked by the powerful National Rifle
Association. Given these contingencies, what can
police do to try to control growing youth homicide
rates? One possibility tested in the Kansas City gun
experiment by Sherman, Shaw, and Rogan (1995)
was that greater enforcement of existing laws
against carrying concealed weapons could reduce
gun crime. With a Bureau of Justice Assistance
"Weed and Seed" program grant, the Kansas City
Police Department selected a target patrol beat and
a control beat. The target beat had a 1991 homicide
rate of 177 per 100,000 persons, about twenty times
the national average. The control beat had a similar
violent crime rate. The research design involved a
matched groups before-after design. The "hot spot"
target area received increased proactive patrols.
The actual technique the officers used to
find guns varied, from frisks and searches
incident to arrest on other charges to safety
frisks associated with car stops for traffic
violations. Every arrest for carrying concealed
weapons had to be approved for adequate
articulable suspicion with a supervisory
detective's signature (ibid., p. 6).
Figure 1 illustrates the differences between
the target beat and the comparison beat during the
Gun crimes in the target beat decreased from
37 per 1,000 persons to 18.9 and guns seized
increased from 9.9 to 16.8. The comparison beat
showed little change in gun crime (22.6 to 23.6 per
1,000) and an actual decrease in guns seized (10.4 to
8.8 per 1,000). There was no displacement of gun
crimes to surrounding areas. Drive-by shootings
dropped from 7 to 1 in the target area but doubled
During Patrols Before Patrols During Patrols
FIGURE 1 Kansas City Gun Experiment-Guns Seized per 1,000 Persons. (Source: Sherman, Shaw,
and Rogan, 1995, p. 1.)
Chapter 3 • Research Design 79
Monthly Firearm Offenses
1990 1991 1992 1993
121 121 121 12
Target Beat Comparison Beat
FIGURE 2 Total Offenses with Firearms by Month in Target and Comparison Beats. (Source: Sherman,
Shaw, and Rogan, 1995, p. 7.)
During the late 1970s, experiments with foot patrols were conducted in Newark, New Jersey,
and Flint, Michigan (Police Foundation, 1981; Wilson and Kelling, 1982; Trojanowicz and
Banas, 1985). The same findings were obtained in both studies (Kelling, 1985, p. 2):
decreased fear of crime, greater citizen satisfaction, and greater appreciation of neighborhood
values by the police. There also appeared to be greater job satisfaction, less fear, and higher
morale for officers who patrol on foot than for officers who patrol in automobiles. The
replications are underway as of this writing. Lawrence
Sherman, the principal author of the report, is on
leave from the University of Maryland and is serving
as Criminologist to the Indianapolis Police Department
in order to further test the program.
Source: Sherman, Lawrence W., James W. Shaw, and
Dennis P. Rogan. "The Kansas City Gun Experiment."
National Institute of Justice Research in Brief, January
1995. Document available from National Criminal
Justice Reference Service, Box 6000, Rockville, MD
20849-6000; call 1-XXX-XXX-XXXX or Internet
EXHIBIT 3.1 (Continued )
from 6 to 12 in the comparison area, again with no
displacement effect. Figure 2 compares offenses by
Other positive findings were a decline in homicides
in the target area but not in the comparison
area. Citizens in the target area were less fearful, but
there was no change in fear in the comparison area.
Two-thirds of those arrested for gun carrying in the
target area were from outside the area.
Finally, only gun crimes were reduced by
the directed patrols, and they had no effect on calls
for service or reduction of other crimes. Further
80 Chapter 3 • Research Design
Flint study showed a decrease of 40 percent in service calls via telephone and a modest
reduction in crime, whereas the Newark program showed no crime reduction. More replications
(repeats of the experiments) are under way. Foot patrols are obviously no panacea but
have been found popular, particularly when selectively implemented in densely populated
In the 1990s, the National Institute of Justice had contracted a large number of
studies of community policing ("Community Policing," 1992). For example, an evaluation
of the Madison, Wisconsin, Police Department's "quality" policing program examined a
community-oriented policing program with a new organizational design based on the work of
management expert Edwards Deming. In comparing the experimental with the comparison
police district, it was found that in the experimental district the managers viewed themselves
more as problem solvers and employee attitudes regarding their work and organization
improved. In addition, citizen interaction improved and their perception of crime as a
problem was reduced. Citizens also expressed a more positive attitude toward the police
(Wycoff and Skogan, 1994).
OTHER EXPERIMENTAL DESIGNS
To assess the impact of consolidated police departments on public satisfaction with policing in
their area, Ostrom, Parks, and Whitaker conducted a posttest-only control group design (1973).
E X O
Three Indianapolis neighborhoods with a consolidated police department were matched
with three communities with independent police departments. The researchers discovered
higher public satisfaction in the communities with independent departments. Without a pretest
it is unclear, however, whether this difference could not in fact have been caused by other rival
factors; for example, perhaps even before police department consolidation, the Indianapolis
neighborhoods had lower degrees of satisfaction with their policing.
The Solomon four-group design (Solomon, 1949) is viewed by some as the purest of research
designs. Basically, it combines the classic experimental design with the posttest-only design. The
Solomon design has four groups, the first two resembling a classic design and the second two
resembling a posttest-only control group design.
E X O2
Chapter 3 • Research Design 81
To illustrate the Solomon design, let us hypothetically suppose, using the same Indianapolis
example, that four areas could be chosen for study from evenly matched communities. The first
two areas would both be measured with respect to public attitude toward the police. Only one of
these would receive the treatment (consolidated policing), and both would be remeasured with
respect to attitude. Two other areas would not receive the premeasure; that is, they would be
measured only after one had received treatment (consolidated policing) and one had not. Such a
design would assess the effect of testing effects as well as provide a premeasure lacking in the
posttest-only control group design. The advantage of the posttest-only control group design is that
it eliminates testing effects and possibly reactivity entirely, although it lacks a measure of where
the groups stood prior to the treatment. The Solomon four-group design obviously has the
advantage of having the premeasure and, by adding the second two groups, has the same advantage
as the posttest-only control group design. It is, however, expensive and difficult to implement
and, therefore, not practical in many research situations.
The classic Solomon and posttest-only control group designs are examples of experimental
designs. Randomization, in which equivalence is obtained by random assignment of subjects to
experimental and control groups, is the key distinguishing characteristic of experimental designs.
Preexperimental designs lack equivalence of groups, and quasi-experimental designs rely on
matching of subjects to achieve equivalence. In many field experiments such as the Indianapolis
example, randomization may be inappropriate or impossible; as long as equivalence is assured, it
can be argued that these are true experiments.
Research designs that lack one or two of the three major elements of experimental designs-
equivalence or experimental and control groups-are designated as preexperimental designs.
O X O
All of these preexperimental designs fail to provide equivalence or any assurance that the group(s)
facto design, or one-shot case study, is quite typical of many early criminal justice demonstration
projects. Our original example of the precinct foot patrol experiment, if it contained no preobservation,
would serve as an illustration. Unlike true experimental designs, one-group ex post facto
(after the fact) studies are subject to many internal invalidity factors or errors. One simply has
chosen for study a group that has already been exposed to a particular treatment. Obviously, things
other than the treatment could explain the outcome. If one finds a precinct that had experimental
foot patrols or an agency with low recidivism rates, without premeasures, equivalence, and control
groups, one is on shaky ground in concluding that lower crime rates or lower recidivism are due to
these factors. Many studies, particularly field studies in criminology and criminal justice, are of
82 Chapter 3 • Research Design
the one-shot case study variety. Cressey's (1957) study of incarcerated embezzlers, for instance,
may have major problems with respect to selection bias and reactivity but, on the other hand, may
be the only way of obtaining exploratory information on a little-known topic. What one-group ex
post facto studies lose in terms of internal control of error may be gained in terms of studying
groups in natural field conditions.
Early research on the XYY or "supermale syndrome" assumed, on the basis of studies of
incarcerated violent offenders, that an extra male chromosome may have been responsible for
violent crime (Witkin et al., 1978). Only later examination of the general population suggested
that it may be as prevalent among the "noncriminal" male population. As mentioned previously,
in lamenting the shortcomings of studies of sentencing behavior, Farrington (1978) points out
that few studies use both before and after measures or compare a sentenced group with an unsentenced
group, thus making it difficult to know whether changes in behavior are due to sentences,
penalties, or other concurrent social changes.
In another example of a one-group ex post facto study, Heussenstamm (1971) reports on a
field experiment in which subjects, none of whom had received traffic violations the previous
twelve months, attached Black Panther bumper stickers to their automobiles. They attracted so
many traffic citations that the experiment had to be canceled. That Heussenstamm had knowledge
of the fact that the subjects had not received citations before the treatment may qualify the
study as being an example of the next type, the one-group before-after design.
The one-group before-after design, or one-group pretest-posttest design, is an example of a
longitudinal design. A group, which is not necessarily chosen on the basis of representativeness,
is observed, exposed to treatment, and again observed. The primary advantage of this design over
the one-group ex post facto design is, of course, the presence of a premeasure. This adds, however,
the problem of testing effects and has the same problem as the one-group ex post facto design in
that one's findings cannot be compared with those for a similar control group not exposed to
An example of a one-group before-after design is Pierce and Bowers' (1979) analysis of
the impact of the Massachusetts Bartley-Fox gun law, which carried a one-year minimum prison
sentence for the unlicensed carrying of firearms. With the use of recorded crime statistics and
observations both before and after passage of the law, the earliest part of a longitudinal design
suggested a decrease in gun-related assaults, robberies, and homicides; however, this was offset
by increases in nongun assaults and robberies using other weapons. Without a control group, the
same problem exists as in our previous burglary reduction program (Garwood, 1978)-other
variables may be responsible for these findings.
The two-group ex post facto design eliminates possible pretest reactivity by studying both an
experimental and a control group after the experimental group has been exposed to some
treatment. The primary problem with this design is that there is no way of being sure that the
two groups were initially equivalent. Skillful selection of groups for comparison may be the
only option a researcher has in some instances. Brown et al. (1970) surveyed two groups of
parolees: those who had succeeded and those who had failed at parole. After the fact, they
were asked to identify factors in the institution and community that assisted or impeded their
Chapter 3 • Research Design 83
adjustment. Such two-group ex post facto designs were heavily utilized in early biological
and psychological theories. These theories in criminology that claimed genetic or personality
differences between criminals and noncriminals suffered from what Reid (2005, p. 89)
describes as the dualistic fallacy, the assumption that prisoners (who are supposed to represent
criminals) and groups from the general population (all of whom are assumed to be
noncriminals) represent mutually exclusive groups (or are nonoverlapping).
CROSS-SECTIONAL AND LONGITUDINAL DESIGNS
Before going further, it is crucial to briefly introduce a general distinction used to describe
research designs. Cross-sectional designs involve studies of one group at one time and usually
refer to a representative sample of this group. Longitudinal studies are studies of the same group
over a period of time and generally are studies of change (Menard, 1991).
Time-series designs involve variations of multiple observations of the same group at
various times. Variations may include dividing the original group into equivalent groups and
observing these portions longitudinally. In Chapters 5 and 6, we discuss the usefulness of
panel designs in such surveys as the National Crime Victimization Survey in providing an
in-depth view over time of the same study population. In a now classic criminological study,
Wolfgang et al. (1972) used existing records to longitudinally trace the criminal or noncriminal
careers of 9,945 boys born in Philadelphia in 1945.
In a replication that included females, Tracy, Wolfgang, and Figlio (1985) tracked the
criminal history of males and females born in Philadelphia in 1958 who continued to live
there from the age of ten until adulthood. Both studies were instrumental in identifying
the concept of serious career criminals, finding that approximately 6 percent of the 1945
group had been responsible for 53 percent of arrests for violent crime and 71 percent for
robbery, whereas 7 percent of the 1958 group had committed 75 percent of all serious crime
by this group. Farrington (1979) conducted a similar study in London begun in 1961 of boys
eight to nine years old in state primary schools.
One of the earliest series of cohort analysis was done by the Gluecks, who studied 500
reformatory inmates over ten years and 1,000 juvenile delinquents for more than fifteen years
(Glueck and Glueck, 1937, 1940). Such longitudinal designs are useful in giving us the longand
short-term variations over time. Another example of an ambitious longitudinal study was
the Cambridge-Somerville Youth Study begun in 1937 and continuing with some interruption
through 1945. Extensive data were gathered on each of the 650 boys who began the project
including delinquency, neighborhood, family conditions, school behavior, intelligence, and
personality. In 1955, the McCord and McCord (1959) reexamined the data and compared official
and unofficial delinquents. Exhibit 3.2 examines a longitudinal study of child abuse and
neglect victims. The Denver Youth Survey (Browning and Huizinga, 1999) is another example
of a longitudinal study. In order to discover correlates of crime, it has been following 1,527
boys and girls from high-risk neighborhoods in Denver who were seven, nine, eleven, thirteen,
and fifteen years old in 1987.
Time-series designs and panel designs are other terms used to refer to types of longitudinal
studies, as are cohort and trend studies which are also variations of longitudinal
designs. Time-series designs involve measuring a single variable at successive points in
time. In an interrupted time-series design, measurements are taken at time points prior to
treatment and for an equivalent period after intervention. The rate of crime committed one
84 Chapter 3 • Research Design
year prior to treatment could be compared with the rate for the first year after treatment
(Schneider and Wilson, 1978). Trend studies analyze different samples of the same general
population longitudinally, whereas cohort studies analyze subgroups over time, although
each time may consist of a sample of the cohort. Panel studies examine the same select group
or sample over time, as we will see in the National Crime Surveys. The most ambitious
longitudinal study ever conducted in the social sciences is the previously discussed "Project
of Human Development" of Chicago neighborhoods.
Although there has been some debate in the field of criminology regarding the
overapplication of scarce federal research funds to expensive longitudinal studies, it is often
the only way of sorting out many trends and causal relationships (Esbensen and Menard,
1990, p. 5).
The Cycle of Violence and Victims of Child Abuse
In previous research, Cathy Spatz Widom (1992)
found that childhood abuse or neglect increased
the odds of future delinquency and adult criminality
by an overall 40 percent. The study consisted
of a longitudinal study of 1,575 cases from childhood
through young adulthood. A group of 908
cases of child abuse or neglect processed by the
courts between 1967 and 1971 were tracked
using official records for fifteen to twenty years.
A comparison group of 667 was matched by
age, sex, race, and family social class. The study
While most members of both groups had
no juvenile or adult criminal record, being
abused or neglected as a child increased the
likelihood of arrest as a juvenile by 53 percent,
as an adult by 38 percent, and for violent
crime by 38 percent (ibid., p. 1).
Using these same cases, Widom (1995) also
examined the relationship between childhood
sexual abuse and later criminal behavior, particularly
sexual offenses. The key finding was that
People who were sexually victimized
during childhood are at higher risk of arrest
for committing crimes as adults, including
sex crimes, than are people who did not
suffer sexual or physical abuse or neglect
during childhood. However, the risk of
arrest for childhood sexual abuse victims
as adults is no higher than for victims of
other types of childhood abuse and neglect
(ibid., p. 2).
Compared with victims of physical abuse,
child abuse victims are more likely to be arrested for
prostitution. Victims of physical abuse were more
likely to commit rape and sodomy than were sexual
abuse victims or the nonvictimized. The long-assumed
relationship between childhood sexual abuse,
running away, and prostitution was not borne out by
All of these findings relied on official statistics
for measuring the dependent variable of crime
commission. Continuing research in this series is
examining other sources. An attempt is being made
to reinterview all 1,575 subjects in order to discover
other consequences of child abuse including social,
emotional, cognitive, psychiatric, and health
outcomes. Also to be examined are factors which
protect child abuse victims from later negative
Sources: Widom, Cathy Spatz. "The Cycle of Violence."
National Institute of Justice Research in Brief, October
1992; Widom, Cathy Spatz. "Victims of Childhood Sexual
Abuse-Later Criminal Consequences." National Institute
of Justice Research in Brief, March 1995. Documents can
be obtained from the National Criminal Justice Reference
Service, Box 6000, Rockville, MDNNN-NN-NNNN
call 1-XXX-XXX-XXXX or InternetXXX@XXXXXX.XXX.
There are many variations of the experimental design. In fact, as mentioned previously, almost all
research in criminal justice can be described using the notation with which we have been working.
Because quasi-experimental designs rely on matching-the use of "comparison groups" or means
other than randomization to obtain equivalence-the value of using comparison groups depends
upon how similar the groups are on key variables to the treatment group. Some quasi- or semiexperimental
designs include single time-series, multiple time-series, and counterbalanced designs.
Time-series designs refer to the analysis of a single variable (e.g., crime rate) at many successive
time periods with some measures taken prior to treatment and other observations taken after the
intervention. It is sometimes called an interrupted time series because the series of observations
is interrupted by a treatment (X).
Interrupted Time-Series Designs
O O O O X O O O O
It is desirable to have at least ten preobservations and a bare minimum of two, but probably
more, postobservations (Schneider et al., 1978, pp. 2-13). Such designs are widely used in criminal
justice research in examining the impact of a new law or treatment upon trends in crime.
Interrupted time-series designs can then be defined as an analysis of a single variable
measured at many successive time points, with some measures taken prior to a treatment
(interruption) and others taken after the treatment. Preproject observations are used as a
basis for estimating the trend, and differences between this projected trend and the trend
observed after treatment can be assessed to determine whether the treatment had an impact.
Figure 3.2 depicts time-series data for a problem-oriented policing program (the treatment)
designed to reduce larcenies from automobiles. The overall reduction in trend lines can be noted
from before to after the intervention.
Month and Year
Number of Larcenies
FIGURE 3.2 Time-Series Data for Larcenies from Automobiles in Newport News, Virginia. The
Intervention (Treatment) Was a Problem-Oriented Policing Approach That Consisted of Special
Tracking and Investigation of Crime Incidents. (Source: Spelman, William, and John E. Beck.
"Problem-Oriented Policing." Research in Brief. National Institute of Justice (January 1987): 7.)
86 Chapter 3 • Research Design
Not significant Significant
FIGURE 3.3 A Comparison of Pretest-Posttest Designs and Time-Series Designs. (Source: Schneider,
Anne L., et al.Handbook of Resources for Criminal Justice Evaluators. Washington, D.C.: National
Institute of Law Enforcement and Criminal Justice (1978): 2-47.)
Figure 3.3 demonstrates the advantage of time-series designs over simple pretest-posttest
In both instances, simple analysis of the last point before and the first point after the intervention
would have led to the conclusion of no significant change where, in fact, examination of
trend lines showed significant change.
Monahan and Walker (1990, p. 66) give an illustration of the superiority of a time-series
design over a before-after design in their analysis of the impact of the Community Mental Health
Centers Act of 1963. This program's goal was the reduction of state mental hospitalization.
In 1963, the year the act was passed, the resident population of state mental hospitals
in the United States was approximately 500,000. In 1990, it was less than
150,000. These before-after figures have been used to persuade Congress of the
effectiveness of the act. When a time-series with more than one measurement
before the passage is used, however, the results seem quite different. A time-series
shows the population of state mental hospitals to have increased each year from
early in the century until 1955, and decreased each year thereafter, with no noticeable
acceleration in the rate of decrease in 1963, the year the act was passed. In this
light, the most plausible hypothesis is that the factor causing the population
decrease began in the mid-1950's, and not in the mid-1960's. Many now view the
introduction of psychotropic medication as the principal method of treating mental
patients, which indeed began in 1955, as the most plausible hypothesis to account
for the "deinstitutionalization" of mental hospitals.
In yet other variations of interrupted time-series designs, the impact of new prisons
(experimental counties) was compared with matched/control counties in order to examine the
impact of the prisons. Twenty variables were followed for two years before and two years
Chapter 3 • Research Design 87
after the prison openings (Smykla et al., 1984). The relationship between "War and Capital
Punishment" was examined by studying the number of executions before, during, and after
three war periods (Schneider and Smykla, 1990).
A distinction is also made between single interrupted time-series designs, which examine one
designs, which contrast one group's performance with that of relevant comparison groups.
Multiple Time-Series Designs
O O O O O O O O
Although a single interrupted time-series design might examine the impact of the
55-mph speed limit upon vehicular deaths, a multiple interrupted time-series design would
compare a state with the new speed limit with one without it during the same period. For
example, Connecticut's 1955 crackdown on speeding reduced traffic fatalities, whereas
neighboring states without this program experienced no decrease (Campbell and Ross, 1980).
In the 1960s, Boston and New York City had very restrictive handgun licensing laws but complained
that their laws were defeated by guns brought in from nonrestrictive states. Zimring
(1975) studied the impact of the Federal Gun Control Act of 1968 on interstate traffic in guns
on homicide rates and found that handgun homicide rates actually grew faster in New York
City and Boston (the restrictive cities) than the average trend for fifty-seven control cities.
Such time-series designs are indispensable and widely used in criminological and criminal
justice research because the subjects often require analysis of trends or long-range effects
rather than short-term outcomes.
Counterbalanced designs are intended to manage or control the problem of multiple-treatment
inference, in which X1 refers to one treatment, X2 a second, X3 a third, and X4 a fourth. By using
four groups that are equivalent and by exposing each group to all four treatments and observing
after each combination of treatments, it becomes possible to isolate the treatment(s), combination
of treatments, or sequence of treatments that produces the outcome. For example, in four
precincts where the treatments were X1-foot patrol, X2-media campaign, X3-police blazers,
and X4-unarmed police, perhaps the desired outcome is obtained only by X3-police blazers
and X1-foot patrol in that sequence of introduction only. Such a design, although complex, is
the only way to uncover such a relationship.
E X2O X3O X4O X1OE X
E X4O X1O X2O X3O
There are many other variations of the experimental model. Familiarity with the notation
and basic designs we have discussed enables one to conceptualize these other designs as offshoots
of the basic ones.
88 Chapter 3 • Research Design
SOME OTHER CRIMINAL JUSTICE EXAMPLES OF VARIATIONS
OF THE EXPERIMENTAL MODEL
Empey and Erickson (1972) and Empey and Lubeck (1971) employed the classic experimental
design to assess the effect of experimental community-based treatment programs in Provo, Utah,
and Silverlake, Los Angeles. The subjects were all juveniles who were either serious and/or
repeat offenders. All were ordinarily candidates for reformatories who were sentenced by a
judge to probation or incarceration. All were male. None were seriously retarded, addicted to
drugs, or had a history of serious assaultive violence.
Although both the Provo and the Silverlake experiments are relatively complex in theoretical
design and methodological execution, a brief review of two elements of the analysis will serve our
purposes. One element in community resistance to community-based treatment programs has been
the fear of crime perpetrated by those undergoing treatment. The Provo researchers, and later the
Silverlake study, compared randomly assigned experimental and control groups from two disposition
conditions. Our discussion is concerned primarily with the Provo Study, although the
Silverlake experience is also briefly discussed.
Provo Research Design
Probation experimental group E O X O
Probation control group E O O
Incarceration experimental group E O X O
Incarceration control group E O O
Equivalence of groups was obtained by randomly assigning those given probation to the
Provo program or regular probation and similarly assigning those scheduled for incarceration to
either the Provo treatment or actual incarceration. With the incarceration group, randomization
broke down because the population was too small and the matched control group was selected
using a state training school.
In evaluating arrest rates as an indicator of crime committed, the probation controls had a
rate twice as high as the probation experimentals. Surprisingly, the arrest rate for "incarcerated"
experimentals was almost as low as that for incarcerated controls. The latter were either on
furloughs or were escapees who, in addition, tended to commit more serious crimes. Similar
findings were obtained in the Silverlake experiment, which added the variables of urban setting
and race (20 percent Hispanic or African American, versus 100 percent Caucasian in the
In analysis of yet another dependent variable, postprogram arrest rates, by means of a
longitudinal follow-up four years after release, the experimental probation types were not
greatly superior to regular probation control types (a 71 percent reduction in crime compared
with the rate four years prior to the experiment, versus 66 percent for the regular probation
group). The "incarcerated" experimental group showed a 49 percent reduction, whereas the
incarcerated control group exhibited only a 25 percent reduction. Maturation or age did not
affect these differences when controlled statistically; that is, these differences operated independently
of age. Among some methodological problems introduced in the Provo experiment
was a breakdown in equivalence, because the judge sentenced too few to the reformatory.
Chapter 3 • Research Design 89
Additionally, a possible Hawthorne effect was present, as suggested by the fact that the
success rate for the control group given regular probation was higher than that for the same
group before the experiment began. The probation officers' knowledge of the study may
have had this impact (Hood and Sparks, 1971, p. 209).
Despite some problems, given the higher costs of incarceration, the Provo/Silverlake
findings were a major demonstration of the utility of the community-based corrections
movement. Exhibit 3.3 reports on evaluations of shock incarceration.
Evaluations of Shock Incarceration
With a burgeoning prison population in the 1980s,
intermediate sanctions such as electronic monitoring,
intensive probation, and shock incarceration became
popular, cost-effective alternatives to overcrowding
prisons. They turned out not to fit the role of a
popular panacea. Often increased recidivism took
place due to an increased "net of control" (Morris
and Tonry, 1990) or increased technical violations due
to greater surveillance during community supervision
(Petersilia and Turner, 1990). While boot camp
programs (shock incarceration) vary in content, most
involve offenders participating in military-type
training and variations of physical exercise, hard
physical labor, ventilation therapy, substance abuse
therapy, and prerelease education. Doris Mackenzie
and associates have examined a number of boot
Examining a Louisiana boot camp (shock
incarceration) program, Mackenzie and Shaw (1993)
looked at shock incarceration releasees after two
years of community supervision and compared these
with similar offenders who had been given probation
or parole. Four groups of offenders were compared:
shock releasees, probationers, parolees, and shock
dropouts. These were contrasted with respect to
technical violations (terms of supervision offenses),
new crime arrests, and new crime convictions.
The shock graduates had higher rates of technical
violations and revocations than the probationers
and parolees, lower rates of new convictions, and,
in some analyses, lower rates of arrests and revocations
for new crimes. There were no differences
between shock alumnae and shock dropouts in the
Louisiana study (ibid.). This study was an example of
the utilization of a quasiexperimental design with
comparison groups in which there was no random
assignment of subjects. The shock experimentals
were compared with offenders who had been eligible
for shock incarceration but had received other treatments
instead. Possible invalidities in such a design
include selection bias (in that offenders were not
randomly assigned) and the fact that shock offenders
were more carefully scrutinized on release, which
most likely accounted for the greater number of
technical violations (ibid., p. 483).
Between 1982 and 1992, the number of shock
programs had increased to forty-one programs. In an
evaluation of eight shock incarceration programs
in Florida, Georgia, Illinois, Louisiana, New York,
Oklahoma, South Carolina, and Texas, Mackenzie and
Souryal (1994) found recidivism rates similar to
comparable offenders who did not go through boot
camps. Where shock offenders had lower rates, this
may have been due to selection bias (specially
selected offenders for the program). It appears that
boot camp experience itself does not reduce
recidivism (Travis, 1994). Successful programs were
followed by six-month intensive supervision in the
community. Participants in boot camps gave higher
ratings to their experience and felt safer.
Sources: Mackenzie, Doris L., and James W. Shaw.
"The Impact of Shock Incarceration on Technical
Quarterly 10 (September 1993): 462-488; Mackenzie,
Shock Incarceration. Rockville, M.D.: National Criminal
Justice Reference Service, NCJ #150062, 1994; Travis,
Jeremy. "Researchers Evaluate Eight Shock Incarceration
Programs." National Institute of Justice Update,
90 Chapter 3 • Research Design
patrol study (Kelling et al., 1974).
Experimental Group I (Reactive Patrols) E O1 X1 O2
Experimental Group II (Proactive Patrols) E O1 X2 O2
Control Group (Usual Patrols) E O1 O2
A fifteen-beat area of the city was divided into three matched five-beat groups. The first
group was reactive in patrol procedure; that is, officers responded only to calls for service and
did not deploy preventive patrols. The second group was proactive, and increased preventive
patrols up to three times the normal levels. Usual deployment, or preventive patrols at their
normal levels, was assigned to the control group.
Outcome variables analyzed prior to the treatments, O1, included reported crime and victimization
surveys of citizens and businesses. Posttreatment outcomes in citizen and business
victimization and perception of security and reported and unreported crime showed no statistically
significant differences among the three types of patrol areas studied. Despite methodological criticisms
such as the location of cars withdrawn from reactive patrol, small size of beats, and small
numbers used in the survey, the study suggested that police administrators have greater leeway
than they supposed in patrol deployment (Davis and Knowles, 1975; Kelling and Pate, 1975;
Larson, 1975; Pate et al., 1975; Chaiken, 1976). Similar replications of the Kansas City experiment
essentially confirmed the same findings (Albuquerque, 1979).
In a review of the Kansas City data, critics concluded that it was not likely that randomization
had been used in beat assignments (Feinberg, Singer, and Tanur, 1985). Kelling, the principal
director of the project, admitted that the police selected the beats on the basis of the department's
needs (Fagan, 1990, p. 110). Research by Sykes (1984) further illustrates the need to measure
different outcomes of increased enforcement efforts. In examining saturation patrol as a deterrent to
drunk driving, he found that it did deter some types of deviant behavior but was not a "panacea."
The Minneapolis Domestic Violence Experiment
A 1977 Police Foundation (1977, p. iv) study in Kansas City found that in the two years preceding
a case of domestic assault or domestic homicide, the police had been at the address of the incident
five or more times in half of the cases. This suggested that the police had an opportunity to attempt
to head off domestic violence. Beginning in the late 1960s, the police had been encouraged to train
their officers and utilize counseling and family crisis intervention strategies in domestic dispute
cases. By the 1980s, concern for the rights of female victims, possible lawsuits against police for
failure to make arrests where subsequent violence occurred, and a more conservative punitivepolicy
orientation led to a questioning of this policy.
In 1983, the Minneapolis Domestic Violence Experiment (Sherman and Berk, 1984a, 1984b;
Sherman, 1985) was undertaken to attempt to provide evidence as to the most effective strategy.
The research design is similar to the Kansas City Preventive Patrol Experiment-that is, it contains
three groups (two with different treatments and one control). In Minneapolis, police officers volunteered
to give up their discretion in handling simple (misdemeanor) domestic assaults and take
whatever action was dictated by a random system: instructions written on a card and drawn from an
envelope at the scene. Three different instructions were given: (1) arrest the suspect; (2) separate or
remove the suspect from the scene for eight hours; or (3) advise and mediate.
Minneapolis Domestic Violence Experiment
Arrest E O1 X1 O2
Separate E O1 X2 O2
Mediate* E O1 O2
After the police intervened, researchers attempted to interview the victims every two weeks
for the next six months, as well as monitor police records to check if there were any subsequent
assaults. The results appeared to be dramatic: 37 percent of the "advised" subjects and 33 percent
of the "separated" subjects had recidivated (committed new assaults within six months); however,
only 19 percent of the "arrested" subjects were repeaters. This reduction was accomplished even
though arrest usually entailed only a night in jail.
The Minneapolis experiment has been both the most widely accepted and the most influential
policy experiment of recent years; no other policy experiment has had quite the same impact on
criminal justice policy. It's effect may be explained in part by the conservative tenor of the times,
which emphasizes a law enforcement orientation for solving social problems; it may also be
explained by the very aggressive dissemination of the study's findings (Binder and Meeker, 1991).
