What would be the best way to check the significance of many hypotheses?

Hi there!

I got an assignment in which I have to test 12 hypotheses, some of which have to be tested using t-test, and some using ANOVA (or even MANOVA). If I test each one of them at p=0,005 (Bonferroni), FW alpha would become 6%, which seems OK, but I'd like to try some multistage comparison, like the Holm procedure. In order to do that, I'd have to order my ratios, and I can't order F-ratios and t-ratios like they are the same (since df1 isn't 1). Is there a way to use Holm procedure on such a mixture?