0

I have some violation of assumptions (normality and equality of variances) is my analysis and I decided to a use robust technique (lmrob function in R). I have a continuous response, one categorical predictor and one covariate. I was wondering which test I should use for the multi-comparisons test/Post hoc analysis?

Farid
  • 83
  • Can you please share how to do the robust pairwise comparisons with Bonferroni correction in R? – Emil Feb 16 '24 at 22:23

1 Answers1

1

Almost all of the usual suspects (closed testing procedures, Bonferroni-Holm, Bonferroni, graphical testing procedures, resampling based methods...) do not really depend on the analysis you performed, as long as it produces a valid p-value.

Björn
  • 32,022
  • thanks for the reply. It means that after running the robust method, I can use Tukey, Bonferroni as the Post Hoc tests? But I heard that the Tukey method is based on the normality assumption. – Farid Sep 24 '18 at 14:19
  • Tukey would be an example of a test that would not work. Those tests that just operate on p-values are the ones where there is no problem, similarly if resampling is the key ingredient. – Björn Sep 25 '18 at 05:36
  • @ Björn, can I use the robust pairwise comparisons with Bonferroni correction. – Farid Sep 25 '18 at 15:14
  • Bonferroni makes no assumptions about where the p-values come from, so if they are valid p-values that is fine. – Björn Sep 25 '18 at 20:51