data modeling help


I am working to analyze my data after a clinical trial. I was brought on partway through the study and was not able to make decisions about data quality, collection times, etc., so what I have is a bit messy.

I have ~30 (22 with attrition) participants with weight loss results as my primary outcome. The trial had several parts to it that participants (ideally) adhered to. I have recorded all adherence rates and have completed all the dietary analysis. What I would like to do now is use the weight loss outcome and tease apart which portion of the program had the greatest impact on it. My data is currently organized in columns: participant ID, 12-month weight loss, daily weighing %, dietary variables (kcal, macros, etc), session attendance %, homework completion %... I do my analysis primarily in R;

With the small sample and messy data, I am not sure the best way to do this. The statistics courses required of graduate students at my institution focuses on pretty 1000+=n samples, rather than the smaller messier samples often seen in human research.

I would appreciate any advice for how best to approach this analysis.

Thank you!!
the primary analysis in nearly every weight loss trial ive seen is weight change from baseline at some time points. usually t-test to compare the groups. attrition is usually handled by LOCF or BOCF. 30 is probably good to go on some t-tests, even with some wackiness. good luck with your study. Did it work?
@fed2 Thanks. Yes, I completed several t-tests either using LOCF or interpolation. Timepoint weight losses were significant, as were dietary changes. I am hoping there is more in-depth analysis I can do, although perhaps this is wishful thinking.