Re: st: logistic and logit and test

[Richard Williams <Richard.A.Williams.5@nd.edu>]

At 06:34 PM 4/13/2004 -0500, David Airey wrote:
> All this has made me wonder whether the OR test is testing whether the OR 
> is different from "1", whereas the test for the logit coefficient really 
> is testing the coefficient being different from 0? Since exp(0) = 1, this 
> suggests as much. But my question is that use of "test x" after either 
> command produces the statement that x = 0 is the test.
>
> Why am I confused?
To add to your confusion, note that when odds ratios are reported, z does 
NOT equal odds ratio/standard error.  But, when coefficients are reported, 
z does equal coef/standard error.

Anyway, regardless of what is reported, the -test- command in your case is 
testing whether the coefficient for x significantly differs from 0.  This 
is equivalent to testing whether the odds ratio = 1.

Incidentally, the -test- command is not considered the optimal way to test 
hypotheses in logistic regression.  (If you are only testing whether a 
single parameter = 0, it also doesn't accomplish much, because the z value 
for the coefficient gives you the same info.) The -lrtest- command (which 
also requires the use of the -estimates store- command) is considered 
better.  pp. 3 and 4 of the following handout give examples:

http://www.nd.edu/~rwilliam/xsoc593/lectures/l24.pdf



-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
FAX:    (574)288-4373
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW (personal):    http://www.nd.edu/~rwilliam
WWW (department):    http://www.nd.edu/~soc

*
*   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/