Running a GLM w/ interaction w/ 2 numeric predictors, neither of which are normal when plotted against the outcome variable. Is there any other transformation I should try, or should I just list this as a limitation? FWIW I'm also mean-centering the variables after I transform them (was told to do this in interaction models.) Thanks!
This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.