Buzawa and Buzawa (1991) note that even though the research project was a modest pilot study with
acknowledged limitations, extraordinary efforts to publicize the results in the national media
resulted in premature police policy changes. In the first published replication of the study in Omaha,
Nebraska, the researchers Dunford, Hiuzinga, and Elliott (1989, 1990) found that arrest alone did not
have any greater impact than mediation or separation.
Binder and Meeker (1991) have provided the most thorough critique of the Minneapolis
study. They cited several objections:
about 28 percent of the cases were being processed by only three officers.
unemployed, whereas a similar study by Ray (1982) in the New York City area found
that most were employed.
had had husband-wife relationships. By comparison, in Ray's study (1982) only 10 percent
had previous records, with two-thirds being married couples.
training were not realistic comparison points.
and of internal factors such as "officer interest in victim's story" raised further questions
regarding the study's broad conclusions.
Further replications will tell us if we have too quickly embraced an "arrest panacea" for
handling domestic disputes. Results from six replications seem to suggest that arrest does not
work more effectively in deterring domestic assault.
The point of all of this is to suggest that research is an ongoing process and one in which
replication is essential; panaceas or simple solutions based on one study are suspect.
* Mediation could be considered a third treatment or X3.
92 Chapter 3 • Research Design
If you refused to be intimidated by the researchese in this chapter and learned your X's and
O's and how research design is a powerful tool for controlling rival causal factors, you are now
conversant in the language and actually have gotten through the worst part. In the following
chapters, we attempt to make you fluent in this language.
THE EXPERIMENT AS A DATA-GATHERING STRATEGY
We have thus far viewed experiments primarily from the standpoint of research design; however, the
experiment is also a data-gathering strategy. Through its three key features of assuring equivalence
of groups, pre- and posttests, and experimental and control groups, the experiment is a powerful
strategy for research.
As a research design strategy, the experiment consists of blueprints outlining the conduct
of the study. Saying that the experiment is the benchmark of comparison for other designs
suggests that by using the X and O notation scheme, we can depict a basic model of a research
study and also the potential strengths and weaknesses of such a design or research plan. The
experiment is also a tool for data gathering, a strategy for obtaining and analyzing data. As a
data-gathering strategy, the experiment has many variations that are defined by the setting.
These variations range from laboratory experiments to field experiments, the former having
greatest control over experimental conditions (thus high internal validity); but because of the
very controlled atmosphere, problems may exist in terms of artificiality or external validity.
Field experiments have fewer internal controls but greater external validity. In discussing relative
advantages and disadvantages of the experiment, it is difficult to distinguish whether critics
are talking about a design or data-gathering strategy or whether they are critiquing laboratory
and/or field experiments or both.
The advantages of experiments are many. They offer the best control for factors that tend to
affect the internal validity of studies. The researcher is able before the fact, by the very design of
the study, to control for many of the rival causal factors that tend to invalidate findings. The
experimenter can control for the effects of many variables by including or excluding them from
the study design.
A second advantage of experiments is that they are relatively quick and inexpensive. In
contrast to many of the other data-gathering strategies that we will examine in Chapters 4-8, an
experiment generally produces the required data necessary for analysis rather quickly.
Depending on the scope of the study and required staff, facilities, and equipment, the experiment
may represent a bargain compared with the expense of surveys or field strategies.
Another advantage of the experiment is its manageability, because the researcher is able to
call the shots by controlling the stimulus, the environment, the treatment time, and even the
degree of subject exposure. The conditions and conduct of experiments are often so rigorously
defined that they lend themselves to replication by which the design and methodology can be
repeated by other investigators. This is a major advantage over some field studies and surveys
where it may be more difficult to repeat all of the ingredients.
Experimental strategies can be applied to natural settings in which the researcher has the best
of both worlds, rigorous control and a more natural setting. If conditions can be viewed as realistic
by subjects, experiments may be the only way of studying certain complex behaviors. Additionally,
"natural experiments," which "occur as part of a natural process, where neither the setting nor the
randomization process are controlled" (Fagan, 1990, p. 13), may present themselves without any
Chapter 3 • Research Design 93
intervention by the researcher. For example, a new treatment program might be implemented, but it
might be applied to only half of the subjects due to funding shortages.
Despite the many advantages of experiments as a data-gathering strategy, there are potential disadvantages
a researcher should take into account in whether experiments are the preferred strategy.
The major disadvantage of experiments is their artificiality. In essence, the very controls imposed
by the researcher to control for rival causal factors internal to an experiment may create artificial
conditions that impede the ability to extrapolate to larger populations which are subject to natural
conditions. In controlling for extraneous conditions, one may literally be creating a mere shadow of
the former entity. This problem is more severe with laboratory experiments than field experiments.
In a typical critique of the contrived nature of some experiments, Field showed how many
laboratory simulations of jury decision making using college students as jurors may be in error.
Using randomly selected students and nonstudents in juror roles, Field (1978) found students to be
significantly more lenient in their sentencing. Thus, experimental results have no assumed built-in
validity. Some of the scorn with which some experimental researchers view data obtained by other
social science and criminal justice researchers using field research methods is misplaced.
Other major problems relate to the general difficulty of doing experimental research in
terms of obtaining human subjects or situations/conditions in which one can properly manipulate
the variables to be investigated. Major ethical issues can be raised by using experimental
research. Luskin points out the difficulty of implementing experimental designs in court
research. Court personnel may decline to experiment with new procedures or be unable to
manipulate key variables (Luskin, 1978). Hackler suggests that in evaluations of delinquency
prevention programs, traditional experimental-control group procedures are nearly impossible,
create unnecessary stress for program staff, and may produce hostility toward the researcher.
After-the-fact statistical analysis is in this instance viewed as far less obtrusive and more useful
than precontrolled studies (Hackler, 1978). Most judges are unwilling to permit treatment
decisions to be governed by pure random selection. In fact, one professor of criminal law
indicated that such assignment could constitute a violation of the right to due process (Glaser,
1976, p. 775), although others point out that most randomized experiments are ethical and legal
(Erez, 1986). Experimenter effects may also occur in experiments in which those conducting the
research actually selectively observe that they wish to see or unconsciously give cues to the
subjects as to the desired behavior or attitudes expected. Experiments provide an excellent
method for controlling for factors regarding internal validity, but they are often weak with
respect to external validity.
Assumptions of causality rest at the basis of scientific
investigation. Three essential steps are necessary to
attempt to resolve the causality question: demonstration
of relationship, specification of time order, and
control for, or exclusion of, rival causal variables.
The experimental model is one of the most powerful
means of controlling for rival causal factors before
the fact through the design of research. Rival causal
factors may be treated as factors affecting internal
validity and those affecting external validity.
Although the former are errors introduced because of
flaws within the study, the latter are factors that
impinge on the generalizability of the study to larger
populations. Factors affecting internal validity
94 Chapter 3 • Research Design
Steps for Resolving the
Causality Problem 67
Rival causal factors 67
Spurious relationship 67
Internal validity 68
External validity 68
Statistical regression 70
Selection bias 70
Experimental mortality 71
Testing effects 71
Hawthorne effect 72
Halo effect 73
Post hoc error 73
Placebo effect 73
Double-blind experiment 74
Research designs 74
Classic experimental design 74
Experimental group 75
Control group 75
Dualistic fallacy 83
Time-series designs 83
Advantages of experiments 92
include history, maturation, testing, instrumentation,
statistical regression, experimental mortality, and
interaction effects such as selection-maturation
interaction. Those impacting on external validity
include testing effects, selection bias, reactivity, and
multiple-treatment interference. Related invalidating
factors include the Hawthorne effect, the placebo
effect, the halo effect, and post hoc error.
The classic experimental design is the benchmark
or point of departure for all other research
designs. That is, in a sense, all forms of research can
be viewed as a variation of the experimental model.
The three basic components of the classic experiment
are pretest and posttest, experimental and control
groups, and equivalence. In addition, familiarization
with the notation of experimental designs is a useful
heuristic device for breaking down the essentials of a
research design. Classic experimental designs
provide for the most rigorous before-the-fact control
over factors of internal validity, and different variations
enable control for rival causal factors.
Illustration of the various designs with examples
from criminal justice research provides the reader
some familiarity with applications of these designs
from the criminology and criminal justice literature.
After this review and examination of examples
of the experimental model and its variations, Chapter
4 will explore the relative advantages and disadvantages
of alternatives to the experimental model in
Experimental methods of gathering data have
distinct advantages such as rigid control over rival
factors within the experiment, the relative quick
and inexpensive manner in which readily quantifiable
data can be gathered, and overall manageability
from the standpoint of the researcher. The
disadvantages of experiments often outweigh their
advantages, particularly in dealing with criminal
justice subject matter. Major shortcomings of the
experimental method include artificiality, which
may hinder its generalizability to wider populations,
and difficulty in applying the approach to
human subjects and situations in criminal justice.
Examples of the Kansas City gun experiment,
child abuse victims and violence, and shock incarceration
illustrated various research designs in this chapter.
1. How does research design control for rival causal
factors? Describe, for example, how the classic
experimental design controls for history and
2. Find a recent journal article that employs an experimental,
preexperimental, or quasi-experimental design.
Name, describe, and illustrate the design, and discuss
any rival causal factors controlled for in this design.
Uniform Crime Reports www.fbi.gov
National Incident-Based Reporting System (NIBRS)
Jastnet (Justice Technology Information Network)
3. Why are time-series designs particularly useful in
criminal justice studies?
4. Design a hypothetical study, and discuss how your
design controls for many rival causal factors.
5. Discuss the Kansas City gun experiment. What type
of research design was employed, and what were the
major findings of the project?
Research Methods in Criminal Justice and
4 The Uniform Crime Reports
The Uniform Crime Reports
The Violent and Property Crime Indexes
Cautions in the Use of UCR Data
Factors Affecting the UCR
Related UCR Issues
Exhibit 4.1 The Crime Dip
National Incident-Based Reporting System
NIBRS versus UCR
Exhibit 4.2 The National Incident-Based
Reporting System (NIBRS)
Types of Sampling
Simple Random Samples
Stratified Random Samples
Exhibit 4.3 Crime Profiling
This chapter, in part, serves as an introduction to the next four chapters, which deal in
detail with alternatives to the experiment as a data-gathering strategy. These alternatives
include surveys, participant observation, case studies, and unobtrusive measures. The
chapter begins with the examination of the Federal Bureau of Investigation (FBI) and Uniform
Crime Reports (UCR). The UCR serves as a data source, and, for a time, it was about the only
source of information consulted by researchers in examining crime and criminal behavior.
Because most of our later discussion of sources, such as victim and self-report surveys, are contrasted
with the UCR, it receives early and separate coverage in this chapter.
ALTERNATIVE DATA-GATHERING STRATEGIES
As has been indicated, the classic experimental design rests as a reference point with which to
compare all other research strategies; however, this by no means suggests that the experiment
as a method of data gathering is better or a more desirable strategy. This debate among social
scientists, who disagree about which means of data gathering is best, XXXXX XXXXX notion of
methodological narcissism (discussed in Chapter 1), which causes individuals to become so
committed to a particular research strategy that they consider all other approaches inferior. The
final resolution of this issue will be discussed in Chapter 9 under triangulation, where we will
argue that the best resolution of many of these issues is simply to employ multiple methodologies
or a wide array of instruments. Arguments that suggest its superiority are misplaced in
that the experiment, despite its obvious strengths with respect to internal controls, tends to
have primary weaknesses that other techniques do not have with respect to external validity.
Outside of captive prisoner research, many topics in criminal justice require "real-world"
strategies or approaches that bring the researcher into the actual environment of naturally
occurring events (Filstead, 1971). Researchers may wish to examine the controlled experiment
as the ideal in this field and consider whether or not alternatives are more acceptable, if not
Experiments are, therefore, by no means the most effective data-gathering strategy.
Depending on one's subject matter, a variety of techniques may be more appropriate or perhaps
necessary, because not all subjects lend themselves to experimentation. Figure 4.1 is an attempt
to illustrate the relative strengths and weaknesses of major data-gathering strategies. As has been
indicated, experiments are an outstanding method by which the researcher can exercise great
control over factors that may impair internal validity, as well as yield relatively quick and inexpensive
quantitative data; however, such an approach trades such control for its chief limitation
As we proceed down the vertical arrows in our illustration, we discover that the subsequent
techniques become in general less quantitative, and the research exercises less control
over rival causal factors impacting on internal validity. On the other hand, as we move away
from experimentation, the researcher gains external validity and moves closer to the natural
environment in which behavior occurs (Bouchard, 1976). This scheme is intended as an
"ideal type" that overgeneralizes to simplify presentation. The model is not entirely true in all
cases; for example, some unobtrusive methods can be very quantitative. In fact, our division
Chapter 4 • The Uniform Crime Reports and Sampling 97
Life History Methods
FIGURE 4.1 Alternative Data-Gathering Strategies in Criminal Justice:
A Heuristic Model.
98 Chapter 4 • The Uniform Crime Reports and Sampling
of methods of data gathering is also somewhat arbitrary but hopefully it organizes these
methods in a manner that clarifies their presentation.
THE UNIFORM CRIME REPORTS
The Uniform Crime Reports (UCR) is a special case in that, of all of the methodological
sources to be discussed, it has traditionally been the most widely cited on crime in the United
States. It is also used as a point of comparison for other data-gathering procedures, particularly
victim surveys and self-reports.
Beginning in 1930, the U.S. Department of Justice instituted the compilation and publication
of national crime statistics. Although participation in the program by police departments and
reports to the FBI, which assumed responsibility as a clearinghouse, was voluntary in nature, the
number of departments and comprehensiveness of reports have continually improved over the
years. Large metropolitan areas were the best participants.
Most newspaper and other media accounts are based on summaries presented in the UCR.
In most instances, these are data presented in an uncritical and alarmist manner without supplying
many of the qualifying problems with the official crime statistics reported. Being the major source
of crime statistics in the United States until 1974, the UCR basically comprises crimes known to,
and recorded by, local police departments. Figure 4.2 illustrates the relationship between crime
committed and other sources of crime statistics including the UCR (crimes recorded).
Although in Chapter 6 we will explore the usefulness of victimization surveys in
estimating the number of crimes committed, it is unclear whether it is possible to obtain an
accurate estimate of the volume of annual crime in American society. For various reasons, not
all crimes committed are discovered; for instance, some crimes involve situations in which
the victim is not aware of having been victimized or there is no identifiable victim. Not all
crimes discovered are reported to the police, and not all crimes reported to the police are
A = Crimes Committed Undiscovered
B = Crimes Discovered
C = Crimes Reported
D = Crimes Recorded
FIGURE 4.2 Theoretical Relationship between Crimes Committed and Official Statistics.
Chapter 4 • The Uniform Crime Reports and Sampling 99
recorded by the police. Although some may be concealed, a number of crimes reported are
unfounded or defined by investigating officers as not constituting a criminal matter. Thus,
even though crimes recorded by the police have an uncertain relationship to actual crimes
committed, until 1974 the UCR represented the available statistics on crime in American
society. The further removed statistics are from the crime committed, the weaker the figures
are as estimates of the true crime level. For instance, number of arrests, indictments, convictions,
incarcerations, and other dispositions such as probation or parole are all inadequate in
estimating the amount of crime in a society. They have much more to do with police efficiency
or allocation to correctional systems or other societal policies toward crime. The researcher
who chooses to utilize official statistics must become familiar with any shortcomings or
sources of bias they may contain.
The FBI receives these data from local police departments. In the majority of states, there
are operational state UCR programs in which the states required local departments to report their
data to the state and then share such data with the federal government. In the 1980s, 98 percent
of the police agencies operating in metropolitan statistical areas (MSAs)1 about 94 percent of
"other" cities, and 90 percent in rural areas reported the data. The Census Bureau estimates that
about 97 percent of the total national population was covered by the report.
THE VIOLENT AND PROPERTY CRIME INDEXES
In 2004, the FBI decided to stop reporting the crime index and report instead two indexes: the
violent crime index and the property crime index. An advisory board felt that the original crime
index had been distorted by including the category of larceny theft. Regardless, for those who
wish to make historical comparisons, it is not hard to combine the two and reconstruct the old
total UCR index.
The UCR is divided into Part 1 and Part 2 crimes. Part 1 crimes consist of index crimes
that are major felonies reported to the police that have been selected for special analysis because
of their seriousness, their frequency of occurrence, and the likelihood of being reported to the
police. The violent crime index consists of:
Murder and nonnegligent manslaughter
The property crime index includes:
Motor vehicle theft
Arson (data are not included in the index).
Part 2 crimes are nonindex crimes such as simple assault, vandalism, gambling, and
drunkenness. In all, twenty-two other crimes (twenty-one, excluding arson) are accounted for
1 MSA includes central cities (over 50,000) and contiguous counties that are functionally integrated economically and
socially with the core cities.
100 Chapter 4 • The Uniform Crime Reports and Sampling
under Part 2 offenses. Although the FBI indicates that it cannot vouch for the validity of data
received from individual police agencies, it attempts to examine all reports for accuracy,
question any unusual changes in trends, undertake special inquiries if necessary, and eliminate,
estimate, and otherwise control for at least the most crass errors.
Table 4.1 contains the Uniform Crime Reports Indexes.
The crime rate is expressed as the number of crimes per unit of population, in this case per
100,000. Such a statistic enables control of population size and thus permits a fair comparison of
different size units. The growth in crime one reads about in the paper is usually based on the
crime rate for the index offenses (excluding arson). Although the notion of scales and indexes
will be treated in Chapter 10, for the purposes of this chapter, it is important to realize that the
crime indexes are unweighted; that is, each offense is merely summed and in a sense given
the same importance as all other offenses. The crime rate is the total number of index crimes
per 100,000 population. Thus, the violent crime rate equals the number of violent index crimes
per 100,000 population, and the property crime index equals the number of property index
crimes per 100,000 population.
CAUTIONS IN THE USE OF UCR DATA
The following statement serves as a caution in using any official government statistics:
The government is very keen on amassing statistics. They collect them, add them,
raise them to nth power, take the cube root and prepare wonderful diagrams. But
TABLE 4.1 The UCR Violent and Property Crime Indexes
Crime Index Offense Number Rate Per 100,000
Murder and nonnegligent manslaughter 17,034 5.7
Forcible rape 92,455 30.9
Robbery 447,403 149.4
Aggravated assault 860,853 287.5
Burglary 2,183,746 729.4
Larceny-theft 6,607,013 2,206.8
Motor vehicle theft 1,192,809 421.3
Arson1 - -
Violent crime index2 1,417,745 473.5
Property crime index 9,983,568 3,334.5
1Although arson data are included, sufficient data are not available to estimate totals for this offense.
2Violent crimes are offenses of murder, forcible rape, robbery, and aggravated assault. Property crimes are
offenses of burglary, larceny-theft, and motor vehicle theft. Data are not included for the property crime of arson.
Source: Modified from U.S. Department of Justice. FBI Uniform Crime Reports: Crime in the United States, 2006.
Washington, D.C.: U.S. Government Printing Office, September 2007
Washington, D.C.: U.S. Government Printing Office, September 2007.
Chapter 4 • The Uniform Crime Reports and Sampling 101
what you must never forget is that every one of these figures comes in the first
instance from the chowty dar (village watchman), who puts down what he damn
pleases (Stamp, 1929, pp. 258-259; cited in Webb et al., 1981, p. 89).
Although the UCR has been steadily improved and refined in its more than half a century
of existence and can serve useful purposes, a researcher using these data for the analysis of crime
must exercise caution and be aware of the limitations of these statistics. An impressive body of
literature on the UCR has accumulated that point out these deficiencies.
Although no attempt will be made here to summarize the many fine critiques of the UCR and
many of the points to be elaborated overlap, some of the primary shortcomings of this source of
crime data can be detailed:
true crime rate of a community. Although we will examine victim surveys in detail later,
the Bureau of the Census victim surveys suggest that there is possibly twice as much crime
committed as appears in the official statistics.
the 1970s was in part explained by better communication, more professional police
departments, and better recording and reporting of crimes. Surprisingly, for instance,
there appears to be a positive relationship between a larger, improved, and professionalized
police establishment and rising crime rates. Higher urban crime rates are
related in part to the fact that there is a higher proportion of formal policing and
professional law enforcement in such areas. Increased urbanization itself also has its
have resulted in a larger proportion of crimes committed being reported to the police.
The usual and expected Saturday night barroom brawl of fifty years ago is now regarded
as a serious assault warranting police attention. In light of the success of the civil
rights movement, many more inner-city residents report crimes to the police, expressing
confidence in the greater willingness of the police to respond to crime in their
in the UCR. Many of the analyses of age, sexual, and racial characteristics of those
arrested, then, describe the inept and poor criminal or concentrate on "crime in the streets"
rather than "crime in the suites."
simply, more attention to better record keeping have had a major impact on crimes
recorded. Obvious inadequacies in statistics from a jurisdiction are checked out by the
FBI. In 1949, for instance, the FBI refused to publish New York City Police Department
statistics. With improved recording, the robbery rate jumped 800 percent. Other changes
in police practices showed jumps of 61 percent in Chicago in 1961 because of a change
in chief, 202 percent in Kansas City in 1959 because of reforms, and 95 percent
in Buffalo in the early 1960s (President's Commission, 1967, p. 25). In 1998, the
Philadelphia Police Department was accused of routine downgrading of crime reports in
which stabbings and beatings were recorded as hospital cases and burglaries became lost
102 Chapter 4 • The Uniform Crime Reports and Sampling
property (Thousands of Rape Claims, 1999). In 2004, it was charged that the city of
Atlanta had understated its crime rate by 22,000 crimes shortly before being approved
for hosting the Olympics. In 1999, Philadelphia's Sex Crimes Unit dismissed as noncrimes
several thousand reports of rape (Associated Press, 2004). In a humorous example
of questionable statistics, the St. Paul, Minnesota, police department reported having
solved 215 rape cases in 1999, but reported that only 200 rapes had been committed. It
had also reported solving eleven more than that took place in 1997. The department
claimed a bookkeeping error. It turns out that the department had been too generous in
labeling cases as "cleared" beyond the guidelines provided by the FBI. The national
clearance rate for rape is about 50 percent.
Crime rates have a mysterious way of dropping if required for political purposes.
Nixon's identification of the District of Columbia as a target of a crime-busting program
resulted in a dramatic drop in the crime rate that was more likely a matter of classifying
crimes out of the index. Until 1973, grand larceny of $50 or more could simply be classified
as under $50 and thus out of the index (Glaser, 1978, p. 58). An interesting example of
statistical shenanigans occurred in the mid-1970s during police negotiations in Cleveland
under the administration of Mayor Ralph Perk. The police were told that if they expected a
5 percent salary increase, they would have to demonstrate greater efficiency by dropping the
crime rate by a similar amount. Predictably, the crime rate recorded a drop in index crimes.
Sherman (1998) has argued that the more crime data is used to evaluate the police, the
greater the temptation is to cheat. He proposed creating separate, independent crime audits
as a solution. Such manipulation of crime statistics is not limited to the United States.
In 2008, charges were levied by criminologists that the British Home Office was manipulating
research on crime by deliberately withholding or disrupting it. The office was accused
of suppressing statistics and studies that might reflect poorly on government policies
crime index. He summarizes a large number of elements that at the very least must be considered
arrest statistics do and do not include. Arrests do not equal crimes solved or suspects
found guilty. Many potential crimes are also unfounded by police. These are complaints
that, on investigation by police, are determined not to be criminal matters.
purposes. Most crimes that are committed are not index offenses. Questions have been
raised that auto theft, a less serious crime with high reportability and clearance, artificially
inflates the crime index and might best be dropped from the Part 1 designation
index crime. In 1973, all larcenies were included as index offenses, thus sharply increasing
the crime rate (Rhodes, 1977, p. 168). At the same time, greater insurance coverage of
property crimes encourages their reporting.
index; that is, a murder counts the same or has the same weight as a bicycle theft. Imagine
two cities each with a crime rate of 100 per 100,000 population. In City A, 100 murders
were recorded, whereas in City B 100 joyrides were recorded. Somehow, these crime rates
of the crime
Chapter 4 • The Uniform Crime Reports and Sampling 103
should not be regarded as the same. Most bodily injury crimes are "nonindex" or Part 2
crimes (Savitz, 1978).
crimes at the expense of other crimes.
been included in the crime index because of the unreliability of arson offense information.
Jackson (1988) compared UCR arson data with a national survey of 683 fire departments
and found significant underreporting in the UCR. He explains that arson is unlike other
index offenses since determining whether a fire was purposefully set or attempted requires
a specific investigation.
media are fond of ranking cities with respect to index crimes the UCR itself correctly
cautions against comparing statistical data of individual reporting units from cities,
counties, and states solely on the basis of their population coverage. Such features as
population density, size of locality, age structure, population mobility, economic and
cultural conditions, climate, strength of law enforcement agencies and their policies,
politics, citizen attitudes, and reporting practices are all factors that complicate such
RELATED UCR ISSUES
Although it does not speak to any inaccuracy in the UCR itself, demographic shifts may provide
some explanation for rapidly rising or falling crime rates. Some criminologists had forecast the
crime dip in the 1980s, a general stabilization or decline in the crime rate. Although other
factors offset the decline, this prophecy was based on changes in population distribution. For
UCR crime rates, the maximal age range of criminality is 15-19, or 20-24 if an adult cohort is
considered. After World War II, the United States experienced an unprecedented "baby boom," a
much larger than usual proportion of births beginning in 1946 and extending through the 1950s.
This group at first overwhelmed the capacity of hospital nursery wards; later, elementary schools
and high schools were hard pressed to meet the demands for space. In the late 1960s, colleges
could not expand fast enough to accommodate demand. Today most of these sectors have either
stabilized or declined in demand as the economy struggles to supply jobs and housing for the
now mature "baby boom" group. Similarly, the criminal justice system was struggling to deal
with the larger than normal group in the maximum crime-committing ages. As this group moves
into middle age and a smaller proportion of the population will be found in the high-crime ages,
the criminal justice system will hopefully find itself with a far more manageable situation.
A counterbalance to this demographic shift is the relationship between arrest rates and race.
If the birthrates and crime rates of minorities remain higher than the general population, we may
not see much of a drop in the crime rate.
Other variables that may also explain this decline include a greater reliance on longer
sentences, incarceration and incapacitation of serious offenders, a decline in the number of state
UCR programs, and an almost unnoticed decline in the per capita number of police in the United
States since during the 1980s. The researcher who decides to make use of official statistics such
as the UCR must become familiar with their inadequacies to avoid drawing inappropriate
conclusions. Despite the shortcomings that have been identified, the UCR remains an excellent
source of information on police operations. Exhibit 4.1 presents attempts to explain the crime dip
of the 1990s and the early twenty-first century.
104 Chapter 4 • The Uniform Crime Reports and Sampling
From the earliest compilation of national crime
statistics in the 1930s by the FBI in its Uniform Crime
Reports (UCR) until the early 1960s, the crime rate in
the United States had been declining, so much that
some prognosticators had unwisely predicted that,
given existing trends, growing affluence in America
might make crime a rarity by the twenty-first
century. By the mid-1960s, recorded crime did an
abrupt about face and rose at unprecedented levels,
producing yet new forecasts of unrepentant
explosions in the crime rate. A slight leveling off in
the early 1980s was interrupted by an epidemic of
youth violence beginning in the mid-1980s with the
advent of crack cocaine and the ready willingness to
use weapons to defend warring turf. By the 1990s,
just at the time that an assumed inevitability of such
high crime rates was setting in, an unexpected
decline took place, particularly in large cities.
Between 1993 and 1997, index crimes dropped over
The cause(s) of this decline remains a subject
of dispute. Some factors associated with the decline
in the crime rate in the 1990s include (Witkin, 1998,
a healthy economy
crime prevention programs
decline in domestic violence
an incarceration binge
Compstat and community policing
decline in the crack epidemic
The healthiest U.S. economy in over thirty
years characterized by low unemployment and low
inflation has been viewed as primarily responsible
for falling crime rates. The relationship is not entirely
clear, however. In the 1960s, crime rates rose sharply
during periods of low unemployment. Sunbelt cities
with low unemployment had higher crime rates
than older cities with high unemployment. In the
1990s, murder in New York City fell over 66 percent
despite high unemployment (ibid., p. 30).
Crime prevention, which holds much promise
with early intervention programs with high risk
youth, has thus far demonstrated only modest
influence on crime rates.
Domestic murders, those among intimates,
showed a 40 percent decrease between 1976 and
1996. This is explained in part by a decline in
marriages among 20-24 year olds as well as
improved opportunities for women to escape
The imprisonment binge in the United States
has been phenomenal from about 744,000 in 1985
to 1,726,000 individuals in 1997 incarcerated in
federal, state, and local prisons and jails. This
represents the largest incarcerated population in the
world outside Russia. Locking up an additional million
prisoners must have an impact; however, "New York
City has displayed some of the most dramatic drops in
crime, but the state prison population-70 percent of
which is from the city-has increased only about
8 percent since 1993. Conversely, the law-and-order
state of Utah raised its incarceration rate by
19 percent from 1993 to 1996-but its violent-crime
rate went up" (ibid., p. 31).
Better policing has also been credited with
the decline. New York City, which showed sharp
decreases, utilized "compstats" (computer or
compare statistics) to computer map and identify
"hot spots" (high crime areas) where they would
concentrate their policing efforts. A focus on small,
nuisance crimes as suggested by Wilson and
Kelling's "Broken Windows" (1982) theory was
pointed to. This postulates that small crimes left
unpunished breed more serious crimes. Nationwide,
however, some cities that employed no new
strategies also experienced plunging crime rates as
others who innovated saw rising ones.
The end of the crack cocaine epidemic, which
began rising in 1986, is a promising explanatory factor.
This epidemic had been combined with the carrying of
guns in order to secure operations. These same guns
were then used to settle disputes previously resolved
by fistfights. By the early 1990s, the crack crisis waned,
as did the violent crime rate. In some cases, territories
stabilized, and in others, the ravages of the drugs on
family members as well as incarceration took its toll.
The Crime Dip
EXHIBIT 4.1 (Continued )
A final, controversial explanation is posed by
Levitt and Donohue (1999), who propose that the
advent of legalized abortion with the 1973 Roe v.
Wade decisions explains half the drop in crime
since 1991. Unwanted, potential criminals were
not born and crime declined beginning in 1992,
just when they would have reached their peak
crime years (18-24).
Source: Witkin, Gordon. "The Crime Bust: What's behind
the Dramatic Drug Bust?" U.S. News and World Report,
May 25, 1998, pp. 28-37.
Beginning in 1977, the International Association of Chiefs of Police, as well as the National
Sheriffs Association, called for a major redesign of the UCR system, and after more than ten years
of effort by committees, task forces, and project staffs, the first major revision in the sixty-year
history of the program was accomplished (Poggio et al., 1985; Rovetch, Poggio, and Rossman,
1984). The lengthy process of the redesign study enabled redesign teams to take advantage of the
decade's revolution in computer technology.
Reporting System (NIBRS), which involves a unit-record reporting system in which each local
law enforcement agency reports on each individual crime incident and on each individual arrest.
