Multivariate Kruskal-Wallis test


I have several count values for two groups in several variables, as seen on the plot.
I have seven observations in the purple group and 74 in the grey group.
I have 10 variables (x axis).

I want to compare the distributions of the purple and grey groups in each of the 10 variables.

I thought I could make a Kruskal-Wallis test, but multivariate.
What do you think? How to implement it in R?


Screen Shot 2020-10-30 at 11.07.22.png


Less is more. Stay pure. Stay poor.
If you have counts you should look into count regression models such as Poisson of negative binomial regression.