I conducted a study to investigate the effect of instagram profiles on mood. Therefore I created two profiles on instagram and used a pre-post-measurement of mood (PANAS-scale). Participants (N=130) were randomly distributed to one of the profile conditions. So I had one within-subject factor (pre- post measurement) and one between-factor (viewed profile). I wanted to use a mixed ANOVA to calculate the results. Unfortunetly, the Levene's test turned out to be significant. Now I was wondering what to do next. I've already tried to transform my data by using the box-cox transformation, but the results turned out to be quite confusing. What can I do? Thank you in advance.

  • $\begingroup$ Hard to say anything without seeing the data. Can you show some plots? How exactly did you run the Levene-test? Generally model assumptions are never precisely fulfilled, and occasionally an analysis is fine even if a test shows that a model assumption is violated. Testing model assumptions and doing transformations and the like conditionally on a significant misspecification test result is controversial, will not always improve matters , and sometimes do harm. $\endgroup$ Jul 6 at 14:49


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.