The original UCR only reported summary counts. The NIBRS uses fifty-two data elements to
describe victims, offenders, arrestees, and circumstances of crimes. While the UCR focuses on
eight index crimes, NIBRS has twenty-two Group A offenses, including bribery, counterfeiting/
forgery, drug and narcotic offenses, extortion/blackmail, fraud, kidnapping, pornography, nonforcible
sex offenses, and firearms violations.
Participation by law enforcement agencies in the program depends upon a department's
data processing resources. If an agency cannot meet full participation requirements, it may limit
itself to reporting details of incidents involving the UCR's eight index crimes rather than the
NIBRS's expanded list of twenty-two crimes (Dodenhoff, 1990, p. 10). NIBRS divides data
collection into two levels. Level I covers all law enforcement agencies and requires basic Group
A incident-based data on the twenty-two categories of offenses. Level II participation includes an
additional eleven-category Group B list of lesser offenses, as well as more detail in submissions.
Agencies serving populations in excess of 100,000 plus a sampling of 300 smaller agencies will
participate in Level II. Exhibit 4.2 reports on features of the NIBRS.
Some key differences between the NIBRS and the UCR program include (Rantala, 2000):
Parts I and II arrest data in aggregate (summary) form. NIBRS, which requires detailed data
on individual crime incidents and arrests, receives separate reports for each incident/arrest.
These reports include fifty-two data elements describing the victims, offenders, arrestees,
and circumstances of the crime.
106 Chapter 4 • The Uniform Crime Reports and Sampling
Theft from building
Theft from coin-operated machines
Theft from motor vehicle
Theft of motor vehicle parts/accessories
All other larceny
Assisting or promoting prostitution
Sex offenses, forcible
Sexual assault with an object
Sex offenses, nonforcible
Stolen property offenses
Weapon law violations
The National Incident-Based Reporting System (NIBRS)
With 1991 data, the Uniform Crime Reports (UCR)
program of the FBI began moving from summary
counts to a more comprehensive and detailed
reporting system known as the National Incident-
Based Reporting System (NIBRS). By 1982, the Bureau
of Justice Statistics (BJS) had already provided over
$11 million to the states to establish centralized statelevel
UCR programs. In a 1985 report, an FBI-BJS task
force that BJS underwrote recommended an incident
based system. When the Attorney General approved
NIBRS, BJS allocated an additional $13 million to the
states to implement the system.
NIBRS versus the Traditional UCR System
The traditional, summary-based UCR system counts
incidents and arrests, with some expanded data on
incidents of murder and nonnegligent manslaughter.
NIBRS, which will eventually replace the traditional
UCR as the source of official FBI counts of crimes
reported to law enforcement agencies, is designed to
go far beyond the summary-based UCR in terms of
information about crime.
One important difference between the two
systems is number of crime categories. The traditional
UCR counts incidents and arrests for the eight offenses
of the FBI Crime Index and counts arrests for other
offenses; NIBRS provides detailed incident information
on forty-six Group A offenses representing twentytwo
categories of crimes (Table A).
NIBRS, unlike the traditional UCR, also makes
a distinction between attempted and completed
TABLE A The NIBRS Group A Offenses
Burglary/breaking and entering
Destruction/damage/vandalism of property
Drug equipment violations
False pretenses/swindle/confidence game
Credit card/ATM fraud
Gambling equipment violations
EXHIBIT 4.2 (Continued )
In September 1982, a BJS-FBI task force
undertook a study of improvements to the Uniform
Crime Reporting Program, which was created in
1930. Law enforcement organizations, state UCR
program managers, and the research community
strongly supported this effort. In January 1984, a
conference considered various recommendations,
and in 1985, BJS and the FBI released the Blueprint for
the Future of the Uniform Crime Reporting Program.
The resulting program, the NIBRS, collected
its first data in 1991. An estimated 40 percent of
the nation reported to the NIBRS by the end of
1994. NIBRS represents a new way of thinking
about crime, providing details about victims,
offenders, and the environments in which they
interact. This report on the first NIBRS data is a
beginning step toward using the data for planning
and evaluating law enforcement responses to crime.
It also illustrates the close partnership among BJS,
the FBI, and the more than 17,000 State and local
law enforcement agencies.
NIBRS collects arrestee information on the
forty-six Group A offenses and an additional eleven
Group B offenses (Table B). Unlike the traditional
UCR, NIBRS requires arrests as well as exceptional
clearances to be linked to specific incidents.
TABLE B The NIBRS Group B Offenses
Bad checks Drunkenness Runaway
Curfew/loitering/vagrancy Liquor law violations Trespassing
Disorderly conduct Nonviolent family offenses All other offenses
Driving under the influence Peeping Tom
In addition to expanded crime categories,
NIBRS definitions of certain offenses are more
inclusive than the traditional UCR definitions. For
example, the NIBRS definition of rape has been
expanded to include male victims.
In incidents where more than one offense
occurs, the traditional UCR counts only the most
serious of the offenses. NIBRS includes information
about each of the different offenses (up to a maximum
of ten that may occur within a single incident). As a
result, the NIBRS data can be used to study how often
and under what circumstances certain offenses, such
as burglary and rape, occur together.
The ability to link information about many
aspects of a crime to the crime incident marks the
most important difference between NIBRS and the
traditional UCR. These various aspects of the crime
incident are represented in NIBRS by a series of
more than fifty data elements (Table C).
The NIBRS data elements are categorized in six
segments: administrative, offense, property, victim,
offender, and arrestee. NIBRS enables analysts to
study how these data elements relate to each other
for each type of offense.
Administrative segment includes the ORI
(originating agency identifier) and incident numbers
that uniquely identify each incident. These tie
together all the records of a single incident. The
administrative segment also includes the date and
hour of the incident and if relevant, exceptional
Offense segment includes the type of
offense(s) reported, whether the offense was
attempted or completed, whether the offender was
suspected of using drugs or alcohol, the type of
location where the offense occurred (such as a store
or residence), the type of weapon or force used,
and whether the offender was motivated by bias
against the victim's race, religion, ethnicity, or sexual
orientation. For certain offenses, the type of criminal
activity (such as possessing, selling, or transporting)
is indicated. For burglary incidents, the method of
entry and the number of premises entered are
Property segment includes (for all property
offenses, extortion, kidnapping, and a few other
specified offenses) the type of property loss (burned,
counterfeited, destroyed, seized, or stolen), the type
of property involved (such as cash or jewelry), the
value of the property, and, if recovered, the recovery
date. For incidents of motor vehicle theft, special
indicators for the number of stolen and recovered
vehicles are included. For drug offenses, the type and
quantity of illegal drug(s) seized are included.
108 Chapter 4 • The Uniform Crime Reports and Sampling
TABLE C NIBRS Data Elements
1. ORI number
2. Incident number
3. Incident date/hour
4. Exceptional clearance indicator
5. Exceptional clearance date
6. UCR offense code
7. Attempted/completed code
8. Alcohol/drug use by offender
9. Type of location
10. Number of premises entered
11. Method of entry
12. Type of criminal activity
13. Type of weapon/force used
14. Bias crime code
15. Type of property loss
16. Property description
17. Property value
18. Recovery date
19. Number of stolen motor vehicles
20. Number of recovered motor vehicles
21. Suspected drug type
22. Estimated drug quantity
23. Drug measurement unit
24. Victim number
25. Victim UCR offense code
Victim segment includes a victim identification
number, the UCR code for offense(s) committed
against the victim, and the victim's sex, age, race,
ethnicity, and residential status. In cases where the
victim is not an individual, codes are used to
distinguish among businesses, financial institutions,
governments, religious organizations, and society at
large. For incidents of homicide or aggravated
assault, codes describing the circumstances of the
incident (such as an argument or drug deal) are
provided. In incidents where the victim is injured,
information describing the injury (such as fractures
26. Type of victim
27. Age of victim
28. Sex of victim
29. Race of victim
30. Ethnicity of victim
31. Resident status of victim
32. Homicide/assault circumstances
33. Justifiable homicide circumstances
34. Type of injury
35. Related offender number
36. Relationship of victim to offender
37. Offender number
38. Age of offender
39. Sex of offender
40. Race of offender
41. Arrestee number
42. Transaction number
43. Arrest date
44. Type of arrest
45. Multiple clearance indicator
46. UCR arrest offense code
47. Arrestee armed indicator
48. Age of arrestee
49. Sex of arrestee
50. Race of arrestee
51. Ethnicity of arrestee
52. Resident status of arrestee
53. Disposition of arrestee under 18
or lacerations) is included. Each victim is linked by an
offender number to the offender(s) who committed
an offense against him or her, and the nature of the
victim's relationship (such as family member,
acquaintance, or stranger) to each offender is
Offender segment includes information on
the age, sex, and race of the offender.
Arrestee segment includes information on
persons arrested in connection with the incident,
including the date of arrest, the age, sex, race,
ethnicity, and the residential status of the arrestee.
An example of how those interested in the
study of crime can tap the potentially rich source
of new information represented by NIBRS is seen in
the Supplementary Homicide Reports data
United States series. Cross tabulations of various
incident based data elements are presented,
including the age, sex, and race of victims and
offenders, the types of weapon(s) used, the
relationship of the victim to the offender, and the
circumstances surrounding the incident (e.g.,
whether the murder resulted from a robbery, rape,
or argument). These data were provided to the FBI
for about 87 percent of the 24,703 murders
reported nationwide in 1991.
For other violent crimes such as rape and
robbery, UCR data beyond the summary counts have
generally been limited to a univariate distribution by
month. With the advent of NIBRS, the supplemental
data elements that were previously available only for
murder incidents can now be used in the analysis of
other violent crimes.
Of course, NIBRS also provides some data
elements that were not previously available for any
violent crimes, including murder. These new data
elements include whether the offender was suspected
of using alcohol or drugs shortly before or during the
incident, the type of location of the crime, the
residential status of the victim, and the nature of any
injuries sustained by the victim.
For robbery incidents, NIBRS also provides
previously unavailable data describing the property
that was lost and its value. Using NIBRS, a researcher
could study carjackings, for example, by selecting
robbery incidents that included a vehicle as the
Source: Reaves, Brian A. "Using NIBRS Data to Analyze
Violent Crime." Bureau of Justice Statistics Technical
Report, October 1993.
criminal incidents in eight offense classifications within the Part I type. NIBRS receives
detailed reports on twenty-two categories and forty-six offenses in the Group A list.
It adds the following list to the original UCR Part I crimes: bribery, counterfeiting and
forgery, vandalism, drug offenses, embezzlement, extortion and blackmail, fraud, gambling
offenses, kidnapping, pornography, prostitution, nonforcible sexual offenses, weapons law
violations, and stolen property offenses. In addition, many of the Part I offenses have been
expanded. For example, the forcible rape category now includes all forcible sexual offenses,
such as forcible sodomy, sexual assault with an object, and forcible fondling.
NIBRS revised the existing definitions of crime. Rape, for example, is defined as "the carnal
knowledge of a person, forcibly and/or against that person's will; or, not forcibly or against a
person's will where the victim is incapable of giving consent because of his/her temporary or
permanent mental or physical incapacity" (ibid.).
place within the same event, only the single most serious crime was reported. NIBRS eliminates
the hierarchy rule and cites all crimes reported as offenses within the same incident.
incidents, NIBRS data will eventually lead to more detailed crime analysis, criminal
profiling, and crime reporting. NIBRS will also have the capability of providing breakdowns
regarding victims, cost, involvement of weapons, injuries, and the like, innovations
that had not been possible in the past.
and "crimes against property," the addition of many new offense categories in NIBRS
necessitated the creation of a new category-"crimes against society." This category
includes crimes such as drug offenses, gambling violations, pornography, and prostitution.
110 Chapter 4 • The Uniform Crime Reports and Sampling
completed ones. The NIBRS system will include a designation of each crime as either
attempted or completed.
determine whether a traditional crime, for example larceny, was committed by computer.
But this specificity will not eliminate the traditional classifications that are important for
historical trend analysis.
between many variables such as offenses, property, victims, offenders, and arrestees.
These features represent the first major overhaul of the UCR system in more than fifty years.
On a final note, mention should be made of the growing international effort with respect
to crime statistics. Organizations such as Interpol, the United Nations, and the World Health
Organization all have programs of collection. Some of the problems in analyzing these data
are the same as those affecting the UCR. Difficulties in analyzing crime data across countries
include: varying definitions of crime; differences in recording practices; differences in the law;
the stage of the system when crime is recorded; factual inequalities among countries in age
and urban/rural structure; and specific problems associated with recording crime (statistics
may be related to politics or measures of system workload) (van Dijk and Kangaspunta, 2000).
Sampling may be used with any of the data-gathering procedures we discuss. The fact that it is
included here with surveys is simply a matter of editorial convenience. Some research involves a
complete enumeration of the total population, households, or the target of study. Ever since
1790, the U.S. Census has attempted to survey every household unit-man, woman, and child-
every ten years. Similarly, city directories attempt to count and obtain information on all persons
eighteen years of age or older who reside within their urban target areas.
Rather than attempting to enumerate an entire population, most studies make use of
sampling. Sampling is a procedure used in research by which a select subunit of a population is
studied in order to analyze the entire population. Sampling enables an inexpensive, relatively
quick assessment, by even small groups of researchers, of a population that is often so large that
complete enumeration is prohibitive. The logic of sampling enables one to make inferences to a
larger population (Kish, 1965).
The initial step in selecting a sample is to develop a sampling frame, a complete list of the
population (or universe) that one is interested in studying. For example, if one is interested in generalizing
all judges in California, a complete list of such judges would constitute the sampling frame.
TYPES OF SAMPLING
The major types of sampling procedures follow.
Simple Random Quota
Stratified Random Accidental
Systematic (Multistage) Snowball
Probability samples refer to samples that permit estimation of the likelihood of each element of
the population being selected in the sample.
Simple Random Samples. Simple random samples (SRSs) are samples in which each
element of the population (or universe) has an equal probability of being selected. Sometimes
the mnemonic device EPSEM samples are used to denote the key features of an SRS (Babbie,
1992, p. 197).
EPSEM, a means of sample selection, is an acronym that stands for Equal Probability of
SElection Method. This method provides a way for selecting a sample in which each and every
unit or person in the population has the same or equal chance of appearing in the sample.
EPSEM or probability samples are very important in the field of statistics because the various
calculations and estimations of statistics assume that the sample was chosen by some probability
method. In describing samples that use an EPSEM, we will use the shorthand acronym EPSEM.
If probability methods have been utilized in selection of the sample, the concept of sampling
error enables researchers to assess confidence limits so that with a given degree of error,
they can assume that what is true of the sample is true of the population and that the sample mean
approximates that of the population.
Procedure. To select an SRS, it is necessary to acquire a clear and complete list of all elements
of the population because all elements must be independently and randomly chosen. Suppose
there were thirty people in a room and a simple random sample of five were to be drawn. One
could give each person a number, drop these numbers into a hat, scramble them, and then draw
five, one at a time. State lottery daily numbers usually make use of an honest gambling device
procedure that is essentially an SRS. What if, as in a large survey of the public, one wished to
draw a simple random sample of adults from a city of a million? Obviously, one would not put
numbers in a hat. To sample such large populations, researchers make use of a table of random
numbers. Figure 4.3 illustrates a theoretical population and a hypothetical typical table of random
numbers. Appendix B contains a larger table of random numbers.
To select a sample of ten inmates, one would first number the list of inmates, then choose
a random start, for example, the top left of Table of Random Numbers. As the entire population
consists of fifty cases, numbers from 00 to 99, or two-digit numbers, would enable each name to
have an equal probability of selection. Numbers from 51 to 99 are, of course, unusable; if these
numbers are chosen, they should be skipped and the selection process continued until the next
two-digit number between 01 and 50. According to Figure 4.3, the first number is XXXXX
Bruno, the second is 00-no case, the third is 16-Mike Federici, and so forth, until ten cases are
chosen. If the same number is XXXXX twice, it is skipped because each respondent should appear
only once in the sampling frame. For complex sampling, various computer programs are available
that provide an SRS of a specified size.
Advantages/Disadvantages. The chief advantage of the SRS is that it enables the use of statistical
probabilities that are necessary in many statistical procedures. The primary disadvantages of
the SRS, however, are that it requires a complete list of the population to be sampled and, if large
numbers are involved, it can become a rather tedious and cumbersome procedure, although this
can be offset by computer. The SRS by no means guarantees a representative sample. On the last
point, by chance it is possible in our San Rocco sample to obtain a sample that is 50 percent
female, even though females represent only 10 percent of the population. As we will see in
our discussion, the probability of this occurring is small, but it certainly is possible. Such a
112 Chapter 4 • The Uniform Crime Reports and Sampling
nonrepresentative sample certainly raises problems for a researcher attempting to infer to the
larger population. Primarily for this reason, much survey research involving sampling utilizes
stratified random samples.
Some examples of research employing SRSs include a study of New York City drug laws by
Japha (1978) in which he randomly selected cases from the Criminal Court of Manhattan of persons
convicted for a nondrug felony who had been given a nonincarceration sentence. In addition,
he drew random samples of cases entering court for arraignment, cases reduced or dismissed at
first arraignment, clients in drug treatment programs, and males held on felony charges in
Manhattan. Sparks (1982), in a study of Massachusetts statewide sentencing guidelines,
constructed a random sample of 1,440 convicted criminals who had been sentenced in the
Massachusetts Superior Court during a one-year period.
Stratified Random Samples. Stratified random samples rely on knowledge of the distribution
or proportion of population characteristics to choose a sample that assures
representativeness of these characteristics. Such characteristics are generally demographic in
nature, such as age, sex, race, social class, or of pertinence to the study, such as area of residence,
nature and type of criminal record, region, or some quality of importance in the analysis.
The general procedure involves dividing the population into strata or groups based on
the variable(s) of stratification and then selecting the sample either proportionately or disproportionately,
samples, sample subjects are chosen in roughly the same ratio as exists in the population.
For instance, suppose that in our San Rocco study of fifty inmates, we wanted to choose a
proportionate stratified sample by sex of ten inmates. Because one of ten is a female in the
population, we must be certain that only one of the ten subjects in the sample is female. Such
a procedure assures representativeness by sex, unlike the SRS in which half of the sample
FIGURE 4.3 Sampling Frame of Inmates at San Rocco Correctional Institution.
*This table of random numbers is provided for illustration purposes only. For actual projects, consult Appendix B.
Angelo, Gerald Clemons, Randy Kozak, Dave Parks, Zeke Thiel, Myrtle
Bell, Earl Dammer, XXXXX XXXXX, Ed Penn, Wally XXXXX, XXXXX
Bender, Harry Dutkowsky, Andrew Mack, Bob Quick, Bob Tierney, Estelle
Benekos, Peter Edsel, Earl McGill, Bill Rapp, Sean Unterwagner, Jim
Bethune, David Erisman, XXXXX XXXXX, XXXXX Rasp, Doug Vance, Lance
Bozo, Boris Frederici, Mike Mucha, XXXXX XXXXX, XXXXX Vega, Terence
Bruno, Albert Goblick, Al Norris, Herb Runt, Juan Wahlen, John
Burns, Rich Hairbreath, Harry Numa, Tod Saxon, Sid Wayne, Mike
Buxton, Bob Harlow, Joe Obernan, Stan XXXXX, XXXXX Weeks, Bary
Buzawa, Eve Johnson, LeXXXXX XXXXX, Omar Simpson, Ted Zeno, Mike
TABLE OF RANDOM NUMBERS*
07001 61569 08812 07344 92880 71728
43102 29751 87806 12031 56214 41387
61622 71481 20091 37658 99612 28143
50126 51296 07509 61483 25143 619747
Disproportionate stratified sampling involves oversampling-taking a larger than proportionate
number of certain groups to assure the appearance of a sufficient number of cases for comparative
purposes of a group that is small in the population. Again, returning to our example in Figure 4.3,
suppose that we wished to investigate differences between male and female inmates at San Rocco.
An SRS could result in a sample of all males, which would certainly destroy our ability to even
conduct the study. A proportionate stratified sample would yield one female and nine males, a
situation that would be quite hazardous because, on every variable of analysis, the 100 percent
response of females would be referring to only one respondent. A disproportionate stratified
sample might take all five female subjects and compare them with a sample of male respondents,
for example, five males. There is generally no problem in comparing males with females using a
disproportionate stratified sample; however, if inferences were to be attempted from a sample that
is overrepresentative of females to all inmates, the sample is obviously nonrepresentative.
Weighting of sample responses is a recommended procedure to adjust sample data to enable
inference to the general population. Basically, weighting involves the differential assignment of
adjustment factors to data to take into account the relative importance of that data.
Table 4.2 illustrates this process.
The responses of the males in the sample would in actuality carry nine times more weight than
those of females. That is, each response of males in the sample actually represents the response of
nine males-the respondent and eight others-whereas the female respondent represents only
herself. Thus, disproportionate stratified sampling permits comparisons between subgroups where at
least one of the subgroups might otherwise be too small. In the early victimization surveys of select
U.S. cities conducted by the U.S. Census Bureau on behalf of the Department of Justice, the
of Population and Housing. To select a stratified sample, the city's housing units were categorized
into 105 strata, for example, own or rent, occupied or unoccupied, and single-family or multipledwelling
unit (Criminal Victimization Surveys in Milwaukee, 1977). Garofalo describes the
sampling procedure utilized in the initial study of eight cities involved in Law Enforcement Assistant
Administration's (LEAA's) High Impact Crime Reduction Program:
Supplemental samples were drawn from new construction permits issued in each
city. Census Bureau interviewers visited the housing units selected and interviewed
residents about personal and household victimizations suffered during the preceding
twelve months. About 10,000 households or 22,000 individuals were interviewed in
each city. . . . The samples were sufficiently large to make reliable estimates of what
the attitude responses would have been if everyone in the city had been interviewed
(Garofalo, 1977, p. 14).
The numbers in the victimization survey report are weighted estimates as if the entire population
were surveyed. The history of the operation of Bureau of Justice Statistics-sponsored
victimization surveys will be covered in depth in Chapter 6.
TABLE 4.2 San Rocco Correctional Institution: Weighting
of Disproportionate Sample
Stratified Sample Weight
Male 45 5 9 ×
Female 5 5 1 × or none
114 Chapter 4 • The Uniform Crime Reports and Sampling
In an analysis of characteristics of high- and low-crime neighborhoods in Atlanta,
Greenberg, Williams, and Rohe (1982) utilized a stratified random sample from three matched
pairs of neighborhoods selected on the basis of crime, racial, and income characteristics. A study
of the relationship between narcotics addiction and criminal activity in Baltimore by Nurco et al.
(1985) involved a sample of 354 male narcotic addicts who were selected using a stratified
random sample of a population of 6,149 known narcotics abusers who had been arrested or identified
by the Baltimore Police Department between 1952 and 1976. The sample was selected not
on the basis of criminality but by race and year of police contact.
Cluster Sampling. Cluster sampling is generally used in surveys that involve field interviews
and is most useful in studies that involve widely dispersed subjects. The population to be surveyed
is divided into clusters, for example, census tracts, blocks, and sections, and then a probability
sample of clusters is selected for study. Such a sampling procedure is less time consuming and
costly, particularly in terms of field staff. Once the clusters are chosen, other sampling procedures,
such as a systematic sample of every nth house, may be employed. Cluster sampling is particularly
useful as a means of reducing travel costs in field interviewing.
An example of the use of cluster sampling is provided by Schuerman and Kobrin's
(1986) study of neighborhood change and criminal activity in Los Angeles. They drew a sample
from census tract clusters in Los Angeles County that were defined as high-crime areas in
1970. They then used a statistical procedure to assemble contiguous census tracts into 192
clusters or neighborhoods and studied the impact of socioeconomic and demographic trends in
these areas on crime rates. Sigler and Johnson (1986; see also Sigler and Haygood, 1988), in a
study of public perceptions of sexual harassment in Tuscaloosa, Alabama, employed a multistage,
stratified cluster sample in which grids on city maps, blocks, and residences were the
Systematic Samples. In systematic samples every nth item in a list is included in the sample. (In
the language of statistics, n represents every second, third, fourth, or nth case.) Purists insist that
such a sample is a nonprobability sample, because various patterns, for example ethnic surnames,
may exist in a list that would destroy its representativeness. If offenders or arrestees were listed in
order of offense seriousness, the final sample may be biased. This writer has chosen to place
systematic sampling in the probability group because the majority of researchers feel that it
satisfies the EPSEM requirement and belongs in the probability group. To illustrate systematic
sampling, let us return to Figure 4.3, the San Rocco example. Suppose we wished to select a
sample often from a population of fifty. Assuming the names are XXXXX XXXXX we would
first select the proper sampling interval, in this case every fifth name. Sampling intervals are
selected by the ratio of sample size to population size-in our example, ten of fifty or one of five.
By choosing every fifth, theoretically every name in the population list has an equal probability of
being chosen so long as one uses a random start. Random start involves randomly choosing where
the interval will begin within the first interval-in this case, 1 to 5. For example, if the number 3
were chosen from a table of random numbers, the sample selected with the every-fifth-sampling
interval would be 3, 8, 13, 18, 23, 28, 33, 38, 43, and 48. As long as one both suspects that there is
no pattern in the population list and uses the proper sampling interval and random start, such that
each individual or unit has an equal probability of being chosen, a probability sample exists. The
obvious advantage of systematic samples is their relative ease of selection, although it may
become burdensome with large populations. In choosing a systematic sample from a uniformly
spaced list of names, a simple procedure, once the random start is selected, is to mark off with a
ruler or other measuring rod and proceed down the list until all cases are selected.
Multistage Sampling. Multistage sampling involves combinations of stratified, cluster, simple
random samples, and/or other sampling procedures. For example, a national survey of neighborhood
crime might stratify first on the basis of region-north, east, south, west. Within regions,
clusters are randomly selected, and within the selected regions, blocks are randomly selected for
door-to-door household interviews. Multistage sampling can become quite complex, as illustrated
by the study of media crime prevention campaigns by O'Keefe et al. (1984).
The population examined included a national sample of the noninstitutionalized U.S.
civilian population of age eighteen and over. A one-call quasi-probability sample design was
employed, based upon the Roper Organization's master national probability sample of interviewing
areas. First, 100 counties were chosen at random proportionate to population after all
counties in the nation had been stratified by population size within geographic region. Second,
cities and towns were randomly selected from the sample counties according to their population.
Third, four blocks or segments were then drawn within each location. Quotas for sex and
age, as well as for employed women, were set in order to assure proper representation of each
group in the sample (O'Keefe et al., 1984, in Loftin, 1987, p. 100).
Any sampling procedure that violates the EPSEM is viewed as a nonprobability sample.
Quota Samples. Quota samples are nonprobability stratified samples. The researcher attempts to
ensure that the sample proportions, for example, age, sex, and race, resemble those of the population,
but does not fill these proportions or quotas on the basis of an EPSEM. Rather than attempting to
ensure that each element of each quota has an equal chance of appearing in the sample, the
researcher uses skilled judgment to select adequate numbers to fill each quota. The data are collected
and analyzed on an ongoing basis until an adequate decision or prediction of outcome becomes
possible. Quota sampling is the favorite technique of many private marketing and consumer survey
organizations. Often at shopping malls, interviewers eyeball shoppers until someone appearing to fit
the requirements of one of their quotas is identified-for example, a black male in his forties. At
times, the interviews are aborted when, on the basis of demographic information, it turns out that the
interviewer guessed wrong on a characteristic and the individual is not needed in the quota.
An illustration of a quota sample is provided in a Philadelphia bail experiment conducted by
Goldkamp and Gottfredson (1984). First, a sample of judges was selected from Philadelphia
Municipal Court, and then cases from court files were selected according to a stratified quota
sampling design by which cases were chosen on the basis of both seriousness of charge and judge.
A quota sample was also used in a study of criminal victimization among the homeless in
Birmingham, Alabama (Fitzpatrick, LaGory, and Ritchey, 1993). The researchers used a previous
Birmingham Homeless Enumeration and Survey Project to construct sampling parameters and
stratified on geographic site, gender, and race. They then conducted a quota sample survey of 150
homeless adults by randomly selecting persons from each homeless shelter or public site to match
the proportions found in the previous enumeration. After determining the number of respondents
required from the site, interviewers selected subjects on the basis of sex and race. The demographic
characteristics of the sample matched those of previous studies conducted in Texas and Tennessee.
Accidental Samples. Accidental samples are the favorite "person on the street" interviews
where the "researcher" makes little attempt to ensure representativeness of the sample. This is
well illustrated by many television commercials, for example, "Nine out of ten doctors recommend
__________ ." Which nine of ten? An interview of the easiest and most accessible
generally will not yield data from which one could infer to larger populations.
116 Chapter 4 • The Uniform Crime Reports and Sampling
Purposive Samples. Purposive (judgmental) samples, on the other hand, represent the
selection of an appropriate sample based on the researcher's skill, judgment, and needs. This type
of sampling is well used on election nights when the major networks, based on sample precincts,
are able to quite accurately predict the likely outcome, often with a small margin of error with
only 2 percent of the votes cast. Marketing studies often use test areas that possess characteristics
quite similar to those of the nation. Both political campaign planners and market analysts have
made use of focus groups. Organizers of these focus groups bring together purposively selected
volunteers in order to measure reactions to or attitudes about products, candidate speeches, and
the like (Krueger, 1994 and 1997; Morgan, 1993; Stewart and Shamdasani, 1990).
The use of focus groups and mock trials as courtroom tools are burgeoning areas of applied
research. Focus groups involve bringing together a group of purposely selected volunteers in order
to measure reactions to or attitudes concerning products, candidates, defendants, and the like. They
are a means of gathering in-depth, qualitative data. In measuring business products, a focus group
of eight to twelve people may be brought together for a round table discussion of one to two hours.
Such participants are often recruited by telephone, matched on key demographic characteristics,
and offered an incentive to participate. Topics to be discussed are planned in conjunction with the
client or research issue. In addition to videotaping the discussion and observation, often through
two-way mirrors, subjects are also requested to fill out a questionnaire.
The dynamics of focus group interviews lie in the group process in which participants
influence each other, opinions change, and new views emerge. The participants learn from each
other (Krueger, 1997, p. 20). In the final analysis, the true test of focus group results is whether they
work in predicting product sales, future behavior, court outcomes, or whatever the topic under
investigation. Researchers can ask a small group of six to twelve people questions that will stimulate
lively discussion. The focus group members are the experts giving their opinion to the researcher.
Wanting to find out about kids who frequent crack houses, Bowser hung out in an inner-city
neighborhood and, once accepted, invited kids to join a focus group at a local pizza place and asked
them how to go about surveying the subject (Bowser and Sieber, 1993). Such groups can operate
very inexpensively and simply. Most researchers employ a discussion leader plus two assistants
(notetakers). Some essential elements of conducting focus groups include (ibid., pp. 81-82):
status, or people who might be guarded around each other.
The purpose of focus groups of mock jurists is to identify and apply information on characteristics
of potential jurists in order to effect positive trial strategy. What are some themes and
tactics that might best communicate the case to the jury (Moore, 1998)?
