skewness-adjusted bootstrap t-statistics

[Tom Smith]

Hi

I want to compare means of two variables which are highly skewed. I tried skewt test and johnson test, but it does not allow me to compare two variables.
Can some one suggest me a command to calculate this?

Thanks

[Nick Cox]

Also cross-posted at http://stats.stackexchange.com/quest...p-t-statistics

These modified tests never seem to attract more than local or transient attention. Postings in

http://www.statalist.org/forums/foru...u%C3%9F-z-test

point to a more modern method, namely using a generalized linear model with an indicator predictor and appropriate family and link. Loosely speaking, moderate skewness matters less than many texts imply, while if skewness really is very high there is usually scope for debate about what is the right question.

[Carlo Lazzaro]

Tom:
as an aside to Nick's comprehensive guidance, you may want to compare the results of -ttest- to the ones obtained via a bootstrapped ttest. This procedure is covered under Example 3 of -bootstrap- entry in Stata 13.1 .pdf manual.