I have three drug groups with measures over 5 time points. I need to study their interaction x time. Since they are non parametric data, I am looking for a non parametric alternative for a 2-way repeated measures anova. I also require post hoc tests for that.
Hello Sooraj R.Kartha, although I still recommend using RM in this case, you can use SPSS software and Friedman's pairwise comparison table.
Faye Anderson Isn't Friendman for one independent variable having one repeating dependent variable? What I am looking is to compare three independent variables having one repeating dependent variable. If the data is parametric I would be considering 2 way anova, but I am not sure what is the non parametric to 2 way anova?
Abolfazl Ghoodjani - You have given a pairwise comparison for which test? I am actually looking for an alternative to 2 way anova.
Sooraj R.Kartha , you are correct. Friedman test won't handle your data. For a non-parametric test, you could use aligned ranks transformation anova. With a good implementation (for example, in R), it can handle repeated measures, and has functionality for post-hoc tests.
Note that the interaction of ranks is something entierly different than the interaction on the quantitative scale. Also on the quantitative scale, the inerpretation of the interaction depends on the transformation. Make sure that the interaction term in your model captures what you really want and that you interpret it correctly.
Karen A. Darbinyan , can you elaborate how one would apply the Mann-Whitney test to a situation analogous to a two-way repeated measures anova ?
Hello Sooraj R.Kartha
As far as I understand we are dealing with a Mixed ANOVA.
where in
Your within-subjects factor is time. Your between-subjects factor is groups.
If the distribution of the data is not normal, I'm sorry, don't care about it (of course, I know this topic has many opponents).
If the Assume of sphericity is not confirmed, use Geisser-Greenhouse correction.
- Badges
- Science topic
- Similar topics
- Mathematics