For Susan Athena (Please have the service cancel the question for Mr. Glenn G then). Thanks.Data set of all hospital discharges in New York with an admitting diagnosis of a heart attack who didn’t have surgery. There are 12, 844 cases. (age in years; sex coded M for males, F for females; diagnosis classified in terms of disease as 121 for heart attacks with cardiovascular complications who did not die, 122 for heart attacks without cardiovascular complications who did not die and 123 for heart attacks where the patient died; DRG’s group patients with similar management; LOS is the length of stay in the hospital; Died was marked 1 for patients who died and 0 if alive; Charge is the total hospital charge).Patient Diagnosis Sex DRG Died Charges LOS Age1 41041 F 122 0 4752 0010 0792 41041 F 122 0 3941 0006 0343 41091 F 122 0 3657 0005 0764 41081 F 122 0 1481 0002 0805 41091 M 122 0 1681 0001 0556 41091 M 121 0 6378.6400 0009 0847 41091 F 121 0 10958.520 0015 0848 41091 F 121 0 16583.930 0015 0709 41041 M 121 0 4015.3300 0002 07610 41041 F 123 1 1989.4400 0001 06511 41041 F 121 0 7471.6300 0006 05212 41091 M 121 0 3930.6300 0005 07213 41091 F 122 0 Yen * (MISSING DATA) 0009 08314 41091 F 122 0 4433.9300 0004 06115 41041 M 122 0 3318.2100 0002 05316 41041 M 122 0 4863.8300 0005 07717 41041 M 121 0 5000.6400 0003 053A. Was the heart attack study a designed experiment or an observational study?B. Looking at assumptions and conditions for inference for proportions list each of them and for each a comment on how they apply to the confidence interval for the proportion of patients who died.C. Why are the median, midrange, range and IWR of little value for 0-1? (midrange average of the largest and smallest values)a. Summary of diedb. Count 12844c. Mean 0.109779d. Median 0e. Midrange 0.5f. Standard deviation 0.312626g. Range 1h. IQR 0d. create a confidence interval for the proportion of patients who are male. Use this to test the hypothesis that heart attack patients are evenly divided between male and female. Already Tried: I included what I got in the data set and questions.
Thank you for requesting me. The question says there are 12,844 cases but only 17 are listed here. Is there more data?
Yes, I apologize. How to upload the excel file?
Attachment: 2012-06-24_232224_heartatk.xls
Thanks for that!Your answer is below.Kind regards,SusanThis is an observational study. The researchers only observed what happened; they didn't put subjects into controlled, randomized trials and apply different treatments or manipulate any variables.The main assumption for the confidence interval for proportion is the assumption of normality. In order to meet this, the smaller proportion has to have at least n = 5. In this case, 1410 patients died, so the assumption is met.The median is of little use because there are only two values. Since more patients lived than died, the median will be 0. The median is essentially useless when all the data is at the ends of the range. The IQR is useless for the same reason. The data through the 3rd quartile is just 0, so there's nothing to compare.The range and midrange are useless because we know the range in advance; there are only two possible values.d.I calculated the 95% confidence interval for the proportion of men.p +/- z * sqrt [p(1 - p)/n]0.6057 +/- 1.96 * sqrt [( 0.6057 * 0.3943)/12844]0.6057 +/- 1.96 * sqrt [0.0000186]0.6057 +/- 0.00845(0.5972, 0.6141)Since the interval does not include 50%, we can be at least 95% confident that the true population proportion is not 50%. The proportion of male heart attack victims is greater than the proportion of female heart attack victims.
Experience: Tutor for Algebra, Geometry, Statistics. Explaining math in plain English.
What about independence assumption, randomization condition, 10% condition, sample size assumption, success/failure condition. With respect to question b.
Hi. Thanks for that. I found this reference with the assumptions you mentioned.http://www. google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0CFMQFjAA&url=http%3A%2F%2Fcims.clayton.edu%2Fmhudachek%2Fbuswell_math_1231%2FClassnotes%2C%2520ppt%2F1231Chap19%2520CI.ppt&ei=QrTnT5ulM_Cu2AWdhJzaCQ&usg=AFQjCNEdUuMtrLFBj5mZXdGP5aGu0CjHTw&sig2=ZN_L4Y7TBtE9s5imAfNMfAKind regards,SusanAssumption of independence:We assume that the data values are independent of each other. That is: there's no reason to believe that one person's death depends on another's or that they have the same underlying root cause.Assumption of randomization:Were the data created from a random experiment? No, they were not. This is an observational, not an experimental condition.10% condition:Is the sample size no more than 10% of the population?The population is "heart attack victims who didn't have surgery". Since New York's population is less than 10% of the population of the US, it's reasonable to assume that the sample studied is less than the population in the same time period just in the US.Sample size assumption/Success-failure assumption (assumption of normality)Some people use n>= 10; I used n>=5. Either way, we expect the smaller proportion to be greater than 5.
I'm sorry my text uses a n>=10 not five when discussing the success/failure condition?. what is the difference and how will that affect the answer related to the above first question.
That's no problem. The result is exactly the same. In this case, n = 1410, so whether 5 or 10 is the threshold, we're well above that.Kind regards,Susan
Can you help me with a problem set of questions? I will tip.
Hi. Thanks for your question! Please post your questions and I'll see if I can assist.Kind regards,Susan