Hi Emilio

If you permit, I'd like to go 'upstream' of your question a bit and ask why you want to adjust for multiple comparisons. I almost never do (though might if I were doing more GWAS or RCT work).

I justify this by appealing to "Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology.1990;1:43-46".

My interpretation of Rothman's piece is that such adjustment only makes sense if I hold a universal null hypothesis of no differences. Were I to hold such a null, I must then believe that “only purely random processes govern the variability of all the observations in hand.” (Rothman, 1990:44).

Back downstream to your question, Glickman and colleagues (www.glicko.net/research/multiple-tests.pdf‎) have shown that roughly half of leading journal articles mentioning multiple comparisons then adjust for multiple testing. Apparently Bonferroni approach was used by most.

Hope this helps

Best

Marco

Thanks a lot, Marco. You know it´s an ongoing debate. I´ll read these papers you kindly share. Greetings.