I'm trying to analyze a pre/post study with 4 treatment groups. The Dependent Variable, Speed, measures how quickly a person's facial expression reaches peak intensity; this measure is normally distributed after log10 transformation. Another variable, Intensity of emotional expression, is also measured at both pre and post and might influence the Speed. Intensity is continuous and not normally distributed.

Question: If Intensity does indeed significantly correlate with Speed, how should this be accounted for, statistically? IF Intensity was "fixed"/static, I would think that (using SPSS), I would choose the repeated-measures option of the General Linear Model, and enter Intensity a covariate. However, since Intensity is measured--and not identical--at both pre- and post-, can Intensity be included as a covariate? If so, how (Enter 2 covariates: Intensity at pre, and Intensity at post?; Create a change-score to use as the covariate?). In one of the treatment conditions, Intensity is significantly different at post-treatment.

Update: I've been advised that mixed-effects modelling (mixed linear model) might be the way to go. After a quick look, the mixed approach seems appropriate, but please advise if you think that there are better/easier options.

My stats and SPSS skills are fairly basic so, if explanations can be calibrated to novice-level, it'd be much appreciated.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.