Is it ever okay to smooth data? I am comparing two data sets that intuitively, we know have some kind of correlation (chicken hatchings versus total chicken population). But the first set (hatchings) can be all over the place month-to-month, because chicken processors are erratic in their orders (one month high, the next month low, etc.). I appreciate that smoothing is generally considered a statistical no-no; however, I feel that if our larger goal is to illustrate that one TREND is correlated with another data set, then I feel I'm doing a DISservice to not smooth the data and kill off some noise. What do you think?
P.S. I am happy to send you further information. Thanks!