0
$\begingroup$

I performed a study where I took 9 measurements from each participant for 2 factors (time/ location) and 3 variables each. These created the following combinations: T1L1, T1L2, T1L3, T2L1, T2L2, T2L3, T3L1, T3L2, T3L3.

My data is not normally distributed. I ran the Friedman test and Wilcoxon signed-rank test for above combinations. All of the combinations are statistically significant.

Now, I am interested in performing the following pairwise comparisons: T1-T2, T2-T3, T1-T3, L1-L2, L1-L3, L2-L3. Any suggestions? One idea I can think of is just taking means and performing Wilcoxon signed-rank test. For example for T1-T2, I take Mean(T1L1, T1L2, T1L3) and Mean (T2L1, T2L2, T2L3) and then perform Wilcoxon signed-rank test. Is it okay to do this? Are there any other better alternatives?

Wilcoxon signed-rank test

$\endgroup$

1 Answer 1

0
$\begingroup$

If I understood correctly, you are looking for a non-parametric alternative to multivariate ANOVA. Although I have never used this, this R package might do exactly what you are looking for npmv.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.