Mock trials are more elaborate in simulating a trial in all respects. This involves attorneys
(or actors) presenting both sides of their cases in a simulated courtroom setting which includes
the voir dire (jury selection), preliminary instructions, opening statements, direct and cross
Chapter 4 • The Uniform Crime Reports and Sampling 117
examination of witnesses, presentation of evidence, and closing arguments. Deliberations are
videotaped, and post-verdict discussions take place in order to monitor the decision-making
process and discover which documents, evidence, or arguments were most persuasive. Such
activities can assist in predicting panel reaction to charges, defenses, and possible awards in civil
cases. Juror profiles for particular cases can also be developed, enabling the targeting of certain
demographic, personality, and attitudinal characteristics. In order to explore this topic further, visit
jury simulation consulting firms on the Internet by entering "jury simulations" as the search item
on a Web browser such as Excite, Yahoo, or Infoseek.
Figure 4.4 describes the use of focus groups to successfully guide jury selection in the
notorious Birmingham Church Bombing Trial, which in 2001 won convictions of a former
Ku Klux Klansman responsible for the attack.
Criminal profiling refers to attempts to construct typical characteristics of certain types of
criminals. Holmes (1989) used a purposive sample and talked to offenders, asking them questions
about their crimes, motivations, and crime scenes. This technique is used by researchers-for
example, by those in the FBI's Behavioral Research Unit-for forecasting purposes and to aid in
the investigation of certain types of criminals such as serial murderers. In a criminal profiling of
forty-one convicted serial rapists, 76 percent were found to have been sexually abused as children.
This same profile found that the majority of serial rapes had not been reported to authorities
(Hazelwood and Warren, 1989). Exhibit 4.3 describes the crime profiling process.
Although purposive samples are not probability samples, their usefulness is judged on the
basis of whether they work in predicting future behavior or attitudes of the target population, for
example, voting patterns and consumer behavior.
The following caveat was issued by Sheley and Wright (1993, p. 3) in their study of juvenile
possession of firearms by selecting purposive samples of 835 male serious offenders incarcerated
in six juvenile facilities in four states and 785 male students in ten inner-city high schools near
It should be stressed that these findings are technically not generalizable to other settings
and populations. The four states serving as research sites for this study were not
a probability sample of States. Moreover, to maximize percentages of respondents
In 2001, a Birmingham, Alabama, jury took only two hours to convict a former Klansman in
the 1963 church bombing. Thomas Blanton Jr. was sentenced to life in prison for the murder
of four black girls who died in the September 15, 1963, explosion at the Sixteenth Street
Baptist Church. Using jury consultants, prosecutors organized two focus groups and polled
500 residents of the Birmingham area. With defense lawyers and judges, they devised a
100-question survey that potential jurors completed at the beginning of the trial. Questions
related to attitudes toward interracial dating and the King holiday were used.
Much of the evidence against Blanton was circumstantial. The prosecution claimed that
they entered the evidentiary phase of the trial confident that they had selected a receptive jury.
The prosecution purported that it struck potential jurists due to their attitudes rather than race
and sex per se. The defense argued that the real intent of the prosecution had been to remove
white men, who had been struck based on racial lines. The judge ruled that the prosecution
had provided race-neutral reasons for its selection.
FIGURE 4.4 Research-Guided Jury Selection in the Birmingham Church Bombing Trial.
118 Chapter 4 • The Uniform Crime Reports and Sampling
Crime profiling (also called crime investigative
analysis) has been practiced on various levels in the
social sciences for years. Classic fictional detectives
such as Sherlock Holmes and Charlie Chan tried
their hand as did a host of profilers of figure such as
Hitler, the Boston Strangler, and the Mad Bomber. In
Profiling Violent Crimes (1996, p. 8), Ronald and
Stephen Holmes indicate that the profiler should
have three goals:
A complete profile provides the criminal
justice system with (a) a social and psychological
assessment of the offender, (b) a psychological
evaluation of the suspected offender's belongings,
and (c) suggestions for the most efficient
and effective way for police to go about interviewing
the suspect once he or she is
Douglas, et al. (1992, p. 310) define crime profiling as
involving seven steps:
Evaluation of the criminal act itself
Comprehensive evaluation of the specifics
of the crime scene(s)
Comprehensive analysis of the victim
Evaluation of preliminary police reports
Evaluation of the medical examiner's
Development of the profile with critical
Investigative suggestions predicated on
construction of the profile
Components of the offender profile include (Federal
Bureau of Investigation, 1991):
age, sex, race
level of intelligence
sexual adjustment and perversions
appearance and grooming
location or residence in relation to crime scene
evaluation and analysis of the criminal act
motive for the offense
prior criminal arrest history
sequence of events during the offense
mood of the offender before, during and after
The crime-profiling approach has been found
particularly useful in investigating arsons, bombings,
kidnaps, murders, child molestations, and serial
murders/rapes. In the profiling process, data is
collected and assessed, the situation reconstructed,
hypotheses formulated, profile developed and
tested, and results reported (Douglas and Burgess,
1986, p. 9). Beginning in the late 1970s, the FBI's
profiling program, housed in its Behavioral Sciences
Investigative Unit, has enjoyed a high-media profile
beginning with success in predictions related to the
Wayne Williams serial murder case in Atlanta in the
early 1980s. Based on interviews with over thirty
serial murderers in prison, the profilers predicted he
would be a black male, early- to mid-twenties, and a
police buff (he impersonated a police officer).
Profiling may not be entirely accepted by old
line investigators as illustrated in the case of criminologist
Bill Tafoya. In 1993, then FBI agent Tafoya
was assigned to the Unabomber investigation in San
Francisco. His profile, which was not accepted, was
uncanny and would most likely have led to an earlier
bust of Ted Kaczynski. Contrary to the prevailing
profile, Tafoya indicated the bomber was probably in
his early fifties (Kaczynski was fifty-three when
apprehended). He predicted the bomber would have a
graduate degree, maybe a Ph.D.; a background in the
"hard" sciences, perhaps engineering or math; and
was an antitechnology Luddite (Witkin, 1997).
Violent Crimes. 2nd ed. Thousand Oaks, C.A.: Sage
Classification Manual. New York: Lexington Books, 1992;
Profiles: The Anatomy of Dangerous Persons, Places,
and Situations. Los Angeles, C.A.: Roxbury, 1998.
Chapter 4 • The Uniform Crime Reports and Sampling 119
involved in the behavior of interest, the study purposely focused on serious juvenile
offenders and on students from especially problematic inner-city schools. Therefore,
the six correctional facilities and ten high schools (and by virtue of the voluntary
nature of participation in the study, the respondents in those institutions) serving as
research sites were not probability samples of their respective universes.
Nonetheless, comparison of inmate respondents' profiles with those known
through studies of youth in similar institutions indicates that the present sample was
not dissimilar to samples of State maximum-security wards serving as subjects of
other studies. Moreover, a 1984 study of inner-city high school students' criminal
activity employed data collected from randomly selected high school students from
inner-city, high-crime neighborhoods in four cities and indicated age and race breakdowns
very similar to those found among the student respondents.
A large number of evaluation studies in criminal justice employ purposive sampling as illustrated
by the following. To assess the impact of determinate sentencing on institutional climate
and prison administration, Goodstein et al. (1984) studied three states that had recently implemented
determinate sentencing and purposely chose states that differed in the types of determinate
sentencing enacted. In an age cohort analysis of arrest rates, Greenberg and Larkin (1985)
chose twenty-five large cities for study on the basis of geographic representativeness. The detailed
planning in choosing a purposive sample can be illustrated in Jacob's (1984) study of ten city
governments' responses to crime from 1948 to 1978. The cities were chosen on the basis of fiscal
strength, type of city government, region, quality of urban life, possession of sufficient research
capabilities, accessibility (cooperativeness in the past) to research, availability of prior research,
and program initiativeness (had received federal grants in the past). Pate et al. (1986) studied fear
of crime in Houston and Newark, the former representing a new, growing city with low population
density and the latter a mature, high-population-density city with declining economic resources.
Toborg's (1981) choice of sites in her study of pretrial release practices in nine jurisdictions was
based on very practical reasons, which probably exist in most purposive samples: geographic
diversity, wide range of (release) types, accurate and accessible records, and a willingness of
agencies to cooperate with the study. Agency contacts and cooperation are essential in such studies
because, without such "hospitality," suspicion will very likely undermine the project.
Snowball Sampling. Snowball sampling is a type of strategy employed particularly in
exploratory studies of little-known or hard-to-obtain subjects (Goodman, 1969; Biernacki and
Waldorf, 1981). It basically entails obtaining a first subject and, on the basis of this subject, obtaining
an entrée and introduction to a second subject, then a third, and so forth. Gradually, as many subjects
as practicable are accumulated. Polsky (1967) employed this strategy in studying uncaught criminals,
as did Solomey (1979) in his study of undercover police. Alex (1969) also employed this strategy in
order to study black police officers in New York City. Such a sampling procedure may be the only
means of obtaining data on little-known or secretive subject matter. In order to study "The Social
Organization of Drug Use and Drug Dealing Among Urban Gangs," Fagan (1989) used a snowball
sample of gang member respondents. Initial subjects were recruited through neighborhood agencies,
and gang members who were recruited later were nominated by these first respondents. All participants
received payment for their cooperation in the form of caps, T-shirts, or coupons to music stores.
In contrasting a history of Asian gangs in San Francisco, Toy (1992) interviewed sixty-four
active gang members as well as nine respondents for historical purposes. Utilizing a snowball
sampling technique, subjects were initially recruited through neighborhood social service
120 Chapter 4 • The Uniform Crime Reports and Sampling
agencies and then asked to refer other gang members. Respondents were paid $50 and another
$40 for each successful referral.
In an Australian study, Dance (1991) utilized a snowball sample to study recreational
intravenous drug users. The author asked everyone she knew if they were familiar with any
intravenous drug users. After asking about fifty people, she met Roger, an intravenous drug
user with whom she developed a rapport. After he told his friends that she was trustworthy,
she was ultimately able to interview others.
The selection of the sample and instrument to be used for data collection is always governed
by time, cost, and staff available to collect and analyze the data. Unless selection probabilities can
be estimated, statistical inference to larger populations is hazardous. In the hands of skillful
researchers, however, and for specific research problems, nonprobability samples may be preferred.
There is no simple answer to the question: "What is an appropriate sample size to choose?"
It depends on a number of considerations, and there is no predetermined appropriate sample
size for all conditions. The choice of sample size can depend on the degree of accuracy
required, the funds available, the expected frequency (or rarity) of the characteristic to be
observed, and the anticipated subclassification of the variables. It's important to note that without
a representative sample, sample size becomes irrelevant. A small representative sample
would yield a better estimate of the population than a much larger sample.
The size of the sample is statistically determined by the size of the sampling error to be
tolerated rather than the total size of the population (Kish, 1965). The larger the sample size, the
smaller the sampling error or extent to which the sampling values can be expected to differ from
population values. Depending on available funds, researchers should attempt to obtain as large a
sample as is practical. Statistical tables for determination of sample sizes are available in standard
statistical texts. For instance, for the 95 percent probability that a sample will have less than
a plus or minus 5 percent error in estimating the population, a population of 500 would require a
sample of 217, a population of 1,000 needs a sample of 286, a population of 10,000 requires 370,
and a population over 100,000 must have a sample of roughly 400. Statistical Package for the
Social Sciences (SPSS) has sample power software to determine appropriate sample size (available
on the Internet at www.spss.com/samplepower/.
The sample size also depends on the expected frequency (or rarity) of the characteristic to
be observed in the population. For example, in the discussion of victim surveys in Chapter 6, it
will be indicated that a sample of 60,000 households is used for the National Crime Survey of the
entire U.S. population. A similar number was required for each city surveyed as part of the original
central city surveys. Why are such large samples required in victim surveys when similar
public opinion surveys are sometimes conducted with only a few hundred in the sample? The
reason is whereas nearly everyone has an opinion, victimization for a specific crime may be rare,
thus requiring a large sample in order to obtain a few cases.
Related to the last point is the fact that too small samples may provide too few cases for
analysis once the sample is subclassified. For example, if the study entailed comparison of three
race categories by ten different crime victimizations, some of the subclassifications (e.g., Asians
who have been burglarized) may yield too few cases for statistical analysis. In such cases, larger
samples are required than minimums expressed in statistical tables.
The reader is advised to examine statistical texts on sampling for more technical detail on
this subject (Kish, 1965; Bachman and Paternoster, 1997), as well as to consult journal articles
and examine sample sizes employed in similar studies.
Chapter 4 • The Uniform Crime Reports and Sampling 121
Sampling involves scientifically selecting a microcosm of a larger population to which
one wishes to infer, usually at great savings in time and cost. There are two major types of
sampling: probability samples and nonprobability samples. In probability samples, which
consist of simple random, stratified, cluster, and systematic samples, an EPSEM is employed.
Each type has relative advantages or disadvantages over the others that must be considered prior
to the decision to employ one. Nonprobability samples do not make use of an EPSEM
procedure and thus make hazardous the employment of statistical techniques that assume this.
They also make problematic any generalization to the larger population from which the sample
was drawn. The major types of nonprobability samples are quota, accidental, purposive, and
snowball. Even though these do not employ EPSEM procedures, careful use of nonprobability
samples can be an effective tool in gaining information regarding larger populations. Because
of its increasing use as a data-gathering strategy in criminal justice, survey research is presented
in two chapters in this text. This chapter concentrated on sampling, mail questionnaires, and
questionnaire construction, as well as self-report surveys of crime. Chapter 6 will cover major
Focus groups are purposively selected groups brought together to measure reactions to
some stimuli. Criminal profiling is an attempt to construct typical characteristics of certain types
Criminal-justice data gathering frequently requires
real-world strategies as well as the use of alternative
approaches, such as social surveys, participant
observation, case study/life history methods, and
unobtrusive measures. None of these methods is inherently
superior to the others, although their relative
strengths and weaknesses can be broken down with
respect to quantitative/qualitative strength, greater or
less control over rival factors, control over factors of
internal/external validity, and artificiality/naturalness
(see Figure 4.1). Each of these alternatives to experiments
is briefly discussed as a prelude to detailed
treatment in successive chapters.
The Uniform Crime Reports (UCR) is given
special treatment in this chapter so that it will not be
lost in our discussion of other studies using available
data or official statistics. Until recently, criminology
and criminal justice in the United States
have heavily relied on the UCR for research purposes.
Most popular presentations on crime rates in the
United States are usually taken uncritically from the
UCR without a full appreciation of the limitations
of these data. The UCR is published annually by the
FBI and represents not crimes committed, but
crimes reported to, and recorded by, the police.
In general, the further from the actual offense
commission, the poorer the official statistics are in
providing an accurate picture of crime. The participation
of local police departments in the UCR
reporting system has improved over the years, with
about 98 percent of the national population covered
by 1978. The original Crime Index consists of a
simple summated index of seven (arson, added in
1978, equals eight) crimes considered more serious,
most likely reported, and most frequently occurring.
The violent index offenses are murder and nonnegligent
manslaughter, forcible rape, robbery, and the
property index offenses are aggravated assault,
burglary, larceny-theft, auto theft, and arson. The
crime rate is the number of crimes per unit of population.
Investigators using the UCR for research
purposes should become familiar with its major
limitations which make it a particularly hazardous
statistic for comparing crime over time or measuring
actual crime commission.
In 1985, a blueprint for a redesigned UCR
program was developed, and by the early 1990s, it
had begun to be implemented. This program featured
a National Incident-Based Recording System
(NIBRS), which will eventually replace the traditional
UCR. This comprehensive system provides far
more detail than the previous summary-based one.
122 Chapter 4 • The Uniform Crime Reports and Sampling
Uniform Crime Reports
Index crimes 99
Crime rate 100
Factors affecting UCR 101
Hot spots 104
Limitations of the crime
Crime dip 103
UCR redesign 105
The National Incident-Based
Hierarchy rule 109
Violent crime index 99
Property crime index 99
Sampling frame 110
Probability samples 111
Simple random sample 111
Stratified random sample 112
Cluster sample 114
Systematic sample 114
Random start 114
Multistage sample 115
Quota sample 115
Accidental sample 115
Purposive sample 116
Focus groups 116
Mock trials 116
Criminal profiling 117
1. What is the UCR? What are its major components?
What are the major components of the crime index?
The calculation of crime rate? What have been some
major identified shortcomings of the UCR?
2. Given the identified shortcomings of the UCR, read
and then discuss how features of the redesigned UCR
may eliminate some of these shortcomings.
3. Discuss the National Incident-Based Reporting System
(NIBRS). What are some of its principal features as
well as advantages over the traditional UCR?
4. What are some possible explanations for the crime
dip of the 1990s?
5. Discuss the various types of sampling and when it
would be most appropriate to use each one.
6. For what is weighting used in disproportionate stratified
sampling, and why would samples be disproportionately
drawn in the first place?
Preventing Crime: What Works, What Doesn't, What's
A Social, Psychological, Educational, and Criminological
Cybrary: The World's Criminal Justice Directory
Research Designs (North Carolina State University)
THE END OF CHAPTER 4
128 Chapter 5 • Survey Research
3. ❏ Household member
4. ❏ Someone else
5. ❏ Police on scene
Some Problems: The use of the term "informed" is a potentially suggestive one. The possibility
may be ruled out that the police were on the scene or happened on-the-scene or
may have been called without the respondent being aware.
Question: 8a. What were the injuries you suffered, if any?
1. ❏ None-Skip to 10a.
2. ❏ Raped
3. ❏ Attempted rape
4. ❏ Knife or gunshot wounds
5. ❏ Broken bones or teeth knocked out
6. ❏ Internal injuries, knocked unconscious
7. ❏ Bruises, black eye, cuts, scratches, swelling
8. ❏ Other-Specify__________________________________________
b. Were you injured to the extent that you needed medical attention after the
1. ❏ No-Skip to 10a.
2. ❏ Yes.
c. Did you receive any treatment at a hospital?
1. ❏ No
2. ❏ Emergency room treatment only
3. ❏ Stayed overnight or longer
How many days?
Problems: Question 8b may be subject to varying interpretations. The interviewer training
manual defines "need" as actually securing aid from a trained medical professional.
However, to a respondent, "need" could be based on a conception of the seriousness of
the injury. Because the interviewer may not always provide the official interpretation of
"need," responses may reflect different interpretations of the meaning of the question.
Question 8a responses 2 ("Raped") and 3 ("Attempted rape") were to be interpreted as
determinants of physical injury, but attempted rape may involve only verbal threats.
That is, the inclusion of attempted physical injury may cause distortions in data on
Question 8c, in obtaining data on hospital care only, fails to identify other types
of medical care, professional or nonprofessional, or institutional or otherwise.
Question: 2. About what time did this (most recent) incident happen?
1. ❏ Don't know
2. ❏ During the day (6:00 A.M. to 6 P.M.)
3. ❏ 6:00 P.M. to midnight
4. ❏ Midnight to 6:00 A.M.
Problem: Noncomparable categories of time provide too broad a category for daytime,
which could be subdivided to 6:00 A.M. to noon and noon to 6:00 P.M.
Question: 11b. How old would you say the person (offender) was?
1. ❏ Under 12
2. ❏ 12-14
3. ❏ 15-17
4. ❏ 18-20
5. ❏ 21 or over
6. ❏ Don't know
Problem: Category 5, "21 or over," is too broad and should be subdivided.
necessary to make a meaningful or informed decision. For instance, a researcher could ask, "Do
you support the current interpretation of the Miranda decision?" Perhaps 60 percent of the
respondents may say "yes"; however, not having asked the respondents if they had any idea what
the Miranda decision was, it may turn out that 90 percent of the subjects did not know what it
was and were too embarrassed to say so.
Some additional pointers include that the first several questions should be easy to understand,
important to the study's purpose, and engage the respondent's interest. Questions should be grouped
into logical order and new sections should feature an introduction so that participants can switch
mental gears (Narins, 1995, p. 8). Avoid too many skip questions; for example, "If the answer is
‘yes,' skip to 5; and if ‘none of the above,' skip to 6." This may be confusing to the respondent. If the
survey has multiple pages, a booklet format is recommended. Attempt to make self-administered
surveys easy to complete, with check boxes and lines easy to see and numbers to be circled far
enough apart. Begin the instrument with a short introduction explaining its purpose, topics to be covered,
how the results will be used, and any incentives for participation. Leave plenty of room for
open-ended questions but do not supply lines, as these may constrain any comments (ibid., p. 9).
Although formulation of dummy tables and a variables list and adherence to general points discussed
so far will assist the researcher in the development of a potentially useful instrument, prior to
using the questionnaire with target respondents a pretest of the instrument is a must. A pretest is a
reconnaissance operation or exploratory testing of the instrument using subjects who are similar to
the group to be studied. The pretest subjects are asked to critique the instrument, pointing out confusions
or misunderstandings and perhaps suggesting more proper wording or issues to be explored.
ORGANIZATION OF THE QUESTIONNAIRE
The order of questions may influence the willingness of subjects to respond to the survey. A common
error in surveys is to begin with the demographic items such as age, sex, and race. Although
these questions are an important part of any survey, they are also routine and boring for most respondents.
Such questions are better asked later or even last in the instrument (see Schuman and Presser,
arouse interest and gain the respondent's attention. The beginning of the survey is also not the
appropriate place for sensitive items; one would not ask a person he or she had just met extremely
personal questions. The following illustrates what not to ask in the beginning of questionnaires:
A Guaranteed Low-Response Questionnaire
What is your name?
How old are you?
What is your sex?
130 Chapter 5 • Survey Research
How much money do you earn?
Do you cheat on your income tax?
How is your sex life?
When was the last time you committed a crime?
Perhaps for good measure one could throw in, "Do you still beat your spouse?" and "Do
you walk to work or carry your lunch?"
The questions should be arranged in a logical sequence that is readable, interesting, and
easy to respond to. In mail surveys, open-ended questions should be kept to a minimum (see
Schuman and Presser, 1996).
Most readers of this book are probably already familiar with the properties of mail questionnaires
because participation in such surveys is becoming a common cultural phenomenon in
North America and other developed countries. The most common type of mail survey is the
self-administered, mail-back variety in which a stamped, addressed return envelope is enclosed.
The mail survey is a popular instrument for research because it promises, at a minimum
of time and expense, to deliver fairly wide coverage for a study. Perhaps this asset of the mail
questionnaire has made it the favorite instrument of a variety of organizations selling products,
soliciting opinions, collecting charitable donations, and those attempting to conduct
social, scientific, or criminal justice research. In the 1970s, a special conference of the
American Statistical Association addressed itself to the growing concern of nonresponse in
such surveys. It appeared as if the potential respondents were becoming overburdened. In the
mid-1960s, large private research organizations could expect roughly a 75 percent response
rate in mail surveys; by the mid-1970s, this figure had dropped to 60-65 percent. This is
assuming even a number of follow-up inquiries to solicit participation (American Statistical
In The Phantom Respondent, John Brehm (1993) notes some alarming trends about the
growing nonresponse problem in polls and surveys such as the National Election Studies
(NES), based at the University of Michigan and the General Social Survey (GSS) at the
University of Chicago. The former has been done every federal election year since 1954 and
the latter every year since 1972 (Morin, 1993). While the NES averaged nonresponse of less
than 10 percent in the 1950s, by the 1990s it had the same 20-30 percent nonresponse as the
GSS. The major media polls have 30 to 50 percent nonresponse (ibid.). A study of 141 articles
from leading managerial and behavioral science journals in 1975, 1985, and 1995 discovered
that the average response rate in academic surveys was 55.6 percent, actually 48.4 percent in
1995 (Baruch, 1999). The key question is whether the nonrespondents differ significantly
from respondents, and the answer is yes. Overrepresented in surveys are the elderly, blacks,
women, the poor, and the less educated. Men, young people, whites, and the wealthy are
Those considering using mail surveys as their means of data collection should consider
the fact that they are competitors for the time of respondents who are becoming increasingly
more difficult to interest in participating. A prudent researcher should, prior to deciding to
employ mail surveys, carefully consider the relative advantages and disadvantages as well as
alternative data-gathering strategies that might make it unnecessary to collect new data. The
definitive source on mail survey and related survey research is the journal Public Opinion
Quarterly, a publication of the American Public Opinion Association, an organization that sets
standards in the field and to which most reputable private research organizations belong. One
practice of such organizations, for instance, is a 5 percent verification check on surveyed subjects
to assure accuracy of data; that is, 5 percent of those already questioned are questioned
again to certify their responses.
Although the problems and prospects raised by a particular survey vary from study to
study, a presentation on general disadvantages and advantages of mail surveys may help one
decide whether it is the appropriate data-gathering method for a study.
ADVANTAGES OF MAIL SURVEYS
wide geographical and perhaps more representative samples at reasonable cost, effort, and
no field staff, thus eliminating transportation and other costs. By the same token, it eliminatesinterviewer bias effects
respondents greater privacy as well as an opportunity to think out their responses, leading to
more considered answers. This is particularly the case for a survey attempting to obtain detailed
information that may require checking records, files, historical documents, and the like. For
example, in a survey of presidents of professions related to rehabilitation, this author asked
questions such as the following:
Question: What was the average budget of your organization from 2000 through 2003?
(If it would be easier, you may wish to supply the information yearly.)
Average Expenditures _____________________________________________
Total Assets _____________________________________________________
Net Assets ______________________________________________________
Optional: 2000 2001 2002 2003
Question: On the next page are a series of events that are believed relevant by some writers
in the field to the history of the development of occupations and their professional associations.
Please supply estimates and answers.
EVENT ESTIMATED DATE APPLICABLE?)
1. At what time did work in your field
emerge as a full-time occupation?
2. When was the first training school
3. When was the first state licensing law
in your field established (if any)?
4. When was the first formal professional
"code of ethics" adopted?
132 Chapter 5 • Survey Research
Obviously, such questions are most appropriately asked in a mail survey, which enables
the respondent to devote adequate time to look things up, rather than in an on-the-spot
DISADVANTAGES OF MAIL SURVEYS
The chief problem with many mail surveys is nonresponse. Inexperienced researchers without
sponsorship may be fortunate to obtain a 20 percent rate in first-wave mailings, that is, a
one-time-only survey without follow-up (Miller, 1991, p. 77). Even with fairly high rates of
and nonrespondents with respect to the issue being investigated. Other potential problems
to follow up attempts, the possibility that a number of respondents may misinterpret the
questions, and escalating costs if several follow-ups are required. Although these and other
problems create difficulties, they are by no means insurmountable, as demonstrated later in
this chapter. Still, these disadvantages must be seriously considered by the researcher and
addressed in some fashion by means of planning, prior to the first canvass.
WAYS OF ELIMINATING DISADVANTAGES IN MAIL SURVEYS
An entire arsenal of techniques is at the disposal of the clever researcher to attempt to outmaneuver
many of the problematic elements of the mail survey. The nonresponse problem can
be broken down into two groups: those who have yet to respond and those who refuse to
cooperate in the survey. It is standard practice in research, unlike encyclopedia sales, to honor
a potential subject's right to refuse to participate in a study. So long as this rate is small, less
than 1 percent for instance, it is an expected loss in surveys. Further pleas to the respondent to
participate, such as "We do hope you will reconsider," stretch a delicate boundary and may be
conceived as harassment on the part of the respondent. If high refusal rates are expected, a far
better strategy is oversampling to create a replacement pool. Although this introduces some
potential error in that the replacement subjects may not match the subjects they replace, at
least the study can continue with a filled sampling frame. The researcher must acknowledge
this potential source of error.
Some ways of increasing responses in mail surveys include, but are not limited to, the
Use of attractive format
The use of techniques to increase response in surveys is limited only by the imagination and
perhaps the time and finances of the researcher. Of major importance in most surveys is the
follow-up with respondents. Continued efforts to solicit response may include renewed mailing(
s) of the original questionnaire, mailings of shortened versions of the instrument, postcards,
telephone calls, interviews, telegrams, mailgrams, and their combinations (Heberlein and
Baumgartner, 1978). Some researchers mail a "reminder/thank you" postcard three days after
the initial mailing to encourage response. An interesting procedure sometimes followed is
enclosure of an identifying postcard with an anonymous survey form so that the respondents
can, at the time they return the form, register that they have responded and should not receive
further reminders (Dillman, 1972, 1976). In general, special delivery and certified delivery are
superior to first-class mail which, in return, is superior to second- and third-class mail. Certified
mail can yield a return receipt verifying delivery. If first-class letters are marked "address correction
requested," postal authorities will notify the sender of the filed forwarding address to
which the letter has been redirected.
A common practice for determining proper timing for the follow-up is illustrated by a
hypothetical study (Figure 5.2).
Beginning on June 1, 1,000 residents of Millvale were mailed questionnaires. Returns
began to arrive on June 4. The daily number of returns peaked on June 16. By June 23, the
replies, encouraged by the second mailing (or first follow-up), began to arrive and peaked around
July 15. At that time, a second follow-up was undertaken, the results of which tailed off in late
August, when a third request was mailed to respondents. The third follow-up had little impact on
encouraging more responses and, because the end of the targeted period for data gathering was
nearing, no further follow-up probes were assumed necessary.
Offering remuneration involves offering rewards or incentives to survey participants. It may,
depending on subject and type of respondents, increase response. Some researchers actually
enclose, rather than just promise, payment on the assumption that people will feel guilty about
keeping the money and not answering the survey. A variation of inducement is the offer to share
July Aug. Sept. Oct. Nov. Dec. Jan.
a First mailing
b Second mailing
c Third mailing
d Fourth mailing
FIGURE 5.2 Millvale Victimization Survey.
134 Chapter 5 • Survey Research
a summary or copy of the report with interested respondents. A word of caution-financial
offerings may be out of order with some subjects; for example, a one-dollar offer might insult
wealthy or influential subjects, or an offer of coupons for free chocolate bars would be in poor
taste in a survey dealing with hunger and starvation. In the latter instance, altruistic appeals to
the respondents' concern for science and humanity would be more effective.
Church (1993) analyzed thirty-eight studies on the impact of incentives in mail surveys
and found that prepaid incentives were more effective than promised ones. Other findings were
that prepaid monetary incentives were better than gifts offered with the initial mailing, response
rates increase with larger amounts of money, and promises of money or gifts do not significantly
increase response rates. Singer et al. (1999), in a replication of Church, found that the effects of
incentives were only modest.
An attractive format for the instrument may impress on the respondent the important nature
of the study. Although cost limitations may determine the ultimate appearance of a questionnaire,
a ditto reproduction is less desirable than a mimeograph, which again is inferior to a
lithograph or a good print job. Other possibilities include colored paper and print,
photographs, illustrations, and booklets. Anything that can attract interest in the survey may
Sponsorship and Endorsements
Sponsorship and endorsements are excellent means of enhancing the potential prestige and legitimacy
reputation of the organization sponsoring or conducting the survey, the greater the potential
response. Unattached researchers or students generally can expect poorer response than known
persons or organizations in the field. For students, letters from professors bearing the college
insignia and urging response would be more effective than the student's own cover letter.
Endorsement cover letters from prominent individuals, for example, presidents of national organizations
of which the respondents are members, may increase response. A survey of police officers
may yield better response if the questionnaire is accompanied by letters urging response
from the Chief of Police and the president of the local Fraternal Order of Police.
In a mail survey of the 100 largest police departments in the United States regarding
police undercover practices, Hamilton and Smykla (1994) were able to achieve an 87 percent
response rate. This high response may have been achieved in part due to the fact that the
cover letter came from the New York Attorney General's Office and one of the authors was a
known practitioner in the law enforcement community.
