Apply Bonferroni correction to Chi2 test

[Ronja Sundborg]

Hi,

I have a dummy variable called "pay" and four different treatment groups.

I wish to examine whether "pay" is dependent on treatment group. By running a simple Chi2 test with the following code, I receive a p-value below 10%.

tab group pay, chi2

Now, I wish to examine the issue further by conducting pairwise comparisons. However, as this implies multiple tests, I also wish to apply the Bonferroni correction. I am unable to find how I can conduct these pairwise comparisons with a Bonferroni correction - can anyone help me with this?

Thank you in advance,
Ronja Sundborg

[Carlo Lazzaro]

Ronja:
welcome to this forum.
If you have 4 different treatment groups to be compared each other, you should perform 6 comparisons (see: https://en.wikipedia.org/wiki/Binomial_coefficient).
Hence, if you arbitrarily set the significance cut-off at 0.05, to reach p<0,05 a head-to-head comparison should reach p<(0.05/6)=0.00833.

[Joseph Coveney]

This has come up on the list before. Here's a suggested approach.

[Ronja Sundborg]

Thank you very much for your help!