Is it possible to get the t-statistics for Pearson Correlation Coefficient?

[Jae Li]

Hello every Statalist,

I am curious to know, do you think it's possible to calculate the t-statistics for Pearson Correlation Coefficients in order to separate those observations with significant correlation coefficients from non-significant ones?

Any comments will be very welcome! Many thanks for your time in advance!

[Clyde Schechter]

If you use -pwcorr- with the -sig- option, you will get a p-values for the correlations returned in matrix r(sig). It does not report the t-statistic itself.

If you want a t-statistic for a correlation you can either calculate it directly yourself:

corr x y
local t_statistic = r(rho)/sqrt((1-r(rho)^2)/(r(N)-2))

Or perhaps simpler, use -regress- instead of -corr-: the t-statistics are the same.

[Jae Li]

@Clyde Schechter Hi Clyde, thank you very much for your help!! It works for me!