My thesis supervisor insisted I do a MANCOVA with 5 dependent variables. However to answer my research questions I need to use the results of the subsequent 5 univariate tests SPSS automatically produces, and the post hoc comparisons for the significant ones. Now I have selected to have SPSS automatically correct my posthoc comparisons using Bonferroni. But do I need to lower the significance levels of the univariate tests (assume significance at .01)? Or is this unnecessary and protected by the MANCOVA?
Asked
Active
Viewed 255 times
1
-
I think you should use p-value adjustment for the univariate tests but you can use other methods that are not as conservative as Bonferroni. – Michael R. Chernick Jul 07 '17 at 00:22