I am doing a non-parametric ANCOVA (my residuals are not normally distributed and transformation did not help), I have been using the sm.ancova function from the sm package.
Now I want to do a post hoc analysis but I'm not sure how to go about it as I need a non-parametric post hoc test on the adjusted means. however, I have not found a way that works for me.
In my study, we have three arms, and I'm actually only interested in the comparison between the control and each of the experimental arms but not the experimental arms to each other, so doing a Dunnett's test but I could not find a non-parametric alternative.
I also considered a Dunns test and simply compared all three but I am not sure how to include my covariate here.
Any advice?
If there is a significant difference by non parametric ANCOVA, you can conduct pairwise Wilcoxon rank-sum tests between the control group and each experimental group, adjusting for the covariate
Hello Chiara,
Since you are conducting a non-parametric ANCOVA and you're interested in comparing the control group to each experimental group separately, you can consider using non-parametric post hoc tests such as the Dunn's test or the Wilcoxon rank-sum test (also known as the Mann-Whitney U test).
Here is how you can approach it:
One approach you can consider is using rank-based methods, the Wilcoxon rank-sum test, while controlling for the covariate.
- Badges
- Science topic
- Similar topics
- Filing