I have some one-group pre-post data where the outcome is a score and an IV of interest is the whether or not the subject has performed the task in the past. I am looking at this in a few ways and am wondering which one is correct:
1) GLM where outcome is the change and the IV is the past exposure.
2) GLM where outcome is the change and the IV is the past exposure, adjusting for pre scores.
3) GLM where outcome is the post and the IV is the past exposure, adjusting for pre scores.
4) repeated measures ANOVA where pre post is time 1,2 and IV is past exposure.
I also looked at the interaction for model 2 for pre score x past exposure and it was close to significant and the IV became significant (it was originally non-significant without the interaction term in the model).
Can anyone suggest which model is more appropriate and how to interpret the results, especially the significance popping up after the interaction is introduced to the model?