Personalization of survey instruments is the attempt to make less impersonal the appearance of
the survey package or follow-up probes. Because all good mail surveys should contain
stamped, addressed return envelopes, some feel the attachment of colorful commemorative
stamps to the envelopes adds more personalization than bureaucratic and impersonal postage
meters. Better yet, commemorative stamps that deal with topics related to the subject matter of
the survey may call additional attention to the survey. A criminal justice survey featuring a
stamp with Justitia, the blind goddess of justice, would certainly be eye-catching. More
Chapter 5 • Survey Research 135
research needs to be undertaken in this regard, however, for Heberlein and Baumgartner
(1978) found the highest response rates for government-sponsored studies in which a franked
or metered postage was used. This was believed to lend an "official" air to the project.
A handwritten "P.S." on the cover letter urging response has been claimed to increase
response, as does personalization of the cover letter by use of the respondent's name (Dillman
and Frey, 1974). In one survey of presidents of national professional associations, this writer
stumbled upon a gimmick which, although prohibitive in larger surveys, ensured a last-resort
response from a few remaining important respondents. After four unsuccessful follow-up probes
that included requesting professionals from these fields to look them up at national conventions
and urge response, a personal touch worked. A visit to a local museum of art yielded some occasional
cards that featured a colorful reproduction of a famous masterpiece. Through sheer luck,
the campus post office was selling commemorative stamps with famous paintings, one of which
matched the cards. The combined visual impact, along with a handwritten final request for
participation on the card, elicited cooperation (Hagan, 1975).
Shortened format of follow-up instruments may encourage response from those who were previously
hesitant because of the length of the original instrument. Although a "reminder/thank you"
postcard sent a few days after an initial mailing is likely to result in a higher response rate, a lastresort
postcard featuring the minimum essential questions will at least salvage information on
key items, as well as give some reading of how a group that would have been nonrespondents
differs from respondents. In a review of the literature on methods of improving response rates in
mail surveys, Heberlein and Baumgartner (1978) claim that longer survey forms were perceived
as more important than shorter forms, and, all, other things being equal, were associated with
Good timing for survey mailing includes avoiding competitive seasons or other historical events
that may impede response. Vacation periods should be avoided. Household questionnaires
should arrive near the end of the week, whereas business surveys are likely to fare better at the
beginning of the week. Other gimmicks have been developed by imaginative researchers. Again,
The Public Opinion Quarterly is an excellent source of such "trade secrets."
Despite valiant efforts, because of time, cost, and other factors, most surveys do not expect
100 percent response rates. One way to assess the impact of nonresponse bias on results is
to compare characteristics of the survey respondents with known characteristics of the general
population. For instance, even though we may have a 60 percent response rate and no way of
knowing for sure how the 40 percent nonrespondents differ from those cooperating in the survey,
knowledge that the respondents were representative of the population with respect to age, sex,
race, income, and other key characteristics may give us greater confidence in these findings. In a
survey of victims in which only 125 of 450 questionnaires were returned and the prosecutor's
office prohibited follow-up mailings to victims, the researchers (Erez and Tontodonato, 1992)
assessed the representativeness of the sample by comparing respondents with nonrespondents,
finding that the former were more serious cases but were similar on other relevant variables.
In an imaginative combination of research strategies Sigler and Johnson (1986), studying
a sample of the general population of Tuscaloosa, Alabama, first sent a postcard to households
indicating that they had been selected for study. Three days later, Johnson personally
5 Survey Research
Some Guidelines for Questionnaire
Organization of the Questionnaire
Advantages of Mail Surveys
Disadvantages of Mail Surveys
Ways of Eliminating Disadvantages
in Mail Surveys
The Tailored Design Method
Self-Reported Measures of Crime
Some Problems with Self-Report Surveys
Strengths of Self-Report Surveys
Use of Other Data
Use of Other Observers
Use of Polygraph
Use of Lie Scales
Measures of Internal Consistency
Use of Interviews
Advantages of Internet Surveys
Disadvantages of Internet Surveys
Procedures in Internet Surveys
Survey research, an area that is emerging as a strength in criminal justice research, is an excellent
tool for primary data gathering.
124 Chapter 5 • Survey Research
In Indiana, legislative hearings dealing with the death penalty for juveniles ended with the
following statement (Hamm, 1989, p. 224):
We have debated capital punishment for juveniles today and have come to various
conclusions. Yet one thing we know for sure. Never fill out a questionnaire from a
Criminologist Mark Hamm was told this by both the press and other persons associated
with the legislature. He was also told that his research "struck too close to the bone." Beginning
in the fall of 1986, he received survey responses from eighty-five legislators (85 percent of
the General Assembly). He was relatively certain that many of the legislators had filled out the
survey themselves rather than relegating it to their staffs because, he said, (ibid., p. 223):
n my testimony on the juvenile death penalty before the Indiana General Assembly,
a number of legislators indicated a familiarity with the substantive content of the
survey. Indeed the survey became a heated topic of debate during these proceedings.
My testimony-one among some thirty given before the General Assembly-was the
only one terminated by the Legislature. It was cut short on the grounds that it was
inappropriate to discuss statistical reasons why some legislators might favor the
execution of juveniles and why others might not.
The fact that, as in this example, participants in a legislative hearing were threatened by the
results of a survey certainly illustrates the power and potential usefulness of such an instrument.
In this chapter, we examine mail questionnaire and self-report studies whose features
successfully illustrate many of the opportunities and pitfalls of survey methods. Other major
data-gathering approaches, such as interviews, victim surveys, and telephone surveys, will be
the subject of Chapter 6.
Surveys have often been misunderstood by some researchers who have been socialized in the
experimental tradition. Many times, hostility appears between some theoreticians and practitioners
with respect to the strong emphasis placed on the experimental tradition as an ideal in social science
research. Part of this methodological argument may result from a lack of full appreciation of the
nature of survey research methods and their potential as tools in investigating many important questions
facing the criminal justice system. The notion of a survey connotes images of a poll or simple
tally (count of opinion), but survey research has many purposes and can address many scientific
problems beyond a simple count of opinion. Descriptive survey research may use statistical
probability theory to assess sampling error. (Is what is true of the sample true of the population?)
Analytic survey research attempts to explore questions of cause and effect similar to traditional
experimental research. The experimenter utilizes research design before the fact to remove the
effects of rival causal factors, whereas the survey researcher tries to remove these rival factors after
the fact (after the data have been collected) through the use of statistical analysis.
A basic quality of survey research that is at times forgotten and is responsible for much
expressed attitude or claimed behavior and seldom the behavior itself. In Chapter 7, we will detail
potential errors in surveys in which the full importance of this statement will be explained, but at
this stage, acceptance of this point as an article of faith will suffice. Previously, we indicated that
surveys are not just useful for political and consumer polls but are also effective means of addressing
scientific questions and the causality problem. Rather than control for rival causal factors
prior to the fact by means of research design, surveys generally employ quantitative methods and
Chapter 5 • Survey Research 125
statistical procedures post hoc to control for extraneous variables and sources of invalidity,
because natural field settings make the control of sources of invalidity more problematic.
Survey research may employ a variety of data-gathering methods ranging from the administration
of structured questionnaires to captive audiences (such as all sections of Introduction to
American Criminal Justice at a college or university), to mail questionnaires, field interviews,
telephone surveys, and their variations. In this chapter, we concentrate on the mail questionnaire. It is
widely used and offers the possibility to discuss many issues that occur in the other types of survey
approaches. We will first examine some guidelines for questionnaire construction.
SOME GUIDELINES FOR QUESTIONNAIRE CONSTRUCTION
Although it would be foolish for anyone to claim that there is only one way of constructing an
effective questionnaire, a number of procedures have been established by practitioners through
trial and error and custom.
The most crucial and most underestimated step in questionnaire construction involves
clearly formulating the research problem and the data required to speak directly to the research
problem. A common method of specifying the relationship between research issues and data is
creation of a variables list which is keyed to questionnaire items and dummy tables.
A variables list is constructed after the initial rough draft of the questionnaire. The
concepts or variables to be measured are listed along with the numbers of the questions that
purport to measure them. By rigorously reviewing a questionnaire in this manner, duplicative
items, unmeasured concepts, or an emphasis that is undesirable may be discovered.
Dummy tables are preliminary blank tables constructed prior to data gathering that suggest
the type of data needed, as well as the type of data analysis. Figure 5.1 illustrates the use of a
dummy table to call attention to data that will be needed for a two-variable cross-tabulation.
Assuming that such a tabular analysis is planned, the researcher is now alerted to the need to
check the data-gathering instrument to see if the questions for these two variables have been asked
in a manner that would render itself to the type of categorization in the dummy table. The real
utility of a dummy table is realized when the researcher discovers that he or she has failed to ask a
question necessary to the study. Although variables lists and dummy tables may strike those
anxious to get on with a study as ritualism, they act to ensure that the data needed are obtained
before the fact, rather than after the fact when it may be too late. A basic maxim of research is no
more data should be gathered than needed; however, blind application of this dictum is myopic. A
study should be viewed as a research opportunity, one in which the basic needs of the present
research enterprise as well as "riders" (related research questions that may be analyzed once the
main project is completed) are present. For instance, in the course of a federally funded project on
attrition among rehabilitation counselors, data gathered on professionalism can later be analyzed to
address the issue of professional developments in the field (Hagan, Haug, and Sussman, 1975).
Victim of Serious Yes
FIGURE 5.1 A Dummy Table for the Relationship between Victimization and Age.
126 Chapter 5 • Survey Research
Sudman and Bradburn's Asking Questions (1982) is a gold mine of suggestions on questionnaire
wording and construction. They caution you to resist writing questions until you have carefully
thought through the research questions or problems. It is also quite useful to collect and borrow
successful questions from other researchers. This may even enable the comparison of questions
across studies (ibid., p. 14).
The language used in questionnaires must be geared to the target population. If the
study group is a specialized one, for example, forensic pathologists, then the use of occupational
argot and technical language would be preferred. In fact, pretests of the instrument with
members of this occupation may suggest appropriate terminology. On the other hand, if the
general population is to be surveyed, a more common language should be used. If the target
group includes significant non-English-speaking populations, it may be necessary to employ
bilingual strategies such as dual-language instruments. Faculty of foreign language departments
at local colleges often prove to be invaluable consultants in this regard. Similarity of
respondent understanding of language is particularly problematic in crosscultural research.
Anyone with exposure to foreign languages realizes that certain ideas, idioms, and jargon are
not readily translated into another language. The same holds true across cultures with similar
languages. Asking if a person has ever been mugged may puzzle others who may confuse the
term with kissing or being served a drink.
Care must also be taken to identify clearly who should answer the questions, for example,
head of the household or any adult member of the household. This writer was once the subject of
a shopping mall marketing survey and, as part of an apparent quota sample, was asked his opinions
regarding some sample cereal box covers. The covers featured pictures of sports figures and,
after the completion of a fairly long interview, my wife asked me when I last purchased cereal.
One could just imagine supermarkets loaded with cereal boxes picturing hockey players while
consumers purchase those featuring smiling children.
Some of the following suggestions on questionnaire wording and construction are not intended
to be either exhaustive or mutually exclusive:
spouse?" No matter how respondents answer that question, they are admitting spousal
something like, "Do you walk to work or carry your lunch?"
citizen and report this incident to the police?" Respondents would feel unpatriotic if they
support the Miranda decision?"-only to discover too late that half of the respondents
thought Miranda was a shortstop with the Pittsburgh Pirates.
so that a positive or negative answer is the most desirable; otherwise, respondents may
answer the first few and then check off the remainder in a similar fashion without even
Chapter 5 • Survey Research 127
Barton (1958, p. 67) provides humorous examples of many of the techniques that have
been used as a means of asking threatening questions (cited in Sudman and Bradburn, 1982,
pp. 54-55). In this case, we want to ask, "Did you kill your wife?"
to what became of your wife?" (Hand card to respondent.)
1. Natural death
2. I killed her.
3. Other (What?) (Get the card back from respondent before proceeding.)
days. Do you happened to have killed yours?"
a. "Do you know any people who have murdered their wives?"
b. "How about yourself?"
rights to anonymity in respect to their marital relations and that they themselves are to fill
out the answer to the question, seal it in an envelope, and drop it in a box conspicuously
labeled, "Sealed Ballot Box" which is carried by the interviewer.
language, such as that to which the respondent is accustomed, and with an air of assuming
everyone has done everything, "Did you ever kill your wife?" Put the question at the
end of the interview. This "everybody does it approach" developed by Kinsey is called
counterbiasing, which means asking the question in such a way that the behavior appears
relatively frequent and normal.
Although Barton's wife-killing example involves a preposterous topic, the techniques used
to ask threatening questions are quite common and useful in survey research.
Researchers must also decide whether open-ended (unstructured) or closed-ended (structured)
questions will yield the necessary information:
Open: Some people feel that certain parts of the criminal justice system do not work.
Do you agree? If so, what parts?
Closed: Some people feel that certain parts of the criminal justice system do not work. Is this
belief _______ True, _______ False, _______ Don't Know. If true, what parts?
_______ Police, Courts, _______ Corrections, _______ Other Specify ___________________________________________________
Although open-ended questions may provide greater detail and permit respondents to express
their attitudes in-depth, such responses pose difficulty as we will see in our discussion of coding in
Chapter 11. Closed-ended (structured) questions, although they ease the coding process and are
easier for respondents, may not give respondents an opportunity to explain fully their views.
To illustrate some question ambiguity in professionally designed surveys, the following
illustrations were drawn from the presentation on analytical issues in victim surveys by the
Crime Statistics Analysis Staff to the Panel for the Evaluation of Crime Surveys (Panel for the
Evaluation of Crime Surveys, 1976, pp. 167-176):
Question: Were the police informed of this incident in any way?
2. ❏ Don't Know-Skip to Check Item G.
Yes-Who told them?
SOME PROBLEMS WITH SELF-REPORT SURVEYS
Nettler (1978, p. 107) states the matter succinctly by pointing out that "asking people questions
about their behavior is a poor way of observing it." In specifying difficulties in the accuracy of
self-reports on other types of behavior such as voting and medical treatment (LaPiere, 1934;
Deutscher, 1966; Levine, 1976), critics wonder why we should expect respondents to be accurate
and honest in admitting deviant behavior. The major problems with self-report surveys relate
to inaccurate reporting, use of poor or inconsistent instruments, deficient research designs, and
poor choice of settings or subjects (Nettler, 1978, p. 107).
American studies of self-reported crime include items that may not be regarded as
The often small and nonrepresentative nature of some of the samples used may limit their
generalizability. Many samples, for example, had poor representation of blacks. In addition to
possible difficulties presented by the lack of complete anonymity in some surveys, self-report
studies may be subject to lying, poor memory, and telescoping, or the moving of past incidents
into the time frame being studied. Many studies fail to provide a time reference during which the
claimed offenses were to have taken place.
Until recently, self-report surveys have been plagued by lack of replication and overreliance
on one-shot case studies often of atypical populations. Bersoff and Bersoff (2000)
claim that the risks to respondents in self-report surveys of having far more personal information
unprotected is greater than in observational or experimental studies. Such anonymous
surveys requiring neither Institutional Review Board (IRB) review nor informed consent may
invade privacy and cause emotional upset by probing into painful experiences.
STRENGTHS OF SELF-REPORT SURVEYS
Despite the preceding criticisms, an impressive body of research has accumulated that highlights
the strengths of self-report surveys. Both the validity and the reliability of this method and
utilization of the method have steadily improved. Junger-Tas and Marshall (1999) indicate that
the self-report method has improved greatly over the years and that many of its problems and
limitations have been addressed.
In examining the reliability of self-report surveys, Hirschi (1969, p. 56) found only moderate
correlations in admissions of the six kinds of crimes he measured. Clark and Tifft
(1966) found about an 82 percent reliability for subjects who were retested regarding selfadmissions
when they were threatened with a lie detector the second time. Dentler and
Monroe (1961) found a 96 percent concordance between first and second self-reporters, and
Kulik et al. (1968) found a 98 percent agreement with little difference if the questionnaire is
anonymous or signed.
Contrary evidence is provided by Farrington (1973). He had English boys respond to
thirty-eight crimes listed on separate cards and place them into piles as to whether they did
or did not commit these acts. On retest two years later, only 75 percent of the original
crimes were readmitted and half of these were more serious crimes. Gertz and Talarico
(1980) point out important and often overlooked sources of unreliability-clerical carelessness
and coding error.
Chapter 5 • Survey Research 139
140 Chapter 5 • Survey Research
The validity of self-report surveys rests on whether people tell the truth or can accurately recollect
of such surveys are:
Validity checks using official or other data
Checks using other observers (peers)
Use or threat of polygraph
"Known group" approach
Measures of internal consistency
Recheck reports using interviews
Self-report data can sometimes be checked against official police records, school records, and
other sources or criteria. There is a paradox of the critics of official statistics using these same
statistics to validate what is claimed to be a superior self-report instrument. Given our previous
discussion of problems with official statistics, it is unclear what type of overall relationship
would be desired.
statistics. Some of these have been discussed in part in our previous analysis of self-report
surveys. In interviews with boys in Utah, Erickson and Empey (1963) found that a check of court
records indicated that none of the boys lied about having been in court or failed to describe the
offense. Voss (1963) found a strong relationship between admissions and official police records
in his Hawaiian sample. Farrington (1973) also found agreement, concluding that self-reported
delinquents were quite similar to official delinquents. Hirschi (1969) had mixed feelings, with
general underreporting among his sample. McCandless et al. (1972) found even a poorer
matchup between admissions and police records. On the other hand, Hardt and Hardt (1977)
found a strong correspondence between self-reported violations and police statistics; based
on this correspondence as well as other checks they made on their data, they concluded that
many of the conflicting reports in previous surveys may have resulted from the use of inadequate
instruments. Hirschi (1969) also checked other records such as truancy reports and admitted
Checks using other informants, peers, or people who might be able to speak to the respondents'
behavior constitute yet another way of obtaining some validation. Gold (1966) interviewed associates
of the respondent to check whether the person was either told about or observed the acts claimed
by the respondent. Short and Strodtbeck (1965) used confirming reports of detached workers.
The use of, or threat of, polygraph validation was employed by Clark and Tifft (1966). They found
less than 20 percent changed their initial responses when threatened with a "lie detector" test.
In known-group validation (Nye and Short, 1957; Voss, 1963), groups whose official transgressions
are already a matter of record are studied and their self-admissions are compared to this same
behavior. Hardt and Hardt (1977) used, as part of their sample, those who had been previously identified
through official arrest statistics. They concluded that such groups yielded valid responses. Nye
and Short, as well as Voss, found significant differences between "known delinquents" and others.
Nettler points out that there is an essential problem in attempting to validate an instrument with a
criterion, in this case official statistics, which itself is of questionable validity. As a possible explanation,
Hardt and Hardt (1977) found that the majority of boys ranking high on the self-report scale did
not have an official police record. Comparing initial with later responses, however, they found that
most respondents changed something, and most of these changes were in the direction of admitting
more deviance. Although minor offenses like truancy and stealing tended to be underreported, major
offenses like violence and sex offenses were overreported.
Another useful tool for checking the validity of responses is the employment of lie scales or
"truth scales," a series of questions that measure truthfulness of respondents in answering a
survey. Previously, we discussed the tendency of respondents in experiments and surveys to be
agreeable or to give the researcher what they think is desired. "Lie scales" attempt to assess this,
usually by asking the respondents to admit to a type of behavior that-it is assumed-no one
person would have performed or by trying to crossup the respondents by having them give inconsistent
responses. Some other procedures involve having the respondent deny behavior that-it is
reasonably assumed-everyone would perform. Such questions are usually weaved among other
attitudinal questions in the survey. Figure 5.3 gives typical "lie scales."
In scoring "lie scales," researchers set a limit for the number of "incorrect" answers to
questions they are willing to tolerate before questioning the truthfulness of the respondent and
thus calling into question all other responses of that individual. These cases would be dropped
from the analysis. Such procedures are quite commonly used in standard personality inventories
such as the Minnesota Multiphasic Inventory.
Related to "lie scales" is the measurement of the internal consistency of an individual's response by
using interlocking items. This involves the repetition of similar items, sometimes expressing them
first in a positive and then in a negative manner. For example, one might say, "I always tell the truth,"
and then later in the survey say, "I never tell a lie." Hardt and Hardt (1977) in their self-report survey
FIGURE 5.3 Examples of a Lie Scale.
1. I always tell the truth.
2. Sometimes I tell lies.
3. Once in a while I get angry.
4. I never feel sad.
5. Sometimes I do things I am not supposed to do.
6. I have never taken anything that did not belong to me.
7. I have never kept anyone waiting for an appointment.
142 Chapter 5 • Survey Research
asked first whether the respondent had ever been warned or questioned by police and later whether
the subject had ever been arrested or ticketed by the police. It was assumed that a positive response
to the latter would require a positive answer to the former question; if not, the subject's responses
were considered inconsistent. In combination with the "lie scale," this measure of inconsistency was
used as the means of discarding questionable respondents. The use of reversals, stating some of the
questions in a negative manner, is a means of checking response sets. The latter refers to the tendency
of subjects to answer all of the items, often without reading many of them, on the basis of their
answers to the initial questions. That is, if the respondents strongly agreed with the first few items,
then they might simply check off this same response for the remainder. This is also a partial check on
socially desirable response patterns.
Subsequent interviewing of subjects permits probing regarding the details and context of the
acts. For example, claims of the respondent can be questioned, and the criminal intent of the acts
can be established (Hood and Sparks, 1971, p. 68). Primary disadvantages of this approach are
that anonymity of subjects is lost (Gold, 1966) and there may be a tendency for concealment of
offenses depending on the characteristics of the interviewer (Coleman, 1961, pp. 16-17).
Although self-report surveys have their limitations, they do provide another measure of
criminality in addition to official statistics. Particularly when combined with victim surveys, they
offer another means of assessing unreported crime. Chilton (1993, pp. 6-7) even suggests that it
might be time for criminologists to propose the creation of a National Self-Report Survey. Such
a representative survey would most likely face serious political and methodological problems but
would certainly give us a broader picture of crime. Requests for such serious data are almost
certain to be "inconvenient for someone's party position on crime and its causes" (ibid., p. 8).
The use of the Internet to conduct surveys is a relatively new and quickly evolving approach.
Changes are taking place so rapidly that what is presented here is only a brief introduction.
Internet surveys take two different forms: e-mail surveys and Web-based surveys; but the
potential is enormous for camera-based, face-to-face interviews, chat groups, and inexpensive
telephone surveys using the Internet.
The potential advantages of Internet surveys are compelling. If research is limited to specialized
populations (e.g., employees or association members), theoretical access of 100 percent is possible.
Internet surveys can be done faster and cheaper when compared to telephone surveys. In fact, they
are free and many surveys can be undertaken at the same time, which is an advantage over the latter,
particularly with large samples which are ordinarily limited by the number of telephones and
interviewers (Schaefer and Dillman, 1998, p. 379). A mixed-mode strategy of using e-mail when
possible and other methods when not possible has been found to be effective. Web surveys
are much cheaper and more accurate than paper and pencil surveys, and data processing is much
faster and cheaper (ibid.). Results can be published online instantly for viewing by the respondent
as a reward for participating. While Internet surveys may incur little coding or data entry costs
because the data are captured electronically, labor costs for design and programming may be high
(Schonlai, Fricker, and Elliott, 2002).
The disadvantages of Internet surveys include that electronic mail has generally failed to produce
comparable results to mail techniques. Such surveys have sampling problems and are limited to
those with e-mail accounts. Although more than half of all households in the United States have
computers, only 42 percent had Internet access in 2000. The risk of coverage error has limited
most e-mail surveys to specialized populations. In one study, e-mail surveys with only one contact
achieved a 28.5 percent response; but two follow-ups raised this to 57 percent (ibid., p. 380). The
sampling bias in Internet surveys of the population would most predictably undercount females,
minorities, the elderly, and the undereducated. Internet surveys, just as mail surveys, are selfadministered
questions and suffer from the problem of misinterpretation of questions. E-mail
respondents may lack anonymity in their responses, and employers have the potential to monitor
Research using the Internet may raise special ethical problems (Hamilton, 1999). The
rapid growth in such studies may have outpaced the ability of those concerned with ethical standards
to monitor them. Most online research sites do not use safeguards, such as signed consent
forms, to protect respondents. Online researchers seldom design Web sites to send participants
to a debriefing page on completion of the study. The confidentiality of online responses is also
a problem. While the Internet is characterized by confidentiality, the researcher might use a
"cookie" to identify the computer from which the response was submitted (ibid.). Hackers
could intercept responses to sensitive items. IRBs have been seldom consulted on such studies
and are ill-prepared to deal with them if they were consulted. Standardized guidelines for this
purpose are necessary.
Market research has found Web-based research particularly useful and is fueling Web site design
and better means of utilizing this strategy (Krauss, 1998). Web surveys can use color photography,
video clips, and other dramatic enhancements. Some helpful hints in Internet surveys include
(McCullough, 1998, p. 32):
KISS (Keep It Simple, Stupid). Be as simple and straightforward as possible with questions.
Do not ask what they can't answer. For example: "Did your toilet training as a child affect
your sex life as an adult?"
Don't sell. This is another way of saying avoid leading questions.
The next step is to put it on the Web. One might respond, "Sorry, I don't do windows"
(ibid.). It is probably easiest to have someone with Web design experience produce the site. Try
your younger sibling or relative.
Ways of increasing response rates on Web surveys (ibid., p. 33) include:
Banner ads on Web sites
Banner ads on sites frequently visited by one's target audience have been found to be
quite effective. The use of purchased lists for e-mail risks having one's message mistaken for
"spam" (unsolicited junk e-mail). One approach is to identify up-front where you obtained
144 Chapter 5 • Survey Research
their name. Calling people on the telephone and asking if they have Web access and then
inviting them to participate can also work. In fact, multiple methodologies can be employed
at any time. Some marketing companies are now using e-mail panels to do longitudinal
E-mail-based surveys are done by means of an attachment that the respondent can answer
and return via e-mail or download and mail. Web-based surveys utilize a Web site to attract usually
invited respondents to answer questions online. McCullough (1998, p. 30) outlines the steps
in a survey utilizing a Web site:
Understand clearly what questions you want answered.
Know with whom you want to talk.
Write a questionnaire and put it on the Web.
Build traffic to that questionnaire.
Analyze the data.
The writing of specific, measurable, and useful questions involves the same procedure
that we discussed under questionnaire construction. Specialized populations such as orchid
growers can be found by visiting Web sites relevant to their interest. Web surveys are very similar
to traditional surveys but may actually contain more flexibility and control. They can be
programmed to check that answers fall within a specified range and utilize amazing research
E-mail can be personalized by eliminating the names of multiple recipients from the top
of the screen. Avoid using the carbon-copy function or sending a group message. This also
prevents the respondent from accidentally sending their reply message to each of the other
recipients. As Internet access becomes more universal, the potential of Internet surveys will
become more common. It is important that we as criminological and criminal justice
researchers are not left behind. Hybrid techniques have been developed to enhance the representativeness
of Internet surveys. For example, random samples of the national population are
contacted by phone and then hooked up to the Internet. Theoretically, this yields a representative
sample to whom questionnaires can be sent over the Internet and be completed in the
privacy of their homes (Morin, 2000, p. 34). With response rates below 50 percent, there is
reason to question whether random samples are truly random. Imaginative use of quota
samples and volunteer samples has been successful in online voter polling. The popularity of
Internet surveys has led to the creation of for-profit Internet survey firms who will design
surveys for you. One such firm, Surveymonkey.com (www.surveymonkey.com), advertises
that it will do the following:
scales, and open-ended questions.
survey is posted.
Chapter 5 • Survey Research 145
Online interviewing also presents many new opportunities as well as challenges for
researchers (Kivitis, 2005). The popularity and relative recentness of Internet surveys have raised
significant ethical and legal questions with respect to human subjects on the Internet. Some
recent concerns about ethical issues and Internet research have arisen as a result of online
researchers not consistently following guidelines with respect to informed consent of subjects,
possible multiple submission of response, and the danger of computer hackers intercepting
messages (Hamilton, 1999). The American Association for the Advancement of Science
(AAAS) also identified some emergent issues (Frankel and Siang, 1999).They raise the question:
What is private information in cyberspace? AAAS saw a particular need for better instruction of
independent review boards regarding online research.
the World Wide Web (2000) is a very useful primer.
It is most important to remember that surveys, for
the most part, measure respondent attitude and not
Characteristics and qualities of mail surveys
have been described in detail because many of
their features are similar to those of other datagathering
strategies in criminal justice survey
research. The chief disadvantages of the mail
survey include nonresponse, unpredictable uniformity
in response, slow replies, possible misinterpretation
of questions, and costly follow-up. To be
weighed against these are the many advantages of
mail surveys, including wide geographical coverage
with a minimum of time, minimal cost and
effort, nonrequirement of field staff, no interviewer
bias, greater privacy, and the opportunity for
more considered replies.
Some possible ways of reducing nonresponse
in mail surveys are follow-up, payment or
altruistic appeals, attractive format, sponsorship,
endorsements, personalization, shortened format,
and good timing.
Some guidelines exist for questionnaire construction.
First, there should be a clear notion of what
is to be measured and a certainty that the instrument
can address this. The use of variables lists and
dummy tables are intermediary steps. Among the
suggestions discussed were the use of language
appropriate to respondents; clear specification of
respondents; avoidance of biased, leading, or objectionable
questions; and the types of questions to use.
A pretest, or trial run, of the instrument is an absolute
necessity. The questionnaire should begin with
the most interesting questions; biographical items
should appear later or last.
Self-report surveys involve asking, usually
anonymous respondents, to admit to a variety of
offenses they had committed in the past. Keeping in
mind our early injunction that reported behavior
does not necessarily equal actual behavior, these
surveys have been criticized for inaccurate reporting,
poor use of instruments and research designs,
and inadequate settings or study subjects. Despite
these shortcomings, defenders of the technique have
demonstrated accuracy and reliability by employing
known-group comparisons and record checks, as
well as "lie scales" and other methodological
devices. Certainly, self-report surveys present the
criminal justice researcher with another tool with
which to measure crime.
146 Chapter 5 • Survey Research
Analytic survey research 124
Variables list 125
Dummy tables 125
Mail surveys 130
of mail surveys 131
Ways of increasing response in
mail surveys 132
Self-report surveys 136
Problems with self-report
Strengths of self-report
Known-group validation 141
Lie scales 141
1. What are some disadvantages of mail surveys?
Discuss ways of eliminating them.
2. Suppose a student group planning on conducting a
questionnaire survey discovered that you had taken a
research course and asked you for some specific suggestions.
What are some general recommendations
that you would give?
3. What are some problems as well as benefits of selfreport
surveys as a data-gathering strategy?
4. What are some techniques for improving response
rates in surveys?
5. What are some considerations one must take into
account in doing Internet surveys?
