Mixed model Anova ??


We are doing a research with 2 different groups (Group A and B).
They undergo the SAME intervention (a rehab program) and we let them take a survey before, during and after the intervention.
During and after the intervention, we expect group A scores better on the survey than group B.

Will a 'mixed model ANOVA' work for our experiment?
Our within subjects would be the time we take the survey and our between subjects consists of the two different groups.

Thank you!


Active Member
How large is each group, and do you want to compare groups at both the intermediate and the final survey? What are the outcome variables? Is it reasonable to treat the outcomes as continuous variables?
Aprox 40 patients per group. We want to compare both groups during and after the intervention, so intermediate and final survey. The outcome variables is a total score on the survey. A score on 2100.

Thank you for your time!


Active Member
Sounds like a mixed linear model might be appropriate for your data. You'd have fixed terms for period, group, and their interaction, and a random term for subjects. You might need a non-standard covariance matrix to handle within-subject correlation and heterogeneity. Frankly, I'd advise you to hire a statistician.
Last edited: