Bonferroni-corrected correlation matrix to Excel

[Andreas Head]

Dear all,

I want to create a couple of correlation matrices (with Bonferroni correction) and include these into MS word.
I want to use Bonferroni adjustments since I have around 20 variables.

Stata's -pwcorr- command allows for Bonferroni adjustments, but does not seem to offer saving the output into a file.
And due to the amount of variables, the Stata output is quite distorted which makes it unreadable.

The stata command -mkcorr- allows to store the output into a file, but seems not to allow the Bonferroni correction.

Is there any alternative way to solve this problem?

Many thanks,
Andreas

[Nick Cox]

You could save the correlations as a new variable to a new dataset using corrci (SJ). Make sure you download the latest version.

The Bonferroni-adjusted P-value is then a one-line calculation using the formula documented at [R] correlate.

You can then export your results to MS Excel.

Not supporting P-value export with corrci was a deliberate decision, as the hypothesis of zero correlation usually should not be taken seriously:

1. If the correlation isn't distinguishable from zero at conventional levels, your sample size is just too small.

2. Wanting to know about the uncertainty associated with a correlation can be reasonable, but the confidence interval calculation tackles that directly.

3. I've seen people evidently wanting to use the P-value as if it were a measure of strength of correlation -- but that is what the correlation does!