The Gallup Organization www.gallup.com
Survey Research Program Links Page (Sam Houston
General Social Survey (Retrievable Data Bases)
National Opinion Research Center (NORC): University
of Chicago www.norc .uchicago.edu/
The Survey System (Creative Research Systems)
THE END OF CHAPTER 5
6 Survey Research
Interviews and Telephone Surveys
Types of Interviews
Advantages of Interviews
Disadvantages of Interviews
Interviewing Aids and Variations
Exhibit 6.1 Public Opinion Polls
General Procedures in Interviews
Training and Orientation Session
Arranging the Interview
Demeanor of Interviewer
Administration of the Structured
Recording the Interview
Vignettes and Scenarios
Advantages and Prospects of Telephone
Disadvantages of Telephone Surveys
Computers in Survey Research
Random Digit Dialing
Techniques Employed in Telephone Surveys
Victim Surveys in Criminal Justice
National Crime Victimization Survey
A Comparison of UCR, NCVS,
and Self-Report Data
Some Problems in Victim Surveys
Cost of Large Samples
Overreporting and Underreporting
Coding Unreliability and Mechanical Error
Problems Measuring Certain Crimes
Benefits of Victim Surveys
A Defense of Victim Surveys
Controlling for Error in Victim Surveys
Reverse Record Checks
Victim Surveys: A Balanced View
Community Crime Victimization Survey
Redesign of the National Crime
Exhibit 6.2 The Redesigned National
Crime Victimization Survey
148 Chapter 6 • Survey Research
Interviewing can refer to a variety of face-to-face situations in which the researcher orally
solicits responses. Berg (2001, p. 57) defines interviews as conversations with a purpose, the
purpose being to gather information. These range from in-depth, lengthy interviews of one or
a few subjects to fairly structured surveys of large groups. As with the other techniques of data
gathering discussed, the advantages and disadvantages of interviewing as a means of obtaining
information should be carefully considered along with other techniques before the decision is
made to proceed.
TYPES OF INTERVIEWS
Researchers use different terms to denote interviews. There are three basic forms:
Although other types of interviews exist, such as the investigative interview used in journalism
(Douglas, 1976) or the preliminary interview employed prior to a larger study, elements
of these are contained in these three principal types.
Structured interviews, sometimes called closed interview schedules, usually consist of
check-off responses to questions that are either factual or to which most responses easily fit an
Question: Compared with last year, what type of job do you feel the local police are
performing in preventing crime in your neighborhood?
______ Much better
______ Somewhat better
______ About the same
______ Somewhat worse
______ Much worse
Question: In which of the following income ranges did your combined family income
fall this past year?
______ Less than $10,000
______ More than $50,000
In structured interviews, the interviewer should avoid soliciting additional comments but,
when they occur, record them verbatim. The principal disadvantage of closed-ended questions is
that they generally elicit only limited response patterns; their advantages are easy administration
and data processing as will be seen later.
Unstructured interviews have many variations depending on the purpose. Sometimes
referred to as focused, clinical, or nondirective interviews, they generally provide for open-ended
Chapter 6 • Survey Research 149
responses to questions. That is, unlike the closed interviews, no predetermined response categories
are provided. To use an example with which all of us have had experience, the structured
interview is comparable to an objective educational test consisting of multiple-choice and trueand-
false items, whereas the unstructured interview is similar to essay tests or tests in which a
person is asked to define or explain the topics being tested. Examples of open-ended response
Question: Do you think the police are better or worse in preventing crime in your local
neighborhood than last year? __________
Question: Why do you feel this way? __________
Question: If you were personally the victim of a crime since January 1 of this year, could
you explain the circumstances surrounding this incident? __________
In comparing the relative advantages and disadvantages of the unstructured questionnaire
vis-à-vis the structured, the previous test question example is a useful touchstone for analogy
purposes. The closed-ended items (e.g., the analogy to true and false) are excellent for recording
simple items in which likely categories of response can be predetermined. Such questions make
codification and tabulation easy but may not provide the depth and quality of response needed.
Open-ended items may present a tabulation nightmare but provide the qualitative detail and
complexity of response that may be required, particularly if the subject of study is little known.
The depth interview is a more intensive and detailed interview, usually of fewer subjects
than is the case in a standard survey, and is particularly useful in life histories or case studies. In a
depth interview, the researcher has a general list of topics to be explored but exercises great
discretion and flexibility in the manner, timing, and direction of questioning. Such interviews are
excellent for hypothesis-generating or exploratory research (Merton, Fiske, and Kendall, 1956).
Smykla (1987), with the approval of a state Department of Corrections, wrote to death row
inmates and later, through further correspondence and visitation, requested from them names,
addresses, and telephone numbers of family members. He was able to arrange interviews with
forty family members and describe what he called their "distorted grief reactions." Classic examples
of the use of depth interviews can be found in many case studies and oral histories such as
Sutherland's The Professional Thief (1937), the result of in-depth interviews with a professional
thief, or Laub's Criminology in the Making (1983), an oral history of American criminology
based on interviews with leading criminologists. Cressey's Other People's Money (1953), which
involved extensive interviews with 133 incarcerated embezzlers, also serves as an example.
Other examples will be discussed in the next chapter.
ADVANTAGES OF INTERVIEWS
researcher and the subject. Such a situation presents many possibilities. Because of the face-to-face
relationship, interviews generally bring about a higher response rate than mail surveys. Being on
the scene, the interviewer can clear up any misunderstandings or confusions the respondent may
have in interpreting questions. Additionally, the interviewer can also act as an observer and not
only record verbal responses, but also make note of his or her own impressions regarding the
respondents and their environment.
Interviews provide an opportunity for the interviewer to make use of cards, charts, and other
audiovisual aids. In asking individuals about their income, for instance, the researcher can hand
150 Chapter 6 • Survey Research
the respondents a card with a list of income ranges and ask them to identify the range within
which household income falls. Having greater flexibility, the interviewer can make return visits if
necessary and pitch the language to the level of the respondent. Perhaps less deliberated responses
reflect the respondents' true attitudes, and the nature of the interviewing process taps this. Unlike
researchers who mail questionnaires, interviewers can determine the actual individual who is
responding and can use their discretion as to the appropriate time at which to ask the more
sensitive questions. With guarded or suspicious respondents, questions regarding income and the
like can be saved for last, so as not to prematurely abort answers to the other questions; or such
questions can be asked at a point in the interview when the subject appears most cooperative.
Interviews are more flexible, may elicit more spontaneous responses, and can utilize more complex
lines of questioning than is often possible in mail surveys.
Feminist researchers challenge what they call "malestream" (mainstream male) approaches
to empirical criminal justice research (McDermott, 1992) for not incorporating feminist views.
One researcher used in-depth interviews as a way of getting at women's experiences that are
obscured in standard surveys (Stanko, 1990).
DISADVANTAGES OF INTERVIEWS
Despite the many advantages of interviews, they possess obvious problems. Some principal
disadvantages of field interviews are that they may be very time consuming and costly.
Although these problems can be offset in part by cluster sampling, covering widely dispersed
households in person can be a problem. Interviewer effect or bias may be responsible for
distorted results. Similarly, the interviewer may make mistakes in asking questions or recording
information. Because of these problems, the use of even a few interviewers requires supervision,
training, and monitoring. In assigning and coordinating field interviews, the supervisor
should be aware of the need for weekend and evening interviews to obtain representative
responses. Interviewing becomes a particularly difficult strategy when attempting to obtain
information from hard-to-reach populations, although by way of trade-offs respondents who do
not own telephones can be reached.
Interviews may be problematic for respondents if factual data that must be looked up are
requested. They are sometimes less convenient to the respondent and afford less anonymity than
mail surveys. Perhaps the chief potential problems rest in the quality, integrity, and skill of the
interviewers, factors that may be uneven in interview surveys.
Question wording in interviews can alter response. Exhibit 6.1 gives some examples from
public opinion polls.
One must weigh the advantages and disadvantages of interviewing and compare them with
those of other data-gathering strategies before deciding on the means of data collection.
INTERVIEWING AIDS AND VARIATIONS
Although most standard interviews are recorded more or less on the spot by the interviewer, using
either an interview schedule, a structured interview protocol, or, in the case of depth interviews,
notes that can be reconstructed into finished form immediately after the interview, a variety of
mechanical aids exist that lend even greater versatility and accuracy. In a small number of important
interviews, videotapes may be used. The American Society of Criminology/Academy of Criminal
Justice Sciences' oral history project conducted videotape interviews of famous criminologists.
Chapter 6 • Survey Research 151
The recording of interviews by means of tape recorders relieves interviewers from the task
of taking on-the-spot notes and enables them to concentrate on conducting the interview.
Dictaphone transcription enables verbatim reconstruction of interviews and, although it produces
an enormous amount of material, presents the researcher with the raw material to digest after the
fact, rather than at the time of the interview. To illustrate that data gathering is limited only by
Public Opinion Polls
Survey research organizations such as Gallup, Roper,
and Yankelovich play a critical role in taking the
pulse of the American public's opinion regarding
public policy. In 1994, 27 percent of the American
public named crime as the most important problem.
Surprisingly, only 3 percent had done so the previous
year. Even though the official crime rate had actually
declined from the previous year, 58 percent of those
surveyed said crime in their community had gotten
worse in the past year and 73 percent said crime in
the country had worsened (Morin, 1994c).
Often, the very wording of survey questions
can produce differing results. In 1992, the American
Jewish Committee was shocked when a poll by the
Roper organization that they had commissioned
discovered that about one-third of those polled felt
it was possible that the Holocaust either never
happened or they were not sure. The actual
question asked was (Morin, 1994a):
As you know, the term Holocaust usually
refers to the killing of millions of Jews in Nazi
death camps during World War II. Does it
seem possible or does it seem impossible to
you that the Nazi extermination of the Jews
In April 1994, Roper simplified the question to read:
Does it seem possible to you that the Nazi
extermination of the Jews never happened?
Only 1 percent said it was possible that it never
happened and only 8 percent were unsure. In thirteen
different polls, the Holocaust doubters varied from
1 to 46 percent. Why? It is because of the question
wording. The high group had questions that were
ambiguous or even contained double negatives, as in
our first example. When questioned specifically as to
whether the Holocaust happened, 98 percent of the
"doubters" changed their position and said it did
In one final example, Richard Morin (1994b),
chief of polling for The Washington Post, points out
how the wording of presidential approval questions
produce skewed results. Most of the media
polls such as by the The Washington Post and ABC
Do you approve or disapprove of the job
that Bill Clinton is doing as President? Is that
strongly or somewhat approve or disapprove?
The Washington Post tested whether a simpler
response scale might produce a different response.
They did a split ballot test in which half of the
respondents were asked the old questions (above)
and half were asked:
Do you strongly approve, somewhat
approve, somewhat disapprove, or strongly
disapprove of the job Bill Clinton is doing as
While the first version produced 58 percent
favorable, 38 percent unfavorable, and 4 percent no
opinion, the second version yielded 62 percent
approving, 32 percent disapproving, and 6 percent
undecided. These illustrations clearly describe the
critical importance of questionnaire construction and
wording in not just measuring attitudes but in some
cases creating them.
Sources: Morin, Richard. "From Confusing Questions,
Edition. 18-24 July 1994a, p. 37; Morin, Richard. "Ask
Weekly Edition. 6-12 December 1994b, p. 37; and
Morin, Richard. "When the Method Becomes the
Message." Washington Post National Weekly Edition.
19-25 December 1994c, p. 33.
152 Chapter 6 • Survey Research
imagination, Albini (1971), in The American Mafia: Genesis of a Legend, reported great success
with mail interviews using cassettes. To cut expenses, time, and travel in interviewing police officials,
organized criminals, and experts on organized crime, he mailed his interview protocol-list
of questions-along with blank cassettes and was able to secure interviews with individuals who
otherwise would not have been interviewed.
Although we will discuss the brief, structured telephone interview survey, the telephone
interview can be expanded under special circumstances by tape recording interviews. An inexpensive
electronic patch can be purchased at any electronics store and easily attached to a telephone
to permit verbatim cassette recording of interviews. Unless one is involved in secretive
measurement, an issue to be discussed in detail in Chapter 8, it is standard procedure to ask,
prior to beginning an interview, the respondent's permission to record. Obviously, problems are
raised with respect to the assurance of anonymity and respondent candor. These problems can
sometimes be circumvented by agreeing beforehand to shut off the recorder for more sensitive
items that may be identified as "off-the-record."
Pictorials, photographs, and motion pictures have all been successfully utilized to enhance
studies involving interviews. For such sensitive items as income, respondents can be handed a
card that contains income ranges and asked to specify the general range. Individuals can be asked
to rank their preferences for various items listed on a card.
response technique (RRT) (Liu and Chow, 1976; Tracy and Fox, 1981). The technique, originally
developed by Warner (1965), basically uses indeterminate questions; that is, the actual
question answered is known only to the respondent and is unknown by the researcher. The interviewer
is blind or unaware of the actual question a respondent is answering. The RRT is rather
complicated to explain, but basically it uses known probabilities in order to estimate unknown
proportions. Neuman (1993, p. 231) provides an example of one variation:
Here is how RRT works. An interviewer gives the respondent two questions. One is
threatening (e.g., "Do you use heroin?"), the other not threatening (e.g., "Were you
born in September?"). A random method (e.g., toss of coin) is used to select the
question to answer. The interviewer does not see which question was chosen but
records the respondent's answer. The researcher uses knowledge about the probability
of the random outcome and the frequency of the nonthreatening behavior to
estimate the frequency of the sensitive behavior.
In Tracy and Fox's (1981) example, 100 married men are brought together in a room and
asked to flip a coin. Next, they are asked to raise their hand if they either get a head on the coin
or if they have abused their wives. If sixty hands are raised, we can assume ten wife abusers
among those with heads on the coin, because fifty heads would be expected by chance.
Additionally, ten of the fifty with tails would also be assumed to be abusers for a total of twenty
estimated abusers. In an actual interview situation, a single respondent would be asked to
respond to "anonymous" sensitive questions in the same manner. Despite some possible shortcomings,
randomized response procedures have been found to yield more accurate results than
direct questioning methods on sensitive items (Fox and Tracy, 1986). A variety of randomization
procedures are employed. For example, Guerts, Andrus, and Reinmuth (1976) posed
questions to subjects in pairs, one sensitive and one innocuous. The question answered is
determined by a coin flip. They used the technique to analyze shoplifting, destruction, price
Chapter 6 • Survey Research 153
altering, and other consumer violations. This technique has also been utilized to study abortion
and fertility control, drug use, child abuse, drunk driving, sexual behavior, illegal gambling,
and shoplifting (Klockars, 1982, p. 454).
GENERAL PROCEDURES IN INTERVIEWS
Much of this discussion on procedures in interviewing is applicable primarily to large, standardized
field surveys; however, most of the specific suggestions have been derived from the experience of
both survey research organizations and individual researchers and is appropriate advice even for
TRAINING AND ORIENTATION SESSION
For interview surveys, an adequate amount of time must be spent on training interviewers. These
training sessions, which may last anywhere from a day to a week depending on the complexity of
the study, should familiarize the interviewers with the organization carrying out the survey as
well as the study's purposes. Details of the project should be provided to make the interviewers
feel that they are an important part of the study and to prepare them to answer any questions
regarding the intent of the survey. Hoinville, Jowell, et al. (1978, p. 117) indicate that an effective
manner of impressing the importance of confidentiality of responses on interviewers is to ask
them to sign a declaration of confidentiality promising not to disclose any information in their
possession. Depending on the sampling plan, either a preliminary letter is mailed or a telephone
call is made to schedule an appointment for the interview.
ARRANGING THE INTERVIEW
Interviewers conducting household surveys should not arrive too early or too late, generally no
earlier than 10:00 A.M. and no later than 8:00 P.M. Surveyors should be furnished with, possess
at all times, and present identification to avoid being taken for door-to-door salespersons.
A prominently displayed name tag featuring the official project name and a picture of the interviewer
is frequently useful. On arrival, be sure that the proper person to be interviewed within
the household is located. At this point, the interviewer should not ask if the respondent wishes
to be interviewed now but rather matter-of-factly indicate that the respondent had received a
letter about the survey and that the interviewer is there to conduct it.
If the interviewer is unsuccessful in scheduling a meeting with the respondent, he or she
should keep written track of callbacks and avoid recalls on the same days or same times.
DEMEANOR OF INTERVIEWER
Advice to interviewers can be as simple as beware of dogs, carry change for telephones in case
of emergency, and dress appropriately for audience and weather (Sanders, 1976, p. 273). Where
possible, the field surveyors should match, as closely as possible, the subjects with respect to age,
sex, race, social class, and dress. Attire should be comfortable, but the interviewer should be neither
overdressed nor underdressed for the occasion. Interviewers should have experienced a few practice
interviews beforehand so that they become familiar with the flow of the instrument to be employed.
The interviewer's language style should also be adapted to the group being studied.
Chapter 6 • Survey Research
In addition to assuring the respondents that their responses will be held in strictest confidence,
the interviewer should attempt to build up rapport with the subjects by being friendly and
diplomatic, as well as convincing, regarding the importance of the study. The interviewer should
attempt to give the impression that the interview will be a pleasant, interesting, and rewarding
experience. Casual conversation can be effective in building rapport with clients. The weather,
children, appearance of the home or grounds, sports, and the like are useful topics. Interviewers
should try to communicate an air of acceptance of respondents' statements but must maintain
their neutrality (Survey Research Center, 1969). In addition to being a sympathetic diplomat, the
interviewer must be prepared to be a persistent boor, a person willing to ask the types of sensitive
questions that generally are considered "nosey" in nature (Converse and Schuman, 1974).
ADMINISTRATION OF THE STRUCTURED INTERVIEW
In structured interviews, it is important for interviewers to become familiar with the flow of the
questionnaire. The instrument should contain good transition statements that in a conversational
style help the respondent to anticipate what comes next. It should be administered in an easy, informal,
and friendly manner to avoid the appearance of an inquisition. For this reason, at no time
should the study be referred to as an investigation. The purpose of a structured questionnaire is to
standardize the manner in which responses are obtained (Survey Research Center, 1969).
Therefore, interviewers should be instructed not to reword or change the questions in any way.
Although procedures may vary depending on the nature and type of interview survey, in
general, questions should be asked in the order listed on the questionnaire. If clarification is
necessary, the interviewer should mark down and list such necessary comments on the questionnaire
If the person being interviewed resists answering sensitive questions, explain that the study
is interested in a group picture of people of different incomes, ages, and backgrounds, and this
information is important to the purposes of the study. In addition, the interviewer must assure the
respondent that the information will be held in strictest confidence and that no individuals will be
identified in the final report.
Often the answer to a question does not provide enough information for the purposes of the
study, and it therefore becomes necessary for the interviewer to probe. Probing involves asking
follow-up question(s) to focus, expand, clarify, or further explain the response given. The interviewer
should be familiar with the responses needed to each question to know when a probe is
necessary. Given the following hypothetical example, it is obvious that a probe is necessary:
Question: Do you think the police in this community are doing an adequate job in
protecting the community?
Answer: Yeah, I guess so.
Probe: What do you mean by that? In what way?
Of possible assistance in the probe is the interviewer's informal mood and responsiveness to
the answers provided by the respondent. The probe should not appear to be a cross-examination
Chapter 6 • Survey Research 155
but should be a natural extension of the interview. Conversation can be stimulated by frequent
"uh-huh's" and "I see's" and by repetition of the respondent's answer while recording it.
Occasional silence, although uncomfortable at times, may encourage more thoughtful and considered
responses. Silence may also indicate, similar to police interrogations, that the interviewer is
not going to accept that response. It is almost like saying "You are going to have to come up with
something better or provide an improved explanation" (Sanders, 1976).
Most beginning interviewers have cold feet and fear hostile respondents. Such respondents
are few; however, a problem can also be raised by the overly friendly respondent. Because the
interviewer has imposed on the respondent's time and good-will, a reasonably friendly socializing
is usually required at the end of the session. What, however, of respondents who, during or after
the interview, account in detail their life history, stories of the big war, or the perils of lumbago? If
this occurs during the interview, some tolerance may be in order to permit the respondent a break
from the demands of the interview schedule (Converse and Schuman, 1974). The digression
can be reoriented by a polite interruption indicating that there is a question directly relating to
that later. Demonstrating inattention such as putting down one's pencil or closing the interview
schedule may also work. If the digression takes place after the interview, a polite excuse such as
another appointment will usually work.
As indicated, once the interview is completed, the interviewer should carry on light conversation
and be alert for any additional comments that the respondent may then offer. Such relevant
remarks should be added to the interview notes as soon as possible after leaving the premises.
Finally, before leaving, the interviewer should thank the subject(s) for their time and hospitality
and should clear up any concerns or doubts the respondent may have regarding the survey.
Informal discussions after the interview can often lead to important "off-the-record" information.
The interviewer can elicit such information by asking in an easy manner, "What do you think?" or
"Is there anything else?" or similar open-ended questions.
Berg (2001, pp. 130-131) describes the ten commandments of interviewing:
1. "Never begin an interview cold." Make small talk and set the stage.
2. "Remember your purpose." Keep the subject on track.
3. "Present a natural front." Be relaxed and natural and not wooden.
4. "Demonstrate aware hearing." Be a real listener and provide appropriate nonverbal
response, such as smiling if the subject makes a joke.
5. "Think about appearance." Dress appropriately for the setting.
6. "Interview in a comfortable place." Meet at a place where the respondent will be afforded
both privacy and safety.
7. "Don't be satisfied with monosyllabic answers." Simple yes or no answers usually call for
further probing questions.
8. "Be respectful." Assure them that you are really interested in what they have to say.
9. "Practice, practice, and practice some more."
10. "Be cordial and appreciative." Do not close the door for future researchers by inappropriate
If you think about it, this constitutes pretty sound advice for a date, job interview, or life
156 Chapter 6 • Survey Research
RECORDING THE INTERVIEW
Although the actual mechanics of data tabulation will be examined in Chapter 12 under Data
Management, a few customary interview procedures bear presentation. Interviewers should be
instructed to write legibly, in pencil, as much of the relevant substance of the interview as possible.
An inexpensive clipboard provides the necessary hard writing surface. The interviewer should
distinguish personal observations from the actual interview by using parentheses. Editing entails
reviewing the interview schedule after completion of the interview and cleaning it up and preparing
it for analysis. The completed interview schedule should be self-explanatory.
The interviewer should have covered each item in the schedule. Unanswered questions
should be marked NA for "not applicable" or simply X to indicate "inappropriate." Where
personal observations are included, it may help the coder (the person charged with assigning
numerical values to the responses) if the interviewer cross-referenced any relevant items. For
example, if an observation has an impact on the understanding of another question, some notation
such as "(see Q. 10 for further explanation)" would be in order.
The interviewer should attempt to record as much as possible during the interview.
Because it is often impossible to record such information verbatim, the jotting down of key
passages for later expansion is helpful. Interviewers should avoid summarizing or paraphrasing
responses but rather try to use the respondent's own words. Paraphrasing requires interpretation
and may change the color and gusto of the real remarks. The interviewer need not ask obvious
questions such as the sex of the respondent. Personal observations can be added at any point they
appear pertinent to an understanding of the response, for example: "(the respondent appeared
very fearful and shaken when relating this incident)". Finally, the end-of-the-interview protocol
should include an opportunity for the interviewer to discuss any other observations that may lead
to a fuller understanding of the context of the interview.
Field interview designers should make use of the face-to-face nature of such encounters to
employ audiovisual and other materials that make the interview more interesting, as well as aid in
the data-gathering process, particularly for overcoming reluctance to answer sensitive questions.
If the research project can afford the luxury of two interviewers per respondent, then
much of the difficulty of recording and conducting an interview can be split, with one interviewer
asking the questions and the other concentrating solely on recording responses.
VIGNETTES AND SCENARIOS
Finch (1993, p.105) describes vignettes as "short stories about hypothetical characteristics in specified
circumstances, to whose situation the interviewee is invited to respond." These vignettes are often
short or typical scenarios (short descriptions of future possibilities) or stories about individuals, situations,
and structures. Target respondents can include individuals or focus groups. In the case of the
latter, vignettes may act as warm-up exercises or icebreakers to get people to interact. The stories are
varied with respect to age, gender, ethnicity, and the like and are useful in exploring sensitive topics.
Decker (2005) points out that interviews with active offenders provide a picture of a different
pattern of offending and perceptions different than those of incarcerated offenders. Such studies
have particularly focused on drug dealers and users, residential burglars, armed robbers, gang
members, and gun offenders. The now defunct Arrestee Drug Abuse Monitoring System used
interviews with offenders to discern their drug use behavior.
Chapter 6 • Survey Research 157
Studies of active residential burglars in Odessa, Texas (Cromwell, Olson, and Avery, 1991),
Delaware (Rengert and Wasilchick, 2000), and St. Louis (Wright and Decker, 1994) documented the
variety of motivations of burglars centered around a lifestyle of partying and keeping up appearances.
Studies of armed robbers documented the robbers' versatility in offending patterns, high
levels of victimization of the robbers themselves, and the pressures of maintaining their lifestyle
(Wright and Decker, 1997). Interviews with gang members and gun offenders, along with the types
of offenders mentioned earlier, were able to note the beginning and end of the offenders' careers, as
well as their responses to various sanctions. The detailed examination of types of criminals overcomes
some of the problems exhibited by developmental criminological studies that use data on
property criminals and infer them to all types of criminals (Hagan, 2008, 2009). Decker (2005, p. 17)
identifies eleven specific procedures or items of concrete advice on conducting offender interviews:
1. Establish the goals of the interview.
2. Choose offenders to interview.
3. Determine who should conduct the interviews.
4. Find appropriate subjects.
5. Convince the subjects to participate.
6. Maintain field relations.
7. Conduct interviews.
8. Sort out the truth.
9. Analyze the interview results.
10. Present the findings.
11. Apply the interview results to tactical and strategic problem-solving.
Most of the projects involving interviews with active offenders have relied on offers of incentives
to participate. This may be in the form of cash (which works best) or vouchers. Maintaining
contacts with subjects may pay future dividends in providing other subjects as well as contacts for
future studies. Most researchers tape-record their interviews and transcribe them at a later date. The
location for the interview should be a place where offenders are not at undue risk. Validation of
interviews is a paramount concern. This may be achieved by repeat interviews, interviews with
peers, and comparisons with other sources of data. Many problem-oriented policing projects have
utilized interviews with active offenders (www.cops.usdoj.gov). These have included interviews
with prostitutes in Lancashire, England (Lancashire Constabulary, 2003), with johns and prostitutes
in Buffalo, and with burglars in Chula Vista, California (Chula Vista Police Department, 2001).
Although the interview ensures a high response rate and possesses many distinct advantages, the
cost, size of staff, and time required often make it prohibitive for many surveys. If use is made of
the widespread ownership of telephones, however, certain advantages of interviews can be
gained without the need for a large field staff and at a fraction of the cost.
ADVANTAGES AND PROSPECTS OF TELEPHONE SURVEYS
The advantages of telephone surveys include not only the elimination of a field staff, but
simpler monitoring of interviewer bias, because the supervisor can be present at the time interviews
are conducted by listening to the interviewers or listening in on calls. Thus, potential bias
or patterns can be caught early and corrected. Although even lengthy interviews can be obtained
through inexpensive electronic patches between telephone and tape recorder, the primary intent
158 Chapter 6 • Survey Research
and quick, generally yield a low nonresponse rate, and provide easy and inexpensive follow-up.
The growing tendency of organizations to obtain flat rate charges that permit fairly unlimited
long-distance calls has made national telephone surveys more of an economic possibility.
Sudman suggests that phone surveys are more effective in obtaining hard-to-locate respondents
than person-to-person interviews. The potential for high refusals can be circumvented,
particularly if short "yes" or "no" answers are used (Sudman, 1980; Glasser and Metzger, 1972).
DISADVANTAGES OF TELEPHONE SURVEYS
Telephone interviews may have difficulty in obtaining in-depth responses or considered answers
over the telephone. There may be some loss of the qualitative detail provided by face-to-face interviews.
In the past, a major objection to telephone surveys has been that they tended to exclude those
who do not own telephones or who have unlisted or new telephone numbers. In some large metropolitan
areas, a considerable proportion of the numbers are unlisted. Also, high mobility in
developed societies may add a large portion of new numbers, and the poor and transient may not own
phones. Although household telephone ownership overall is estimated by the Census Bureau at
approximately 92 percent, it is less than this for African American and minority groups. Ownership
for Hispanics in areas of the Southwest is as low as 65 percent. While telephone surveys can be quick
and inexpensive, if the survey is global in nature, the cost could be prohibitive.
Additional, but not insurmountable, difficulties with telephone surveys include possible
questions. These are initial queries made by the interviewer to determine whether the person
who has answered the telephone fits the target population, for example, income and occupation.
The following describes screening procedures used in a national survey of stalking (Tjaden
and Thoennes, 1998, p. 17):
Because much confusion exists about what it means to be stalked, the National
Violence Against Women Survey did not use the word "stalking" in its screening
questions. Including the word would have assumed that victimized persons knew
how to define stalking and perceived what happened to them as stalking. Instead, the
survey used the following behaviorally specific questions to screen respondents for
Not including bill collectors, telephone solicitors, or other sales people, has anyone, male
or female, ever
Respondents who answered yes to one or more of these questions were asked whether
anyone had ever done any of these things to them on more than one occasion. Because stalking
Chapter 6 • Survey Research 169
A DEFENSE OF VICTIM SURVEYS
Despite the shortcomings of victim surveys that have been elaborated, it should once again be
pointed out that no method of data gathering is perfect. Many of these sources of error are not the
sole province of victim surveys but may apply equally to some of the other techniques of data
gathering. Victim surveys are a relatively young endeavor in criminal justice. Much has already
been learned, and much has yet to be learned in future methodological analyses.
The Department of Justice's Bureau of Justice Statistics (BJS) is constantly monitoring
and attempting to update the methodological accuracy of the NCVS. In 1985, a panel of experts
in criminology-The National Crime Consortium-was charged with the task of devising better
screening questions to obtain better control over forgotten as well as sensitive items. Redesign of
the NCVS reflects this effort.
CONTROLLING FOR ERROR IN VICTIM SURVEYS
Some common means of controlling for error in victim surveys, some of which have already
been mentioned, are:
Use of panels
Bounding of target groups
Evaluations of coding
Reverse record check surveys of known groups
Reinterviews of the same group
Interviews with significant others
The use of panels in the NCVS permits the researcher to achieve bounding. The first interview
with residents results in a panel that can then be followed up five times every six months before
being dropped out of the sample. At each six-month interview, respondents are asked about
events since the last interview. Bounding is made possible beginning with the initial interview
place, can be established. Any events recalled later can be tracked since the previous interview,
thus eliminating telescoping of reports.
Simple coding procedures-the assignment of responses to categories-have high degrees of
measurement error and intercoder discrepancies (Crittenden and Hill, 1971; Sussman and Haug,
1967). The utilization of different coders to classify the same data enables assessment of coding error.
Reverse record checks involve validation of reported behavior on the basis of studying a group
whose behavior is already known. Pilot studies of known victims who had reported their incidents
to police were conducted in Washington, D.C., Baltimore, and San Jose (Panel for the
Evaluation of Crime Surveys, 1976, p. 33). The surveys found that recall deteriorates more
quickly after six months, and particularly forgotten are crimes committed by close acquaintances.
