-1

I have a dataset in Stata with two variables, perc_vol (numeric) and low_medium_high (values "L", "M", "H").

I want to run a Mann-Whitney test to compare perc_vol between H and L, H and M and L and M.

For two values I can use code like ranksum perc_vol_fin_lei, by(group_liquidity) but for three, I don´t know how I can do this.

Nick Cox
  • 35,529
  • 6
  • 31
  • 47
Marcel
  • 1
  • 3
  • Mann-Whitney is for comparing two groups, and two groups only. The most common similar technique for three groups is Kruskal-Wallis, and one such test is preferable to three applications of Mann-Whitney. It's not obvious here that you can't use ANOVA or some other generalized linear model. – Nick Cox Aug 31 '23 at 07:45

0 Answers0