Question: Number of Comparisons in Bonferroni Correction
0
gravatar for khatami.mahshid
6 days ago by
khatami.mahshid10 wrote:

Hi,

I have a list of variants and each of them has a different number of donors in 3 groups. for example:

var1:
group1:10-
group2:20-
group3:25

var2:
group1:20-
group2:15-
group3:20

var3:
...

I have a p-value of each variant, but I don't know what is the number of test for converting p-value to adjusted p-value of Bonferroni.

can anyone help me?

thank you

ADD COMMENTlink modified 6 days ago by ATpoint21k • written 6 days ago by khatami.mahshid10
2

The number of variants, is the number of tests you have used.

ADD REPLYlink written 6 days ago by Benn7.4k

but each variant is independent and I do not compare variants with each other.

ADD REPLYlink written 6 days ago by khatami.mahshid10
3

Bonferroni (and all multiple testing corrections) corrects for each test that you performed. If you have 1000 variants, and you have 1000 p-values, then you've performed 1000 tests. Even if they're independent, with a p-value of 0.05 performing 1000 tests would still results in 50 "significant" results just by chance.

You might want to consider performing a Benjamini-Hochberg (or FDR) correction instead of a Bonferroni as it's less conservative.

ADD REPLYlink modified 6 days ago • written 6 days ago by shawn.w.foley840
2

Okay it sounds like you do not completely understand the 'why' you should correct for multiple testing (no offense). Best is to read more about it, for example here some background about the test.

If you have many variants, you would be better off using a less conservative method, like Benjamini-Hochberg.

ADD REPLYlink written 6 days ago by Benn7.4k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1413 users visited in the last hour