Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Re: hypothesis tests


From   "Dr. Stephen Owusu-Ansah" <sowusuansah@rgv.rr.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Re: hypothesis tests
Date   Tue, 30 Dec 2003 11:02:21 -0600

Hi Kit

Would you consider a sample size of 168 small, and as such any
inferences drawn on the basis of LM or LLR instead of Wald
could be wrong?

Stephen


Christopher F Baum wrote:
On Dec 30, 2003, at 2:33 AM, Richard wrote:

Thanks Kit.  Getting back to the original question, Stephen had asked

If I use the "test" command after estimation, the results reported by
Stata is in F-distribution, but I want the results to be reported in
chi-square distribution with its associated p-value.

If it is simply a matter of him preferring to report chi-squares rather
than F values, then rather than go through the tedium of -lrtest-, could he
just take the F value reported by test, and compute

chi-squared = (numerator degrees of freedom) * F

That seems to be the implication of this FAQ from Stata on chi-square versus F:

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

Indeed, that simple solution answers Stephen's original question. Your earlier response proposed switching from Wald to likelihood ratio tests. Asymptotically, all three tests are the same. In small samples, one may draw different inferences, depending on whether the Wald, LM or LLR tests are used.

Kit

*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   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   |   What's new   |   Site index