skewness adjusted tstat

[seb nan]

Hi,

I am currently analysing the long-run abnormal returns of stocks. To compute the statistic, literature (barber et al 1999) advise to use a skewness adjusted method to calculate the statistic. Now I already read that there is a package by Johnson which can be used to compute a similar tstat. However, can someone explain to me how this should be used? if I simply write johnson 'var' = 0, I don't think I am getting the correct t stat...

Not sure what to do here.

thanks in advance

[Carlo Lazzaro]

Seb:
welcome to this forum.
Typing -search johnson- from within Stata looks more promising.

[seb nan]

Thank you Carlo,

Yes I have searched and downloaded the package. If I run the test, I obtain the original mean and t stat, and a (what I think is the adjusted) t1 an t2 with both a pr>t. how should I interprete these ??

[Carlo Lazzaro]

Seb:
provided that I'm totally unfamiliar with the community-contributed module -johnson-, as far as I can get the issue no matter the skewness correction both -ttest- support the evidence on no rejection of the null.

[Rich Goldstein]

I am the author of that command but I do not understand your question - the help file is clear that both the standard t-test and the modified are presented; what exactly do you not understand? have you read the original STB article (available for free at the StataCorp website? or maybe the Johnson article (citation is in the STB article)?

[seb nan]

thank you very much Rich, I was not aware of the STB article. This helps a lot.