How to add more decimal points to p-values?

[Sabrina Khan]

Hi everyone! I'm running several logistic regressions and I was wondering if there was any way to get more decimal places for p-values? I am using the Bonferroni correction so I want to make sure I have enough decimal points.

Thank you!

[Clyde Schechter]

There is a -set pformat- command that modifies the number of decimal places in the regression outputs. See -help pformat- for details. But the most you can get with that is 4 decimal places, which may not be enough for your purposes, and isn't much different from the default display of 3 decimal places.

But you can also pull the pvalues out of the -r(table)- matrix that is left behind after regression, and if you don't like the way they are displayed there, you can always -display- them yourself with the desired format. For example

sysuse auto, clear
regress price mpg
matrix list r(table)
display %12.11f r(table)[4, 1]

[Joseph KAMTCHUM TATUENE]

Good morning Clyde, thank you very much for your tips.
I wanted to know how I can get the p-value fully displayed after running the "pwcorr" command (for correlation). It does not appear in the list of stored results/outputs.
Thanks in advance.

[Nick Cox]

The sig option for pwcorr displays P-values.

[Joseph KAMTCHUM TATUENE]

Thank you, Nick. I see that but my p values are very small and I only see 5 zeros after the comma. I wanted to display more decimal places so that I can convert the value into the corresponding scientific notation (requested by a reviewer).

[Nick Cox]

If you want P-values in more detail AND also the number of non-missing values varies from correlation to correlation, then I know no alternative beyond looping over pairs of variables and showing each P-value as well as the number of values used and the correlation itself.

In many fields, but not all, once you know that the P-value is <0.00005 you stop caring. The P-value is often dubious on other grounds any way, e.g. lack of independence.

[Joseph KAMTCHUM TATUENE]

Thank you very much! Have a great day!

[Joseph KAMTCHUM TATUENE]

I think I found the answer. If not mistaken, the command below provides the scientific notation of the p-value.

matrix list r(sig)

[Nick Cox]

You're right!