Hi,
I would like to ask if I have considerably positively skewed raw data for my task performance measures, should I: 1) adjust the variables for covariates (e.g. age, sex and years of training) first, and then transform the residuals using maybe box-cox transformation; or 2) transform the raw variables before adjusting for the covariates?
I've initially done (2) but then I noticed in this paper (http://www.ncbi.nlm.nih.gov/pubmed/21336711) that transformation was done on the residuals, which seems to make more sense.
On a related note, how do I obtain the residuals after adjusting for the covariates?
Thank you very much!
-Yi Ting