2
$\begingroup$

I have used the following code in Stata to calculate the p-values using the Holm method. However, I can't understand what to report and how to interpret these results.

*Holm procedure
parmby "reg anderson Icouples ", by(sample) norestore
by sample: qqvalue p, method(holm) qvalue(q)
by sample: list 

And for this I get: enter image description here enter image description here

At first I thought we had to compare the p-values with q-values but apparently, that is not it?

$\endgroup$
5
$\begingroup$

Each "sample" corresponds to a test, and the "q" is the transformed "p"-value to which you can compare to the nominal $\alpha$ level to evaluate overall statistical significance. In each case you have two tests. The Holm procedure begins with the "most significant" (minimum) p-value and multiplies it by the overall number of tests, i.e. 2. The next p-value is divided by the number of tests minus 1, which is untransformed. In each case, you work from "smallest" to "largest" in the untransformed p-values: if the q-value is significant at the nominal alpha level, report that test as significant, and so on for the second test, etc., until at least one test does not "meet statistical significance". Only the first example has only one test meeting significance. All others show no significant results for any test.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.