In the San Jose survey, 52 percent of assaults known to police were not reported in victim surveys
(National Advisory Committee, 1976, p. 146). Reiss interviewed a sample of respondents who
170 Chapter 6 • Survey Research
had reported crimes to the police the month before and discovered that one out of five failed to
report to interviewers crimes they had reported to police the previous month (Reiss, 1967).
Reinterview of the same group enables an assessment of potential bias and interviewer
bias. Results obtained by the initial interviewer can be compared with those obtained by the
second interviewer. Such information can then be reconciled or resolved through agreement
(Panel for the Evaluation of Crime Surveys, 1976, p. 63).
Interviews with significant others, peers, teachers, and the like provide a cross-check on
VICTIM SURVEYS: A BALANCED VIEW
In 1974, when LEAA released the findings from its thirteen central city surveys, Donald Santarelli,
then head of the organization, was quoted as having remarked "For the first time in history, we now
have an accurate measure of crime in America-at least in these 13 cities" (Burnham, 1974, pp. 1, 51).
Other popular reviews pointed to how victim surveys, in covering the dark figure of crime, showed
that there was actually twice as much crime as appears in official police statistics. After review of
the advantages and disadvantages of victim surveys, it seems fair to conclude that:
1. For the types of personal and household crimes, both victim surveys and the UCR measure,
the true rate is most likely somewhere between victim surveys, which overestimate crime,
and the UCR, which underestimates crime.
2. For other types of crimes such as occupational, corporate, and public order crime, both
measures underestimate crime.
3. Despite shortcomings, victim surveys present a needed separate and independent assessment
of crime and other criminal justice matters of importance.
The problems encountered with most methods of data gathering are not inherent to the nature
of the method; rather, the problems arise because the method is used as the sole means of assessment.
COMMUNITY CRIME VICTIMIZATION SURVEY SOFTWARE
The BJS has produced Crime Victimization Survey (CVS) software that can be used by local
government agency researchers to conduct their own victimization surveys. It uses the same
question asked in the NCVS. In addition, it contains questions that can measure citizen attitudes
toward crime, their neighborhood, and local policing services. The Justice Survey Software is a
free Web-based software for justice agencies to conduct their own surveys using standardized
questions available from various sources (www.bjsjss.org).
REDESIGN OF THE NATIONAL CRIME VICTIMIZATION SURVEY
In 1974, in response to an evaluation by the National Academy of Sciences and an internal
review by a predecessor of the BJS, a project was begun to evaluate and redesign the NCVS. The
evaluation was put together by a consortium of universities and private research firms and by the
staff from BJS and the Census Bureau. Implementation of some of these redesign plans was
begun in 1986; others were phased in later (Taylor, 1989, p. 1).
The first changes, those which would have minimal impact on NCVS victimization rates,
were introduced in July 1986. Most of these items related to the expanded list of questions, those
that were added to the questionnaire which had remained fundamentally unchanged since 1972.
These new questions related to: XXXXX XXXXX alcohol use by offenders, self-protective measures taken
by victims, police actions, victim contact with the justice system, location of crime, victim's activity,
Chapter 6 • Survey Research 171
and expansion of several existing questions (Whitaker, 1989, p. 2). Among other changes was the
decision to use CATI technology (computer-assisted telephone interviewing). With this CATI technology,
questionnaire items can be flashed on a video monitor, and then interviewers can immediately
enter responses using the keyboard. The redesigned NCVS program is also considering but
has not implemented CAPI technology (computer-assisted programmed interviewing), which
makes possible the use of portable laptop computers in field interviews. Although the proposed
changes involve too many details for our presentation, it is worth noting some other modifications.
These modifications include altering the scope of crimes measured or adding new topical supplements
to the NCVS on a regular basis. Special questions will be periodically added to deal with
timely topics, such as school crime or victim risk (Bureau of Justice Statistics, 1989).
Efforts were made in redesigning the NCVS to avoid disrupting the integrity of its longitudinal
design. The new survey instrument was phased into the NCVS so as not to compromise
trend data (Bachman and Taylor, 1994, p. 502). Exhibit 6.2 details the redesigned NCVS.
The Redesigned National Crime Victimization Survey
Additional Details About the Redesign of
the National Crime Victimization Survey
In the mid 1970s, the National Academy of Sciences
evaluated the NCVS for accuracy and usefulness.
While the survey was found to be an effective
instrument for measuring crime, reviewers identified
aspects of the methodology and scope of the NCVS
that could be improved. They proposed research to
investigate the following:
better stimulate respondents' recall of
concepts of criminal victimization and diminish
the effects of subjective interpretations of
the survey questions
of victimizations that would yield
useful data for analysis
domestic violence, rape, and sexual attack to
get better estimates of these hard-to-measure
The Redesign Has Improved the
Measurement of Domestic Violence
Respondents may be reluctant to report acts of
domestic violence as crimes, particularly if the
offender is present during the interview. In addition,
victims may not perceive domestic violence as
discrete criminal acts but as a pattern of abuse.
Though these issues still pose measurement
problems, the redesigned screening section includes
explicit questions about incidents involving family
members, friends, and acquaintances. Screening
questions also include multiple references to acts of
domestic violence to encourage respondents to
report such incidents even if they do not define
these acts as crimes. The survey staff review these
reported incidents using standardized definitions of
crimes. Thus, within the categories of violent crime
measured by the NCVS, the redesign will produce
fuller reporting of those incidents that involved
intimates or other family members.
A Comparison of the Old and New
Questionnaires Illustrates the Expanded
Cues That Help a Respondent
Recall an Incident
1. People often don't think of incidents committed
by someone they know. Did you have
something stolen from you OR were you
attacked or threatened by-
a. Someone at work or school-
b. A neighbor or friend-
c. A relative or family member-
d. Any other person you've met or known?
172 Chapter 6 • Survey Research
EXHIBIT 6.2 (Continued )
2. Did you call the police to report something
that happened to YOU which you thought
was a crime?
3. Did anything happen to you which you thought
was a crime but did NOT report to the police?
1. Did you call the police to report something
2. Did anything happen to YOU which you
thought was a crime but did NOT report to
The new NCVS has resulted in more victimizations
being reported than when the old instrument was
used. The survey now includes improved questions
and cues that help victims to remember victimizations.
Interviewers now ask more explicit questions about
sexual victimizations. Victim advocates have also been
instrumental in encouraging victims to talk more
openly about these experiences (National Crime
Reasons for Differences in Violent Crime
Rates Because of the New and Old
The new screener questions provide more specific
cues regarding kinds of items used as weapons and
kinds of offender actions that better define the
in-scope crimes of violence for the NCVS. In particular,
the explicit cuing of rape and other sexual
assaults has been added to the new screener. A sideby-
side comparison of the new- and old-screener
questions is provided.
Furthermore, two frames of reference have
been added or more explicitly defined in the new
screener. The first relates to crimes being committed
by someone the respondent knows. The second
relates to the possible location of a crime or activities
the respondent may have been involved in. This
screener question takes the few sporadically
mentioned cues of location/activity in the old
screener questions and creates another specific
frame of reference with a greatly expanded list of
Violent Crime Screener Questions
1. Has anyone attacked or threatened you in
any of these ways-
a. With any weapon, for instance, a gun or
b. With anything like a baseball bat, frying
pan, scissors, or stick-
c. By something thrown, such as a rock or
d. Include any grabbing, punching, or choking,
e. Any rape, attempted rape, or other type of
f. Any face-to-face threats-
g. Any attack or threat or use of force by anyone
at all? Please mention it even if you are
not certain whether it was a crime.
2. Incidents involving forced or unwanted sexual
acts are often difficult to talk about. Have you
been forced or coerced to engage in unwanted
sexual activity by-
a. someone you didn't know before
b. a casual acquaintance OR
c. someone you know well?
1. Did anyone take something directly from you
by using force, such as by a stickup, mugging,
2. Did anyone TRY to rob you by using force or
threatening to harm you?
3. Did anyone beat you up, attack you, or hit
you with something, such as a rock or bottle?
4. Were you knifed, shot at, or attacked with
some other weapon by anyone at all?
5. Did anyone THREATEN to beat you up or
THREATEN you with a knife, gun, or some other
weapon, NOT including telephone threats?
6. Did anyone TRY to attack you in some other
All Types of Crimes Screener Questions
1. Were you attacked or threatened, OR did you
have something stolen from you-
a. At home including the porch or yard-
b. At or near a friend's, relative's, or neighbor's
c. At work or school-
d. In places such as a storage shed or laundry
room, a shopping mall, restaurant, bank,
e. While riding in any vehicle-
f. On the street or in a parking lot-
g. At such places as a party, theater, gym,
picnic area, bowling lanes, or while fishing
h. Did anyone ATTEMPT to attack or attempt
to steal anything belonging to you from
any of these places?
2. People often don't think of incidents committed
something stolen from you, OR were you
3. Did you call the police to report something
4. Did anything happen to you which you
1. Was anything stolen from you while you were
away from home, for instance, at work, in a
theater or restaurant, or while traveling?
3. Did anything happen to YOU which you
Reasons for Differences in Burglary Rates
Because of the New and Old Screener
In general, the same frame of reference is
established for burglary in the new and old
screener. However, the new screener has several
more specific cues. These additional cues relate to
how the offender might have gotten into or
attempted to get into the respondent's home
and/or other types of buildings that may be on the
Burglary Screener Questions
1. Has somebody-
a. Broken in or ATTEMPTED to break into
your home by forcing into door or window,
pushing past someone, jimmying a
lock, cutting a screen, or entering through
an open door or window?
b. Has anyone illegally gotten in or tried to
get into a garage, shed, or storage room?
c. Illegally gotten in or tried to get into a
hotel or motel room or vacation home
where you were staying?
1. Did anyone break into or somehow get into
your home, garage, or another building on
your property illegally?
2. Did you find a door jimmied, a lock forced, or
any other signs of an ATTEMPTED break in?
3. Did anyone take something belonging to you
or any member of this household, from a
friend's or relative's home, a hotel or motel, or
Motor Vehicle Theft Rates
There is no significant difference in motor vehicle
theft rates between the new and old methods. One
reason is that the new- and old-screener questions
are very similar. Another reason is that motor vehicle
174 Chapter 6 • Survey Research
TABLE 1 Changes in Totals Reflect the Headings Under Which Offenses Are Counted
Type of Crime
(New Classification) 1992 Crime Rate
Personal crimes 126.8 Personal crimes 51.1
Crimes of violence 49.3 Crimes of violence 49.3
Rape/other sexual assault 2.9 Rape/other sexual assault 2.9
Robbery 6.2 Robbery 6.2
Completed 4.1 Completed 4.1
Attempted 2.1 Attempted 2.1
Assault 40.2 Assault 40.2
Aggravated 11.1 Aggravated 11.1
Simple 29.1 Simple 29.1
Crimes of theft 77.5 Purse snatching
Household crimes 180.8 pocket picking 1.8
Burglary 58.7 Property crimes 325.3
Household larceny 103.5 Burglary 58.7
Motor vehicle theft 18.6 Motor vehicle theft 18.6
*The theft category is a new crime category. It includes those crimes that were previously classified in two other crime
categories: household larceny and personal larceny without contact (a subcategory of crimes of theft).
Source: Bureau of Justice Statistics, 1994a, p. 10.
thefts are highly salient events (demonstrated by the
fact that they have the highest percent reported to
police), suggesting little room for improvement in
their measurement. Similar results were observed in
the Computer-Assisted Telephone Interviewing (CATI)
research. While CATI increased ratios for most types
of crime, it had no significant effect on motor vehicle
Motor Vehicle Theft Screener Questions
1. Was it-
a. Stolen or used without permission?
b. Did anyone ATTEMPT to steal any vehicles?
1. Did anyone steal, TRY to steal, or use it without
Redesign of Type of Crime
A major reclassification scheme has shifted most of
what were previously categorized as personal crimes
of theft into property crimes of theft. Under the old
scheme, theft was characterized as a personal or
household crime based on location of the incident. If
an item were stolen from the grounds of a home, it
was considered a household theft; if the same item
were stolen from someplace away from the home, it
was considered a personal theft. This distinction was
rather arbitrary and unwieldy, since many items are
jointly owned by members of a household. The
redesigned NCVS classifies all thefts as household
thefts unless there was contact between victim and
offender. Personal thefts with contact (purse snatching
and pocket picking) are now the only type of theft that
are categorized as personal theft. Table 1 compares
the old and new type of crime classification scheme.
Overlap Between the Old and New
As discussed previously, an integral part of the
planned transition from the old methods to the new
methods of kconducting the NCVS was to include a
substantial overlap period in which both methods
were implemented concurrently. Besides being used
for comparing crime estimates, the overlap data can
be used to extend earlier time trends data. Statistical
models will be developed to adjust for the effects of
the new methods on victimization reporting.
Adjustment factors will be estimated at least for the
major crime categories and possibly for other
important variables if reliable differences are found.
Although many of the issues regarding survey
research discussed in Chapter 5 are also applicable to
this chapter, the purpose of this chapter has been to
explore the major elements of the interview, particularly
as it is used in criminal justice research.
Interviewing, which basically involves face-to-face
interaction between the interviewer and the respondent,
has many variations depending on the purpose
of the interview. Principal among these are structured
(closed-ended response), unstructured (open-ended
response), and depth (focus) interviews.
Some general advantages of the interview
method are personal contact, which affords observation,
clarification of misunderstandings and control
over respondents, and the opportunity to employ
visual aids, make return visits, and gear language to
the level of the respondent. Interviews are also more
flexible than mail questionnaires. Disadvantages of
interviews include their sometimes time-consuming
and costly nature, potential interviewer bias and
mistakes, need for field supervision, and difficulty
in reaching certain respondents. Also, question
wording in public opinion polls has been shown to
radically alter response.
Some interview situations may lend themselves
to the use of various electronic recording equipment or
other aids which can be a considerable bonus. The randomized
response technique, in which the interviewer
is blind to the specific item (either a sensitive one or
probability one) being answered, may assist in overcoming
respondent reactivity to sensitive questions.
Some appropriate procedures in conducting
interviews have been detailed. The interviewer
should receive an orientation and training session to
be made aware of the organization and the survey and
to practice interviewing and become familiar with the
instrument to be employed. Arrangement of the
interview, proper protocol, demeanor of interviewers,
and administration of the questionnaire, including
probing and exiting, were detailed. Interviewers
should follow established procedures in recording
and editing their survey schedules.
The use of the telephone survey to assess
victimization holds promise as a means of reducing
the cost of victim surveys. Phone surveys have such
limitations as reduced scope, less in-depth responses,
high refusal rates, and exclusion of disproportions
of certain populations, particularly the poor
and minorities. On the other hand, they have the
advantages of no field staff, simple checks on interview
bias, inexpensiveness, quickness, and easy
Computer software such as CART (Continuous
Audience Response Technology) has greatly expanded
the versatility of the interview. Random digit
dialing enables the coverage of unlisted numbers, a
previous shortcoming of phone surveys of victims.
The use of more careful screening questions and
branching procedures for sensitive items may reduce
nonresponse. Clever procedures have been developed
to ask even attitudinal scale questions by telephone.
Victim surveys have been surprisingly
ignored as a means of measuring crime until relatively
recently. As a result of pilot studies in the
late 1960s, LEAA, in cooperation with the Census
Bureau, began two major types of surveys that
176 Chapter 6 • Survey Research
involved direct questioning of persons as to
whether they had been victims of crime. The first
type, the NCVSs, collected information from both
central city households and commercial establishments.
The second type, called the "National Crime
Panels," consisted of a national stratified multistage
cluster sample of households and a two-stage probability
sample of businesses. The unique characteristics
of the crime panels were bounding of panels
and reinterviews of respondents every six months
until they were rotated out of the sample and
replaced by a new unit.
Victim surveys are not without their problems.
Principal among these are high cost, false reports,
mistaken interpretation of incidents as crimes, memory
failure and decay, sampling bias, over- and/or
underreporting, telescoping, interviewer effects, and
coding and mechanical error. In defense of surveys
of victims, many of their shortcomings are also
present in other techniques, and many of the problems
identified are in part controllable through the
use of panels, bounding, quality control, overcoding,
reverse record checks, studies of known victim
groups, reinterviews, and interviews of persons who
know the victim. As was the case with self-report
surveys, victim surveys provide a valuable additional
assessment of crime.
Beginning in 1986, redesign of the NCVS
began to be implemented. This redesign included
plans to use CATI (computer-assisted telephone
interviewing) and, in the future, possibly CAPI
(computer-assisted programmed interviewing). The
redesigned NCVS improved particularly the measurement
of domestic violence.
Structured interviews 148
Unstructured interviews 148
Depth interviews 149
Interviewer effect 150
technique (RRT) 152
Editing the interview 156
Telephone surveys 157
telephone surveys 157
Screening questions 158
Random digit dialing 160
Branching procedure 161
Dark figure of crime 162
Victim surveys 162
Crime panels 164
Some problems in victim
Demand characteristics 167
Benefits of victim
1. What are some distinct advantages of interviewing as
a data-gathering strategy? Discuss some interviewing
aids which further enhance this technique.
2. Suppose you were the director of a research project
and assigned the task of running a short training program
for the interviewers. What are some specific
points you would present?
3. Compare the NCVS with the UCR as measures of
crime in the United States.
4. What are some methodological problems in victim surveys
as well as some means of controlling for them?
5. What are some techniques employed in telephone
surveys, particularly in those that are designed to
overcome identified shortcomings of telephone
6. What impact does the wording of questions have on
response in surveys, public opinion polls, and victim
7. Discuss some redesign features of the NCVS particularly
as it relates to the measurement of domestic
Crime Victimization Survey Software (Version 1.3)
Telephone Interview Methodology (Washington State
Computer Assisted Qualitative Data Software Resources
Electronic Journals in Qualitative Research
Office of Juvenile Justice and Delinquency Prevention
Guide to BJS Web Site www.ojp.usdoj.gov/bjs
THE END OF CHAPTHER 6
7 Participant Observation
and Case Studies
A Critique of Experiments and Surveys
Verbal Reports versus Behavior
A Defense of Quantitative Research
Types of Participant Observation
Characteristics of Participant Observation
Objectivity in Research
General Procedures in Participant Observation
Other Recording Methods
Tips on Participant Observation
Exhibit 7.1 American Skinheads: The
Criminology and Control of Hate Crime
Announcement of Intentions
Reciprocity and Protection of Identity
Concern for Accuracy
Examples of Participant Observation
Exhibit 7.2 Islands in the Streets
Exhibit 7.3 This Thing of Darkness:
A Participant Observation Study of
Idaho Christian Patriots
Advantages of Participant Observation
Disadvantages of Participant Observation
Life History/Oral History
Some Examples of Case Studies
Exhibit 7.4 Confessions of a Dying Thief
Journalistic Field Studies
Who are Chic Conwell, Doc, Long John, Vince Swaggi, XXXXX XXXXX, and Stanley the
"Jack-Roller"? They are pseudonyms of legendary subjects of social science field
studies and "native guides" to the criminal "turf" as well as to some of the most
fascinating literature in criminology. In this chapter, you will meet Chic Conwell, Sutherland's
The Professional Thief (1937); Doc and Long John, members of Whyte's Street Corner Society
(1943); Vince Swaggi, Klockars' The Professional Fence (1974); XXXXX XXXXX, Chambliss'
safecracker in The Box Man (1975); and Stanley, a mugger in Shaw's The Jack-Roller (1930)
and Snodgrass' The Jack-Roller at Seventy (1982).
A CRITIQUE OF EXPERIMENTS AND SURVEYS
Some researchers feel that social science and criminal justice research have been overdependent
on the artificial elements of questionnaires, interviews, and experimental settings. Such
data-gathering approaches are viewed as creating, as well as measuring, attitudes and bringing
about atypical roles and responses (Webb et al., 1966). Such strategies intrude into a setting as
foreign elements, are limited to cooperative and obtainable populations, and tend to elicit
stooge effects or response sets, particularly with much of the early corrections research.
Whyte (1981) laments the recent ascendancy of quantitative methods and decline of
qualitative field work and says that if a history of the current period of sociology were to be
written, one of the chapters would have to be "Captured by Computer."
Critics of such quantitative research suggest that little relationship exists between attitude
and behavior and that more "sensitizing" strategies involving field studies contain greater
accuracy (Deutscher, 1966; Phillips, 1971). Such writers typically cite a classic study to
illustrate their point of nonconvergence of attitude and behavior-LaPiere's "Restaurant
Study" (LaPiere, 1934). LaPiere traveled with a Chinese couple to a large number of restaurants
on the West Coast and observed the treatment they received. Only one of 251 establishments
refused them service. Later, he sent the same establishments a questionnaire; more
than 90 percent replied that they would deny service. The disparity between what people say
(attitudes) and what people do (deeds) illustrates the hazards of attitudinal measurement of
Although Chapter 9 will discuss in detail the validity of verbal reports, Levine cites
a number of studies in which respondents misreported known behavior that could have been
checked. People were inaccurate in reporting voting behavior, time of vaccination of children,
money in savings accounts, level of loan debt, sexual activity, class attendance, and school
grades. Studies of known crime victims found that a significant number failed to report to interviewers
victimizations that they had already reported to the police (Levine, 1976).
In Chapter 3, it was suggested that error is ever present, even in the best research. Critics
of more quantitative and artificial means of measurement indicate that instead of speaking of
error in measurement, it would be more accurate to speak of the error of measurement
(Deming, 1944; Phillips, 1971). These errors in surveys and experimental studies include
variability in response as a result of the noncomparability of studies and differences caused by
the methodologies employed. For example, telephone surveys may turn up greater victimization
than face-to-face interviews. The degree and kind of canvass will impact on results.
Interviewer bias, if unchecked, may produce error. In doing a study of rural elderly using five
interviewing teams, this writer was startled to discover in a preliminary analysis the high level
of fear of crime in what was assumed to be a pastoral setting. A quick check, however,
revealed that one interview team produced the majority of cases in which crime was a
perceived problem. During our weekly staff meeting, it came to light that one of the interviewers,
an elderly woman, had recently been victimized and unconsciously led her respondents
into seeing crime as a primary problem. Fortunately, this was caught and corrected before it
did major harm to the study results (Hagan, 1972).
Bias of auspices or sponsorship may compromise the results of many studies. For this very
reason, many criminal justice programs bring in outside, objective evaluators to analyze program
180 Chapter 7 • Participant Observation and Case Studies
outcomes. Certainly, a study of the benefits of a product would be more suspect if conducted by
the manufacturer of a product, rather than by an independent laboratory.
Design imperfections, in either the instrument or the analysis, can produce inaccurate
results. Failure to account for nonrespondents may compromise the results of surveys.
Nonrespondents may differ considerably from those who cooperate in a survey. Much survey
research, although planned and designed by professionals, is conducted by "hired hands":
individuals who may have little interest in the accuracy or are unaware of the idiosyncrasies and
subtleties in data. Sussman and Haug (1967) point out that unchecked mechanical errors in
coding and data entry may be more serious in survey data than many assume. Sampling errors
and nonrepresentative samples may lead to error, as might errors in interpretation of findings on
the part of the researchers.
Orne found that subjects in his experiments were willing to put up with boring, uncomfortable,
painful, and ridiculous tasks if asked to do so by the experimenter. In fact, he was
unsuccessful in finding experimental tasks that the subjects would refuse to perform. Most
would yield to any request because "it's an experiment" (Orne, 1974). In one attempt to
assign an obnoxious task, respondents were asked to perform serial additions of rows of
digits and then tear up their answers and start all over again. The subjects so eagerly continued
this dull and senseless task without refusal that the experimenter finally gave up (Orne,
1974, p. 142).
For these and other reasons, some critics feel that findings based on surveys and experiments
are questionable, that much "artificial" research is really measurement error ("error of
measurement"), and therefore more natural methods of data gathering should be employed.
A DEFENSE OF QUANTITATIVE RESEARCH
In defense of quantitative methods, it should be reiterated from Chapter 3 that error is ever
present in all research. Many of these errors are not the exclusive problem of surveys and
research is no research. Also, many errors are not additive in their effect and may cancel
each other, just as many of these potential errors are not inevitable but can be controlled before
the fact, through research design, or after the fact, through statistical analysis.
Chapter 9 will illustrate that no data-gathering methodology alone has any guaranteed
inherent superiority over another. So, although the problems with surveys and experiments may
have been overstated, these criticisms remain the major reasons why some prefer more natural
field methods, ethnographic or qualitative measures, such as participant observation.
Participant observation has long been the favorite tool of anthropologists in studying preliterate
tribes (Bernard, 1994). This is so much the case that it has been suggested in jest that
the typical Navajo family consists of a grandparent, mother, father, three children, and an
Participant observation refers to a variety of strategies in which the researcher studies a
group in its natural setting by observing its activities and, to varying degrees, participating in its
Chapter 7 • Participant Observation and Case Studies 181
A very moving call for field studies in an unpublished statement by Robert Park in the
1920s was recorded by Howard Becker, one of Park's students at the time:
You have been told to go grubbing in the library, thereby accumulating a mass of
notes and a liberal coating of grime. You have been told to choose problems
wherever you can find musty stacks of routine records based on trivial schedules
prepared by tired bureaucrats and filled out by reluctant applicants for aid or
fussy do-gooders or indifferent clerks. This is called "getting your hands dirty in
real research." Those who thus counsel you are wise and honorable; the reasons
they offer are of great value. But one thing more is needful: first hand observation.
Go and sit in the lounges of the luxury hotels and on the doorsteps of the
flophouses; sit on the Gold Coast settees and on the slum shakedowns; sit in
Orchestra Hall and in the Star and Garter Burlesk. In short, gentlemen, go get the
seat of your pants dirty in real research (McKinney, 1966, p. 71).
The Chicago School of Sociology in the 1920s not only got the seat of their pants dirty, but
positively wore them out, along with shoes and pens. Much of the early ethnographic work in
criminology was pioneered by the students of this school. Perhaps with the demise of the funded
"golden age of criminal justice research" (the 1970s), researchers may return to the qualitative
methods and people-oriented Chicago School style research (Reichel, 1985).
The target populations of criminal justice research-the public, victims, criminals, and
criminal justice functionaries-have been subject to a variety of methodological analyses.
Although some fine examples of the use of participant observation exist in the field, it has been
viewed as a neglected and underused technique in criminal justice. Despite this belief and police
suspicion, clandestineness, and resistance to researchers, a disproportionate number of police
studies have utilized participant observation as the major means of data gathering (Manning and
Van Maanen, 1979; Manning, 1972; Sanders, 1977).
Contrary to advice offered by writers of leading criminology textbooks such as Sutherland
and Cressey (1978), Polsky suggests that it is not unwise or impossible to study criminals in their
natural environment. In his book, Hustlers, Beats, and Others, Polsky (1967) describes how he
successfully employed participant observation in studying uncaught organized criminals, pool
hustlers, drug dealers, and con artists. Advocates of participant observation such as Polsky feel
that we have been too dependent on studies of imprisoned criminals in an unnatural environment
or on unquestioned use of official statistics, and that this has led to an inaccurate view of criminals
and criminal behavior.
Previously, the point was made that all research may be viewed as a variation of the
be viewed as the beginning point of all other research. Before one can design a survey or experiment,
one must observe the subject of the investigation sufficiently to know the proper areas to
probe. As we have indicated, participant observation has been most heavily used in anthropology
where it was often the only way of studying preindustrial groups without a written language, by
employing methods other than questionnaires or other standard methods. It represents a commitment
to a more inductive or sensitizing strategy. Weber (1949) referred to such strategies as
illustrative of a Verstehen approach, one in which the researcher purposefully attempts to
understand phenomena from the standpoint of the actors or to gain critical insight through an
understanding of the entire context and frame of reference of the subjects under study.
182 Chapter 7 • Participant Observation and Case Studies
FIGURE 7.1 Types of Participant Observation.
Those of you who have had internship or service learning experiences have a sense of
verstehen. It is difficult to describe the changes that take place and perspectives gained, but
those exposed to the experience gain critical insights that enable a greater appreciation for
others and the setting in which they live or work.
Ethnography, ethnomethodology, and field studies are other labels for techniques similar
to, if not the same as, participant observation (Garfinkel, 1967; Denzin and Lincoln, 1993).
Glaser and Strauss (1967) call for a grounded theory approach, by which a theory is
developed during the data-gathering process, thus grounding it in the real world, rather than
artificially predetermining which hypotheses will be looked at.
TYPES OF PARTICIPANT OBSERVATION
Previously, participant observation was defined in terms of the degree to which participation
and observation may vary. Figure 7.1 suggests how each of the following items may vary
(Gold, 1958): complete participation, participant as observer, observer as participant, and
Complete participation takes place when the researcher not only joins in, but actually
begins to manipulate the direction of, group activity. Such a strategy is rare and tends to violate an
essential element of good participant observation-that the researcher attempt to avoid influencing
the attitudes or behavior of the subjects under study. Another way of viewing complete participation
is to label it "disguised observation," a subject to be treated in detail in Chapter 8.
The most frequently cited example of complete participation in the social science literature is
the case of a group of researchers who joined a small doomsday cult called When Prophecy Fails
(Festinger, Riecken, and Schachter, 1956). Because the group was small, questions were raised as to
the extent to which the researchers brought about the behavior they wished to investigate.
Describing his research strategy as "complete participation," Marquart (1986) worked as a
prison guard for nineteen months while collecting data on prison life. He was able to enter into
more sensitive aspects of guard work, particularly after he established his credibility by successfully
defending himself against an attack by an inmate. Complete participation is very similar to,
if not the same as, disguised observation, a topic which will be discussed in the next chapter.
The participant as observer is the type that most people identify as constituting participant
observation. The researcher usually makes his presence known and, although attempting not to
influence situations, tries to objectively observe the activities of the group. Most of our discussion
in this chapter will relate to this type. The observer as participant describes the one-visit interview.
Even though the interviewers may not think so, they are also short-term participant observers.
Holzman and Pines (1979) employed in-depth interviews of thirty primarily white, middle-class
"johns" and were unable to find support for the "pathology-ridden depictions of the clients of prostitutes."
Money, which involved interviews with 133 incarcerated embezzlers; Klein and Montague's (1977)
Chapter 7 • Participant Observation and Case Studies 183
interviews with imprisoned, retired, and uncaught check forgers; and Letkemann's (1973) study of
forty-five bank robbers and burglars. The last study provided much of what criminologists know
regarding "casing" (looking over) of banks and the bank robber's dependence on uniformity of
bank design, as well as handy parking. Experimental and unobtrusive measures may be viewed as
a type of participant observation that stresses complete observation. Stein (1974), employing oneway
mirrors, was able to secretly observe and record hundreds of sessions between prostitutes and
their clients. In reality, specific studies seldom fall into any one of these categories, or "ideal types,"
and might better be viewed as falling along a continuum, as in Figure 7.1.
Henceforth, our discussion of participant observation will focus on the participantas-
observer type. It should be noted at this point that descriptions of and distinctions between
types of participant observation, field interviews, and unobtrusive measures become somewhat
arbitrary at times and that what one writer might call participant observation another might
view as a type of interview or unobtrusive measure. Some studies are examples of more than
one method; for example, an experimental design may be employed to conduct a simulation
that involves disguised observation in which subjects are interviewed.
