I am having trouble deciding what type of analysis to use for the following data:

I have two groups (one control, one treatment) and took a pre-measure of my dependent variable, and then 3 post-measures.

I am interested in understanding if the treatment affected the dependent variable over the course of the study.

I originally did a rmANOVA, and found no significant interaction (treatmentxtime), but found a main effect of the treatment. I think this means that the treatment group was different than the control over the entire study, including before treatment began.

My question is, is there a better analysis that takes into account if the groups were in fact different before treatment? With a t-test, they were not different, but somehow with all of the data throughout the experiment, they are different.



1 Answer 1


One simple way would be to compute gain scores - post1 - pre, post2 - pre, and post3 - pre and do your analysis in terms of treatment and period (1st, 2nd, or 3rd post meas) -- or average the 3 gain scores so that you just have 1 value for each subject.

A lot depends on what exactly those pot-treatment measures are. There must be a reason for 3 of them rather than just 1. Can you say more about that, and I may add to my answer.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.