There are 10 dependent variables and 2 groups. When I wanted to compare 10 dependent variables between two groups, I performed MANOVA.
I wonder if there is a multiple comparison problem in that case. Should I have performed Bonferroni correction because of 10 dependent variables? Or is Bonferroni useless because MANOVA just does analysis once?
Best Answer
There is no need for correction.
Think about it like that: you could conduct 10 independent ANOVAs for each of the 10 dependent variables (in case of only 2 groups ANOVA reduces to t-test, but it does not matter for this question). Then you would get 10 different p-values, and certainly would need to correct for multiple comparisons, e.g. by applying Bonferroni correction. But if you use MANOVA, then you get only one single p-value and so there is no need for any corrections. The test is done by considering all 10 variables together, so there is no "multiple testing" (meaning that you are not doing any test "multiple times").