Hi,

In the logistic regression function with the Wald test, is there an automatic correction like Bonferroni for the pvalues ?

Thanks

Hi,

In the logistic regression function with the Wald test, is there an automatic correction like Bonferroni for the pvalues ?

Thanks

(Disclaimer: I am not a statistician)

I think of the Bonferroni correction as a correction to the *significance threshold*, not the p-values. To compute the Bonferroni-correct significance threshold, you can divide 0.05 by the number of tests:

```
0.05 / mt.count_rows()
```

Yes, it’s a conservative correction to the significance threshold, and we don’t do that automatically, in part because the appropriate correction depends on the context. In GWAS, often the divisor is not the number of tests, but an estimate of the number of “independent” tests, which is smaller because of linkage disequilibrium. A typical threshold is `5 * 10^-8`

, see this paper for a longer discussion.

Yes I agree, I was thinking of a function like p-adjust in R :

https://www.rdocumentation.org/packages/stats/versions/3.5.1/topics/p.adjust

I don’t have so much significant p values that’s why I was thinking that some kind of correction was done implicitly.

Thanks for the paper.