0
$\begingroup$

I am making two separate comparisons between populations - I am comparing the frequency of multiple alleles between populations - Population 1 Vs Population 2 and Population 1 Vs Population 3.

So my resulting data frame of p values will look like this (I have just populated with x's instead of numbers):

        Pop1_vs_Pop2 Pop1_vs_Pop3
Allele1       x            x
Allele2       x            x
Allele3       x            x

So my question is this: with Bonferroni corrections, so I correct by 3 (ie - divide 0.05 by 3 to determine the significance cut off level, since I am running two independent tests at three alleles each), or do I correct by 0.05/6 (the total number of tests)?

$\endgroup$
0
$\begingroup$

It depends on whether you want to control the familywise eror rate (FWER) overall, in which case you use $p=\frac{\alpha}{6}$, or to control them within each comparison of two populations, in which case you use $p=\frac{\alpha}{3}$. I recommend the former for reasons of interpretability.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.