I need to compare two groups and each one has multiple measures over time, but my concern is my variables may not meet the assumptions of sphericity or homogeneity of two-way ANOVA.
Is there a solution to this problem or a non-parametric test (such as Freidman) for two-way ANOVA?
Hello Ali,
1. Of course, you may try transforming the DV to see if that helps data to conform better to the assumptions.
2. Sphericity issues can be addressed by imposing one of the common df reductions (e.g., Greenhouse-Giesser; Huynh-Feldt; Box). You probably know that sphericity only applies as a concern when the repeated measures factor has three or more levels.
3. One option is that of exact/resampling/bootstrap error estimates for the three principal tests (main effect of group; main effect of time; group x time interaction). More statistical software is including one or more of these as an option, but here's one link that may help you understand the application: https://rpubs.com/howelb/39050).
Good luck with your work.
If you need a post-hoc analysis after that, you can use the Mann-Whitney U test with alpha Bonferroni correction for comparison number.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey