Question: Question on Bonferroni correction

0

brunobsouzaa •

**510**wrote:Hi Everyone.

So, my question is quite simple but raises me a little doubt. I have a data like the following:

```
Patient Sex Age SNP1 SNP2 SNP3 SNP4 Outcome
1 M 56 TT CC AG AT 1
2 M 47 AT CG AA AA 0
3 F 85 AA GG AA AA 1
4 M 90 TT GG AG TT 0
5 M 75 TT GG AG TT 0
6 F 45 AA CC AG TT 0
7 F 80 AT CG GG AT 1
8 M 53 AT CG GG AT 1
9 M 54 AT CG AG AA 0
10 F 61 TT CC AG AT 1
```

It's a case-control study, so for `Outcome`

, `0`

= Control and `1`

= Case

For this data, I've performed Univariate logistic regression and multivariate logistic regression like the following:

```
SNP1 ~ Outcome
SNP2 ~ Outcome
SNP3 ~ Outcome
SNP4 ~ Outcome
SNP1 + Age ~ Outcome
SNP2 + Age ~ Outcome
SNP3 + Age ~ Outcome
SNP4 + Age ~ Outcome
```

Also chi-square test was performed to compare alleles x Outcome in both groups.

So, the question here is simple... Should I correct my p-values considering SNPs isolated which would lead me to 3 tests for each SNP or should I correct by general number of tests? To clarify, SNPs are not related to each other!

Thanks.

May be more appropriate for CrossValidated StackExchange. However, I recall searching for information on a similar question and I found conflicting answers about how to perform p-value adjustment in this scenario.

71k