Bonferroni or not...

#1
We run a study, comparing 3 treatment groups (same 6 individuals), and we are comparing the mean blood flow (in an area) between them. The data is not normally distributed. Would you perform a Friedman and then if there is a difference analyze it with a Mann Whitney U? Do you have to apply a Bonferroni correction here? Because you are comparing means of one variable between 3 groups.. or given that just one variable between the 3 groups is analyzed, we don't need a Bonferroni?
 

Karabiner

TS Contributor
#2
Mann-Whitney is for between-groups comparisons. For repeated measures (cf. "same 6 inidividuals"),
you can use Wilcoxon Signed Rank Test.

Personally, I would not use Bonferroni in a post-hoc pairwise comparison, after the omnibus test
turned out significant. But if it is very important for you to prevent false-positive results, you
can do it.

With kind regards

Karabiner