Multiplicity: Mixed feelings [Design Issues]

posted by ElMaestro  – Denmark, 2012-11-30 16:24 (4950 d 19:26 ago) – Posting: # 9646
Views: 25,007

Hi Helmut,

❝ Out of curiosity: Did FDA want you to adjust α to deal with multiplicity (>90% CIs: Bonferroni, Holmes, Hochberg, :blahblah:)?


Some personal observations from the bag of EU/US/other memories:
  1. Agency statisticians seem to very much like the the idea of multiplicity adjustment. So if you ask if they want you to do it, the answer is often a clear Yes (Bonferroni) in autopilot mode referencing this book.
  2. When the assessment report comes and the poor assessor -who isn't the statistician- realises that there are no 90% CI's in the dossier, there will be a deficiency letter and 90% CI's must be submitted.
  3. I see companies trying to adjust alphas even when they do one test and two refs (as in Ref1 EU sourced and Ref2 US sourced). It results in funny and endless discussions around e.g. the EU guideline stating that BE must be shown on data reflecting the products of interest, pt. 1 above, and the fact that sponsors are hoping for BE to be shown against both Refs.
So while I think the underlying idea of multiplicity adjustment is appealing in theory I find the whole matter slightly pseudo-scientific in practice. Am eager to hear your views or any other user's views.

Pass or fail!
ElMaestro

Complete thread:

UA Flag
Activity
 Admin contact
23,655 posts in 4,993 threads, 1,570 registered users;
113 visitors (0 registered, 113 guests [including 14 identified bots]).
Forum time: 12:51 CEST (Europe/Vienna)

Science is simply common sense at its best that is,
rigidly accurate in observation, and
merciless to fallacy in logic.    Thomas Henry Huxley

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5