0

I performed 8 ANOVAs on the same subjects with different dependent variables. We tested the significance of each term & interactions using F-tests (ANOVA). A reviewer asks us if we have applied corrections for multiple comparison.. But it's not usual practice to adjust p values of the main effects and interactions... Indeed, p-values adjustments are usually applied to post-hoc tests performed of omnibus ANOVAs, and not directly on the F-tests.. Could you help me to explain to the reviewer ?...

(I know that same kind of questions (here & here..) have been asked before but there is no real answers..)

learners
  • 579
  • 6
  • 20
  • If you ran 8 ANOVAs this sounds like a fishing expedition that is in need of serious discounting, e.g., serious multiplicity adjustment. – Frank Harrell Mar 09 '22 at 12:23
  • What does it mean exactly? I have 3 tests for each ANOVA (two main effects and one interaction), so it means I have 3*8 comparisons that I should correct ?! – learners Mar 09 '22 at 12:41
  • Since you are fitting 8 separate models you should multiplicity adjust for the 8. Not as sure about the 3. – Frank Harrell Mar 09 '22 at 13:04

0 Answers0