Parallel box plots and significance

#1
I once knew this but have forgotten. I think that when the IQR of one box plot doesn't overlap the IQR of another, this can be interpreted as indicating that (H0: equal medians) can be rejected. What have I forgotten? :(
 

JohnM

TS Contributor
#2
I don't believe this is the case. You need to do a Mann-Whitney U test, also called the Wilcoxon test (the nonparametric counterpart to the t-test) to do the hypothesis test.