st: significance level after ols, robust

["Viktor Slavtchev" <viktor.slavtchev@uni-jena.de>]

Dear Stata-listers,

I’m estimating a simple OLS with correction for intragroup correlation:

reg y x1 x2, cluster(race).

Now Stata reports estimated coefficient, SE, t–values and P>|t|. For example:

2.910862   1.303605     2.23   0.089

t-value of 2.23 means significant at 5%, at least according to a standard normal distribution table.  Significance level reported by Stata is, however, 8.9%. Why?

Thanks

Viktor