Re: st: Testing the extent of difference between two coefficients in the same model

[Maarten Buis <maartenlbuis@gmail.com>]

On Wed, Feb 1, 2012 at 12:55 PM, Erik Aadland wrote:
> When performing the one-sided z-test of one of my new coeff >= old coeff sets, I get a non-significant result from the Wald test in the first step.
> I.e. prob > chi2 = 0.15. This result suggests that I cannot reject the equality hypothesis for this set of coefficients. However, in the second step, I get a p-value that is larger than .05 suggesting that I cannot reject the hypothesis that new coeff >= old coeff.

You need to say exactly what you mean with first and second step. This
means the _exact_ Stata code that you used.

> Do these results simply suggest that the I cannot rule out that the coefficients are equal? Should I pay equal attention to both steps in the test, or do I focus on the results obtained in the second step?

Both p-values are larger than .05, so you cannot reject whatever
null-hypothesis you are testing (at the customary but arbitrary 5%
significance level).  The real trick with statistical testing is to
know exactly what the null hypothesis of your test is. You do not give
equal weight to different test, you focus on the test that tests the
null hypothesis of interest. What the null hypothesis of interest is,
depends on what you find interesting...

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany


http://www.maartenbuis.nl
--------------------------

*
*   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/