Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

re: st: test multiplicity and use of F or Chi2 distributions after ANOVA or regress


From   "Airey, David C" <david.airey@Vanderbilt.Edu>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   re: st: test multiplicity and use of F or Chi2 distributions after ANOVA or regress
Date   Mon, 21 Jun 2010 16:41:00 -0500

.

Just found the FAQ:

http://www.stata.com/support/faqs/stat/wald.html

> In doing some ANOVAs and using different approaches for post-hoc tests of interaction simple main effects, I noticed that use of the Chi2 and F distributions are the same when appropriately scaled, but their p-values differ.
> 
> . display chi2tail(2,45.26)
> 1.486e-10
> 
> . display Ftail(2,30,22.63)
> 1.019e-06
> 
> I'm told when df2 -> infinity, Ftail(df1,df2,F) = chi2tail(df,x). Also, smaller F and Chi2 statistics have more comparable p-values.
> 
> While the discrepant p-values did not impact my conclusion, I'm assuming when one is stuck in a multiple test situation (100s of tests), when it pays to know if the probability is 1e-5 or 1e-10, that using the F distribution following ANOVA or regression is the right way to go?
> 
> Is there a good reason -test- after -margins- after -regress- or -ANOVA- reverts to use of the Chi2 distribution, or is it convenient because margins get used after other than regression?
> 
> -Dave

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index