CHARACTERISTICS OF PARTICIPANT OBSERVATION
personal cost. Cost here refers not to financial obligations but to personal involvement. As a
general rule, the researcher is committed to studying a group for a considerable period, ranging
from several weeks or months to several years. Participant observers hope to understand the
frame of reference of the group they are studying. This is done by joining the group in their
normal activities to experience things as they do. In doing so, the researcher may temporarily
become a "different person" (Weinberg and Williams, 1972, p. 165).
The observer must attempt to operate mentally on two different levels: becoming an insider
while remaining an outsider. The observer cannot be so far "inside," or socialized into the group,
that everything seems so normal as not to be worth reporting. By the same token, the observer
must be able to report patterns of behavior and interrelationships objectively and without moral
bias. The role of "outsider" can be very valuable in that subjects may be willing to share important
information because she or he is an outsider. Informants may be more willing to open up to
neutral, reliable outsiders (Trice, 1970; Plate, 1975).
The researcher must avoid oversocialization, or "going native." Some become so enamored
by the new lifestyle of the group they are studying that they pass into a new identity and become
too much a part of the group. Polsky, for instance, describes how some of the "uncaught criminals"
he was studying told him that he would make a fine "wheelman" (driver of a getaway car) or
"steer-horse" (accomplice who fingers the score in a con game) (Polsky, 1967). In doing his
participant observation study of vice squads, Skolnick was asked to play a "john" (prostitute's
customer), drive a stakeout van, surveil a bar for suspects, and offer advice on the legality of an
arrest (Skolnick, 1966; National Advisory Committee, 1976, p. 131).
The researcher must avoid not only overidentification with the study group, but also aversion to it.
The ability of social scientists to remain objective despite personal subjective bias is illustrated by
the work of anthropologists. They occasionally find some of the attitudes, beliefs, and behaviors of
the societies they study repugnant and immoral; however, they are trained not to judge, but rather to
184 Chapter 7 • Participant Observation and Case Studies
record the meaning of these behavior patterns to the people who practice them. Famous anthropologist
Pacific (1922), and other fine early studies of the Trobriand islanders, has been pointed to as a
model of objectivity. Social scientists and others were shocked in the late 1960s when
Malinowski's personal posthumous memoirs, A Diary in the Strict Sense of the Word (1967), were
released. Although personally he found the group revolting, this bias was controlled and cannot be
detected in his scholarly writings. This reflects an interesting debate in criminal justice research
wherein Yablonsky, author of such field studies as Synanon (1965a) and The Violent Gang (1962),
criticizes Polsky's view that the field researcher in criminal justice should avoid moralistic stances.
Yablonsky (1965b) feels that such a posture is going too far, whereas Polsky claims that such a
position is necessary to gain a full picture of group activity:
Until the criminologist learns to suspend his personal distaste for the values and
lifestyles of the untamed savages, until he goes out in the field to the cannibals and
headhunters and observes them without trying either to civilize them or turn them
over to colonial officials, he will be only a veranda anthropologist. That is, he will be
only a jailhouse or courthouse sociologist, unable to produce anything like a
genuinely scientific picture of crime (Polsky, 1967, p. 147).
Manning looks at this same issue from the other side of the fence, in "Observing the Police":
Does the observation, if it occurs, of brutality, harassment, incompetence, or malfeasance
obligate the researcher to reveal immediately to the policeman's superiors, or
should he overlook them and pussyfoot in the interest of completing the study? Will
a complete study have an even greater cumulative impact on the organization than
revelation of instances of wrongdoing? (Manning, 1972, p. 258).
On this same general issue, some researchers take the stance that participant observation is
obnoxious manipulation and immoral (Shils, 1961), or "psychological espionage," or necessary
deception in order to obtain needed data (Gans, 1968). Becker perhaps best stated the argument
supportive of Polsky's view of a nonmoralistic stance:
In spite of the romantic yearnings of researchers and the earnest ideological assurance
of some deviants, scientific requirements do not force us to join in deviant actions.
But our scientific purposes often require us to hear about and on occasion to observe
activities we may personally disapprove of. I think it equally indisputable that one
cannot study deviants without foregoing a simple-minded moralism that requires us to
denounce openly any such activity on every occasion. Indeed the researcher should
cultivate a deliberately tolerant attitude, attempting to understand the point of view
from which his subjects undertake the activities he finds distasteful. Amoralism that
forecloses empirical investigation by deciding questions of fact a priori is scientifically
immoral (Becker, 1978a, p. 99).
Soloway and Walters, in discussing ethnographic fieldwork, state the point succinctly: "Approaches
must be found to avoid the dilemmas and pitfalls of a facile and unconstructive ‘hipness' on the one
hand and stagnating righteousness on the other" (Soloway and Walters, 1977, p. 176).
Chapter 7 • Participant Observation and Case Studies 185
Perhaps an even more common outlook resulting from close contact with a new group over
a sustained period is the tendency of observers to over-identify with groups. There are examples
in the literature of an anthropologist who married a cannibal chief and of other individuals who,
native" is a situation in which the researcher identifies with and becomes a member of the study
group and in the process abandons his or her role as an objective researcher. Toby (1986, p. 2)
attacked criminologist John Irwin for a speech he gave before the American Society of
Criminology, accusing him of romanticizing criminals:
Irwin talks about prisoners as though all of them are victims of an oppressive society.
And, in an aside to emphasize his point, he alluded to his personal history. (As is
well known in the profession, Irwin served a prison term before becoming a criminologist.)
Of course, John Irwin is not alone among criminologists in romanticizing
criminals, in seeing virtue rather than moral flaws in offenders. And I can well
understand that a person who has himself served time in prison is aware of decent
people, who, through adverse circumstances, committed crimes, were convicted
and were sentenced to incarceration. I can even understand criminologists, who,
like Edwin Sutherland, get to know and become quite attached to professional
criminals. However, loving the man and hating the fault is quite different from
denying the existence of the fault because criminals are human beings. . . .
I think of criminology as a discipline. By "discipline" I mean more than
subject matter. I mean that we ought to restrain impulses, including benign impulses,
that prevent us from seeing the world realistically. Just as anthropologists cannot be
trusted (intellectually) when they go native to the extent that they glorify rather than
study their preliterate societies, so a criminologist who has gone native cannot be
trusted to tell us what criminals are like.
In 1999, a researcher at John Jay College of Criminal Justice was removed from a Health
and Human Services-sponsored research project and was accused of using as well as buying
heroin for his research subjects. Allegedly he gave some junkies heroin as an incentive for granting
interviews (Schneider, 1999).
GENERAL PROCEDURES IN PARTICIPANT OBSERVATION
There is no one and only method of participant observation. There exists continual debate among
field researchers themselves regarding specific operations; however, a core of customary procedures
have accumulated with which researchers would not radically disagree, although they may
on specific points.
McLuhan (1989, p. 167) whimsically stated that "the last person to ask how the water is
a group in which one has been a lifelong member. First, the researcher is too far immersed in
the culture to maintain objectivity. Second, the members of the group get to know the
researcher too well and may often be unwilling to treat him or her as a researcher. The old
speaker's rule of thumb holds that one must travel at least 100 miles from home grounds to be
regarded as an expert.
186 Chapter 7 • Participant Observation and Case Studies
field notes or diaries. Webb et al. (1966, p. 196) state that "the palest ink is clearer than the best
memory." Initial participant observation is often exploratory, and presumably, the researcher
does not fully understand the culture of the group. Thus, it is necessary to take complete notes
on as many details as possible, even those that appear trivial, because it may be these very
"unimportant details" that later provide the key to some important facet of the study. A good
investigator records observations as often as possible and does not rely upon memory. As much
importance and time should be allotted to recording observations as to participating, observing,
and gathering such information. The participant observer, unlike the vacationer, cannot tell the
reader that "you just had to be there to appreciate what I am telling you."
Berg claims that there is a 4:1 ratio of field-note writing to time in the field (Berg, 2007,
p. 156). Novice field researchers in particular may wish to limit their initial time spent in
actual fieldwork in order to assure that they will have the time and energy to put together
quality field notes.
"Word crunching" (Dennis, 1984), the use of computers in qualitative analysis, has increased
tremendously. Software programs such as Ethnograph (Tesch, 1990, 1991) are quite popular, as is
the adaptation of standard software packages such as Lotus 1-2-3, Gofer, HyperCard, WordStar,
and WordPerfect (Berg, 2007, p. 32). Word searches in WordPerfect, for example, can be utilized
for content analysis purposes (Fielding and Lee, 1991).
Often it is unwise or impossible to record notes on the spot. It may, however, be crucial that
one be able to reconstruct in exacting detail much of what has taken place. Skilled
researchers train themselves in an ancient art used by preliterate tribes to pass down word for
word their traditions. Most are familiar with the closing chapter of Alex Haley's Roots (1976)
in which Haley, after asking the tribal elder whether a Kunta Kinte ever lived in this village,
received no direct answer. Instead he had to listen hour upon hour to an oral recitation of the
chronological history of the tribe until the wise one mentioned the sought-for information.
Using mnemonics, a system of memorizing, the elder was merely using his mind as a vast
computer memory bank and effortlessly providing an extensive oral printout of information.
We have all used mnemonic devices. In elementary school, "ROYGBIV" for the color spectrum,
"every good boy does fine," and "FACE" for the musical scale, and poems to remember
spelling or calendars, all provide useful tools for remembering. Anagrams (words or names),
first letters of important lists, caricatures, and mental associations all permit later recall,
reconstruction, and recording of important information.
The recording of detailed notes on a dictaphone and their later transcription can be effective,
although this method may prevent the researcher from thinking through the material and may
yield huge transcripts that must be boiled down.
Sanders (1977), in a participant observation study of detectives, quickly realized that taking
notes made his subjects nervous. He cut down on this practice and began to leave the notebook
open in the detectives' offices when he left each evening to show he had nothing to hide. Later he
ceased taking notes on the spot altogether and used the time for relaxed observation to improve
Chapter 7 • Participant Observation and Case Studies 187
rapport. Sanders also took photographs to improve recall. These acted as a second type of field
notes (Sanders, 1977, p. 200). Certainly, in the proper circumstances, tape recorders, videotapes,
films, and other mechanical aids can greatly improve recall. Nevertheless, such devices must be
used with caution or perhaps not at all if the subject matter is criminal activity, although they
have successfully been employed with this group also (Klockars, 1977, p. 210; Sherizen, 1976).
The use of photographs in field studies has a long tradition, particularly in anthropology
(Hurworth, 2003; Emmison and Smith, 2000). Classic anthropologist Franz Boas used the technique
in his study of the Trobriand Islanders. He used "photo-elicitation," in which photographs
were shown to the subjects in order to get them to talk about specific rituals. There are a variety of
other techniques that can be used. In "autodriving," the respondents themselves are asked to take
photographs and to comment on them. "Reflexive photography" involves giving respondents
cameras and asking them to take pictures that are then explored in subsequent interviews. "Photo
novella" (picture stories) are another form of photo interviewing in which taking photos gives
people a photo voice by later discussing particular needs or problems illustrated by these photos.
Such techniques encourage more detailed interviews (Hurworth, 2003; Rose, 2001). Cecil Greek
(2005) describes visual criminology as using photography as an ethnographic research method in
criminal justice settings. Photographs can also be used for news media (crime photojournalism
and war crimes photography) and for the collection of evidentiary forensics and other legal material.
He views visual criminology as a useful supplement and alternative to the dominant research
paradigm of combining mathematics and text in academic criminology.
Whether the researcher is aware of it or not, a newcomer to a group is assigned a role by
some time before the newcomer is accepted to the degree that the group becomes less suspicious
and begins to act more naturally. Usually, the smaller the group, the greater the potential disturbance
introduced by the researcher; however, the longer the participant is present, the less the
disturbance over time because he or she is eventually accepted by the respondents.
TIPS ON PARTICIPANT OBSERVATION
In a classic statement on participant observation in criminology, Polsky (1967) offers some tips
on studying criminals in their natural environment. He suggests that one should keep in mind that
the subject is in greater jeopardy as a result of being studied in the field than is someone in jail.
The researcher is more of an intruder and the subjects are certainly freer not to cooperate.
In studying criminals on their "turf," researchers should avoid taking notes on the spot and using
spend their time observing and listening and avoid asking a lot of questions (Spradley, 1970).
The researcher's middle-class language styles and probing may prove irritating to subjects.
William Foote Whyte, in Streetcorner Society, was never able to do an entirely successful analysis
of the rackets in Cornerville because he blew an opportunity with a racketeer to whom he had
been introduced (Whyte, 1943):
One has to learn when to question and when not to question as well as what questions to ask.
I learned this lesson one night in the early months when I was with Doc in Chichi's
gambling joint. A man from another part of the city was regaling us with a tale of the
188 Chapter 7 • Participant Observation and Case Studies
organization of gambling activity. I had been told that he had once been a very big
gambling operator, and he talked knowingly about many interesting matters. He did
most of the talking, but the others asked questions and threw in comments, so at
length I began to feel that I must say something in order to be part of the group.
I said: "I suppose the cops were all paid off?"
The gambler's jaw dropped. He glared at me. Then he denied vehemently that
any policemen had been paid off and immediately switched the conversation to
another subject. For the rest of the evening I felt very uncomfortable.
The next day Doc explained the lesson of the previous evening. "Go easy on
that ‘who,' ‘what,' ‘why,' ‘when,' ‘where' stuff, Bill. You ask those questions, and people
will clam up on you. If people accept you, you can just hang around and you'll
learn the answers in the long run without even having to ask the questions" (Whyte,
1943, p. 303).
These suggestions need not be limited to studying deviants, but apply to much field
research in general. As an American Fulbright scholar in Uruguay, Smykla (1989, p. 29)
described how his initial involvement entailed "inoffensive social interaction." This wins the
subjects' confidence, identifies important contacts and who has the most prestige and insight,
and permits the researcher to learn the words and symbols that will elicit response without
forcing the researcher's own agenda and preconceptions.
Polsky advises that one learn the "argot" (specialized jargon) of the group being studied but
avoid overusing it or trying too hard to be an insider. Initial introductions to criminals in the field
may be gained by frequenting their haunts or sharing other common recreational interests.
Becker (1963) recommends cabdrivers, reporters, bartenders, and cops as good sources of information
on deviant hangouts, although his avenue to studying drug users was his performance in
a jazz band. Polsky (1967) was very adept at playing pool, whereas Bryan (1965) was a counselor
and gained access to other prostitutes through one of his clients who was in the trade. Some
criminals spend a lot of time in court. Talese was successful in establishing contact with Bill
Bonanno during a trial recess. Initially, he simply indicated that "Someday, months or years from
now, I would like to sit down with him and discuss the possibility of writing a book about his
boyhood" (Talese, 1971, p. 13). After many months Talese began to establish contact with young
Bill and his family, although at first the family was suspicious and skeptical:
Nor did I question them: I was sensitive to the situation and at this juncture I was
more interested in the domestic atmosphere and the style of the people than in
specific information. I was content to observe, pleased to be accepted. At night, after
I returned home I recorded (Talese, 1971, p. 501).
Becker (1978a) suggests various strategies for studying deviants. If previous status
provides access to deviant groups, it should be taken advantage of. Ianni and Ianni were able to
gain cooperation on the basis of mutual ethnic identity (Ianni and Ianni, 1972). Tewksbury and
Gagne (2006, p.56) suggest the following roles to gain entry: knowledgeable insider, potential
participant, marginal member, emphatic outsider, and knowledgeable outsider working with a
Chapter 7 • Participant Observation and Case Studies 189
If access is totally lacking, begin with incarcerated individuals. If the type of deviance being
studied is prevalent, subjects may be obtained from larger samples of the population. Steffensmeier
(1986, p. 1) describes how he first decided to undertake his study of The Fence, "Sam Goodman":
I first met Sam Goodman, a white male, nearly sixty years of age, through the
recommendation of several burglars I had been interviewing as part of a research
project on the topic of female criminality. "Talk to Sam," they advised, "He's an
‘old head,' knows his way around if anybody does." I did interview Sam-in
January of 1980 in the Midstate Penitentiary where he was serving a three year term
for stolen property. During this interview, in questioning Sam about the types of
crimes women commit and the criminal roles they play, I found my interest shifting
to questions about Sam himself, his life and his colorful criminal career. Since then,
I have regularly interviewed and studied Sam, even after his release from prison in
the summer of 1981 and on into the present time. [In 1994, Sam passed away, but
not before contacting Steffensmeier and having him visit.]
Advertising for subjects in periodicals geared to their interests has been successfully
employed (Lee, 1969; Hamm, 1993). Finally, the offering of services that deviants need or
cannot readily obtain elsewhere may induce them to reveal themselves. Bryan, for instance,
offered counseling services to prostitutes (Bryan, 1965; Becker, 1978a; Johnson, 1990).
Cromwell, Olson, and Avary (1991) gave a stipend of fifty dollars for each active burglar referred
to their study of thirty active (uncaught) burglars. In addition, confidentiality and anonymity
were promised in encouraging respondents to reconstruct their past burglaries as accurately as
possible. Such insider accounts are very useful in acquainting researchers with the perspectives
of the subjects (Cromwell, 1996). Exhibit 7.1 describes gaining access in a study of skinheads.
In 2008, the Pew Center on the States reported that more than one in 100 American adults
was incarcerated, the largest per capita rate in the world. This fact makes research on what is
taking place in such settings even more imperative. Scholars who undertake ethnographic research
in prisons complain of growing hurdles to doing such investigations. These include suspicious
wardens, state approval boards, and institutional review boards that can greatly slow down the
Gendered Organization, suggests that scholarly research be proposed that can help wardens with
practical problems in their prisons. Trulson, Marquart, and Mullings (2004) provide guidelines for
those attempting to gain entry to hard-to-access places such as prisons. This includes not just gaining
access, but also maintaining access. Before the late 1970s, researchers gained entry to prison
settings on the basis of employment (Clemmer, 1940; Goffman, 1961; Irwin, 1985; Carroll, 1974;
Marquart, 1986) or a relationship with a connected person (Sykes, 1958; Jacobs, 1977). Since that
time, gaining entry to correctional environments has become more complicated.
Human subject consent, institutional research boards, and issues such as researcher liability
and institutional litigation made entry more difficult. Trulson, Marquart, and Mullings (2004,
pp. 477-479) provide ten tips for "breaking into prisons" and other criminal justice organizations:
1. Get a contact.
2. Establish yourself and your research.
3. Little things count, such as being on time and showing up when it is convenient for them,
4. Make sense of agency data by keeping in contact.
190 Chapter 7 • Participant Observation and Case Studies
5. Deliver competent, readable reports on time.
6. Request to brief the agency and give a formal presentation of your findings.
7. Write a personal thank-you note to everyone involved.
8. Deal with adversity by planning ahead.
9. Inform the agency of data use, including providing copies of the publication.
10. Maintain trust by staying in for the long haul and keep in contact.
American Skinheads: The Criminology and Control of Hate Crime
Entering the world of violence, hate, and racial
paranoia of skinhead subculture, "idiots with ideology,"
invloves more than the usual challenge to field
researchers. Similar to that of other field investigators
(Thompson, 1967; Yablonsky, 1962), Mark Hamm's
attempts were on one occasion greeted with violence.
He was attacked by members of the American Front at
the corner of Haight and Ashbury in San Francisco.
Kicked in the shins with a pair of steel-toed Doc
Martens was his punishment for having been seen
conversing with an Indonesian prostitute.
Hamm (1993, pp. 100-101) describes his
I began to systematically collect data on
the American skinheads in the fall of 1989.
Drawing from the native field study approach,
I started the investigation by visiting various
U.S. cities where I tracked down skinheads in
their natural habitat (street corners, bars,
coffee shops, record stores, survivalist outlets,
rock concerts, and motorcycle shops). These
subjects were not hard to identify. They had
shaved heads and wore white power and/or
Nazi regalia. Two skinheads, for example, were
tattooed in the middle of their foreheads with
the mark of the swastika. To gain a broader
context for the research, I used the same
methods to interview skinheads on the streets
of Montreal, Vancouver, London, Amsterdam
He simply approached these youths and
asked if they would consent to an interview in a
research project that was attempting to "set the
record straight on skinheads." No electronic devices
were used. Hamm offered ten dollars for a
completed interview. Another technique involved
getting membership lists from the underground
teen press and Tom Metzger's publication WAR
(White Aryan Resistance).
After sending letters to twenty-six skinhead
leaders promising twenty-five dollars for a collect call
telephone interview, Hamm obtained nine interviews
plus an unexpected dividend. Three additional
unsolicited leaders heard about the research, called,
and were interviewed. Nonrespondents were sent
five copies of a questionnaire and asked to fill one
out and pass the others to four other members and
they would be paid ten dollars. This produced eight
additional usable questionnaires. Using the Internet,
Hamm next logged onto the WAR board. He identified
himself as a sociologist with no axe to grind and
offered ten dollars to any skinhead who came online
and conversed. This yielded two responses. Finally,
Hamm used the prison field methods approach and
with the cooperation of authorities was able to
interview incarcerated skinheads in four states.
Hamm (1993, pp. 102-103) concludes:
In summary, then, I conducted thirty-six
original skinhead interviews using native field
methods, clandestine community agency
techniques, and prison field study strategies.
I controlled for paranoia by primarily focusing
on subcultural leaders-or core members who
are likely to display the highest rates and
severity of hate crime violence-and by
presenting myself as a sociologist operating
independent of law enforcement and
community service agencies. The early Haight
street assault notwithstanding, I experienced
no life-threatening violence.
Criminology and Control of Hate Crime. Westport,
CT: Praeger, 1993.
or person who is willing to accept the purpose of the study and vouch for the researcher's presence.
A community worker's introduction of Whyte to Doc, leader of the Norton Street Gang, made access
to Cornerville possible because Doc told everyone that Whyte was his friend (Whyte, 1943). Agar
(1977, p. 145) indicates that transfer to the street with a trusted "native" has several advantages:
1. You have a guide to the territory. You quickly learn the social spaces in the neighborhood
and the kinds of persons and activities that occupy each.
2. You have an introduction into at least some groups on-the-scene. The importance of this
cannot be overemphasized. A straight outsider is often a "mark" just waiting for a disaster
to occur. An introduction from a trusted insider immediately establishes an openness
together with certain rights and obligations as so-and-so's friend.
Walker and Lidz (1977) suggest the employment of "indigenous observers," paid
researchers from the ranks of those to be studied. Such remuneration is viewed as tangible
reciprocity, or evidence of respect; such employment helps some to improve their circumstances;
however, researchers must be certain that they are not eliciting demand characteristics or the
creation of work as a result of the pay offer.
Adler in "Researching Dealers and Smugglers" (2000) gained access to drug dealers and
smugglers by befriending a neighbor who was a member of one of the drug smuggling crew.
XXXXX XXXXX observed the neighbor at work and got to know other members of the crew and
their women. The crew advanced the author's research by serving as key informants and giving
Polsky (1967) suggests that if researchers gain access on the basis of some common interest, for
instance, gambling or drinking, they should very early on indicate their true purpose: "do not
pretend to be one of them." Most subjects will accept the simple explanation that the researcher
is writing a book on the subject, although Orenstein and Phillips (1978, p. 312) correctly recommend
that a far more detailed explanation be given to the leaders, sponsors, or contact who must
answer for the investigator's presence.
Because of the very nature of most participant observational studies in criminal justice, particularly
of sensitive subjects, the use of standard sampling procedures is inappropriate. Chapter 4
discusses snowball sampling as a much-used technique in field studies, in which the investigator
builds subjects on the basis of faith and the introductions of former subjects. This further
reinforces the necessity of being "up front" with key contacts.
Reciprocity involves a system of mutual obligations. The research subjects help the investigator;
now what is owed them?
Reciprocity would entail that the researcher permit the subjects to study him or her by
answering questions they may ask. Of great importance in participant observation, particularly of
(aliases) to shroud the actual names of the subjects. Some of these pseudonyms have now become
192 Chapter 7 • Participant Observation and Case Studies
legendary in criminology and criminal justice: Sutherland's professional thief "Chic" Conwell
(1937), the Iannis' Lupollo organized crime family (1972), and Klockars' professional fence
degree to which they wish to be privy to criminal activity. Klockars (1977, p. 214) struck a deal
with Vincent: "I also told Vincent that I would not reveal his identity unless it meant that I was
going to jail if I did not, and he told me that he really could not expect me to do more." Polsky
(1967) tells us that although a researcher should not pretend to be "one of them," he or she should
also not stick out like a sore thumb. For instance, Polsky wore short-sleeved shirts and an expensive
watch in studying heroin use and trafficking. This signaled that he was not a user since an
addict would have track marks from "shooting up" heroin and would have sold any items of value
such as a watch in order to support his or her habit.
Finally, as we will discuss later in this chapter, not everyone should attempt to gather
data by means of this technique. First, long hours of boredom may make huge demands on
time before the few things the researcher wishes to observe occur. Second, there is danger; for
instance, organized crime member Bill Bonanno was concerned with Talese's welfare during
the "Banana Wars" in New York (Talese, 1971). Polsky (1967, p. 141) tells us that "most of the
danger for the field worker comes not from the cannibals or headhunters, but from colonial
Participant observers should, where possible, employ other methods to further validate findings.
The Iannis developed a pecking order or scale that they used to assign validity to the data they
gathered (from highest to lowest):
1. Data gathered by direct observation where we were participants.
2. Data gathered by direct observation where we were not direct participants.
3. Interviews which can be checked out against documented sources, for example, records of
arrest or business ownership.
4. Data corroborated by more than one informant.
5. Lowest priority is assigned to data gathered from only one source. In addition, informants
were graded from "always reliable" to "unreliable" (Ianni and Ianni, 1972, pp. 188-189).
Similarly, Steffensmeier, in The Fence: In the Shadow of Two Worlds (1986, pp. 4-6),
indicates the following validity checks:
1. The interview format provided a cross check: did the second, third, and fourth interviews
all say the same thing? Some interviews were also tape-recorded and some were checked
by Sam Goodman (the subject) himself.
2. Documents such as newspapers, personal documents, court records, letters, sales receipts,
advertising, and the like were examined.
3. Observations of Sam at work were supplemented with interviews and meetings with
customers, friends, and dealers.
4. Consultation took place with police and legal officials.
5. The data were consistent with biographies and autobiographies of thieves.
Another way of verifying the accuracy of observations is to use "member checking."
This involves having participants check your report to see if you are misinterpreting or misunderstanding
EXAMPLES OF PARTICIPANT OBSERVATION
This writer's personal interest in participant observation may be related to the fact that I was an
People of the Inner City, a series of studies he conducted in the Pittsburgh and New York City areas.
Part of his study of Millvale (near Pittsburgh) was a description of a corner gang-"the bridge
boys"-with whom I occasionally had contact. From my youth, I recall a man (who, in retrospect,
I now assume was working under the cover of a caretaker at a local private girls' high school) who
would strike up conversations with us. In his book, Gittings described quite accurately local political
corruption, gambling operations, and a gang war that occurred while I was away at college. The
competing sides in the battle ironically consisted of boys this writer had grown up with in his old
neighborhood (in Pittsburgh) versus "the bridge boys" and others in his new neighborhood.
Although we have discussed a number of participant observation studies, some further
examples demonstrate the versatility of such studies. Having previously conducted a participant
observation study of a chapter of the Guardian Angels in Detroit, Albini (1986) conducted a field
study from 1983 to 1984 of all Guardian Angels chapters in the United States and Canada.
He underwent training, became a member of the organization, and patrolled with every chapter,
Underworld performed a two-year field study of uncaught professional criminals in the London
underworld, whereas the Adlers in Wheeling and Dealing (1985) interviewed and observed for
six years upper-level cocaine and marijuana dealers and smugglers. Sullivan (1989) spent more
than four years studying youth gangs and crime on the streets of Brooklyn, while Sanchez
Jankowski (1991) spent over ten years studying gangs in three cities (see Exhibit 7.2). Eleanor
Miller (1986) did field research interviewing sixty-four prostitutes in Milwaukee, Marquart
(1986) worked as a prison guard, and Hopper (1991) studied outlaw motorcycle gangs. Wright
and Decker in Armed Robbers in Action (1997) and Burglars on the Job (1994) conducted field
interviews with uncaught, active burglars and armed robbers.
In addition to participant observation studies of criminals and the public by Polsky
(1967), Whyte (1943), Thrasher (1927), and Humphreys (1970), a variety of such studies have
been done with the police as subjects. Kirkham (1976) in Signal Zero was a professor who
became a police officer. Reiss (1971) and Skolnick (1966) did field studies of police operations.
Numerous other examples exist in the literature, including case studies that concentrate on
fewer subjects. The sometimes humorous subjects of participant observation are represented in
such titles as "The Milkman and His Customer: A Cultivated Relationship" (Bigus, 1978) or
"The Cabdriver and His Fare: Facets of a Fleeting Relationship" (Davis, 1959).
Hate groups have also been studied. Quarles (1991, 1997) studied members of the Ku Klux
Klan (KKK), while Aho (1994) did a participant observational study of Christian Patriots in Idaho
(see Exhibit 7.3). Goldstein et al. (1990) have conducted participant observation studies of drug users
utilizing "ethnographic field stations," outposts established in the community (such as store fronts) for
the purpose of collecting data and as a setting for interaction between researchers and subjects
(Goldstein et al., 1990). Such sites facilitate the collection of longitudinal data from fairly large
samples and enhance the researcher's legitimacy in the neighborhood. Indigenous observers (people
from the neighborhood) were employed by Hagedorn (1994) to study gang members. They characterized
four types of gang members: legits, homeboys, dope fiends, and new jacks. While "legits" were
those who had matured out of the gang, "homeboys" represented the majority of African American
and Latino adult gang members who alternated between legitimate jobs and drug sales; "dope fiends"
sold drugs to feed their own habits, while "new jacks" pursued such sales as a career.
194 Chapter 7 • Participant Observation and Case Studies
Other unusual subjects of participant observation have been Skipper's "Stripteasers"
(1979), Weinberg's "Sexual Modesty: Social Meanings and the Nudist Camp" (1968), and
Cavan's Liquor License (1966), a study of mating behavior in singles bars. It is incumbent on the
researcher using participant observation to consider some of the relative advantages and disadvantages
of the technique over other means available for gathering data. Mimicking the style of
the Chicago school, Grazian's (2008) On the Make: the Hustle of Urban Nightlife combines his
own participant observation of singles bars in Philadelphia with 811 student narratives of lies and
deceit in such nightclubs. One of the themes of On the Make is the intense emotional labor
of workers in such settings involved in maintaining jovial, flirtatious faux relationships with
hundreds of customers each week (Glenn, 2008b, p. 316).
Two recent works examine the extremes or edge of participant observation. Miller and
Tewksbury in Extreme Methods: Innovative Approaches to Social Science Research (2001) and
Islands in the Streets
From 1978 to 1989, Martin Sanchez Jankowski
conducted a participant observation study of about
thirty-seven gangs in Boston, New York, and Los
Angeles. The gangs were African American,
Jamaican, Puerto Rican, Dominican, Chicano, Central
American, and Irish, as well as gangs of mixed ethnic
origins. Jankowski's entree required two steps. First he
contacted community individuals or agencies that
worked with these gangs and arranged introductions
and subsequent meetings with gang leaders. He
simply explained to the leaders that he was a
professor and wanted to write a book, the idea of
which many found interesting. Despite a Polish last
name from his adopted father,