Multiple Hypothesis Correction
Posted on
Sabha
Joined: 05/18/2010
Forums
Hi ,
I am running ACE and nested submodels on 100 twin pairs ( 50 MZ and 50 DZ twin pairs) to estimate A,C and E for 10,000 genes. Do I need multiple hypothesis correction on these estimates?
Thanks in advance.
Sabha
I'd need a lot more
Log in or register to post comments
Like Ryne said, it's tough to
The way to think about it is like this: On a standard significance test with alpha at .05, you've got a 5% chance of erroneously rejecting the null hypothesis. Every additional test to reject the same null carries its own 5% chance of a Type-I error.
If you are performing several tests, and allowing a significant result on any of them to be evidence for the same hypothesis, you need to do some sort of multiple testing correction.
That means if you're doing 10,000 tests to reject the null hypothesis that specific genes do not influence a behavior, you should probably be doing multiple testing correction.
It's also worth noting that multiple testing correction doesn't change your estimates. It only changes your significance levels.
Log in or register to post comments
In reply to Like Ryne said, it's tough to by tbrick
I apologise for not making it
I have expression value of 10,000 genes from 100 twin pairs. I am running ACE and nested sub models AE,CE and E for each gene seperately in other words running 10,000 chi-square tests for each model. Based on your reponses it looks like I need to adjust the pvals for multiple hypothesis correction.
Log in or register to post comments