Re: st: testing difference in partial coefficients with -test-

[David Greenberg <dg4@nyu.edu>]

It is usually a dubious enterprise to test for differences between
standardized coefficients whether in a multilevel model or not.
Differences in standardized coefficients can arise because of
differences in the underlying unstandardized coefficients or because
of differences in the variances of the two predictors. The finding of
significant differences or of non-significance is consequently
difficult to interpret. David Greenberg, Sociology Department, NYU

On Fri, Feb 24, 2012 at 2:10 AM, Martin Engeler <martin.engeler@unisg.ch> wrote:
> Hi all!
>
> I try to test the difference between two coefficients from the same sample of a multilevel model (xtmixed) by using the -test- command. (postestimation command test coeffA = coeffB) However, I am unsure how to interpret the resulting chi2 outcomes properly.
>
> I want to test whether the two standardized coefficients are significantly different, and thus whether one is significantly larger than the other. Am I right to conclude that the coefficients are different and one is larger than the other if the resulting Prob Chi2 is smaller than 0.05?
>
> Many thanks for considering my question. Best, Martin
> *
> *   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/

*
*   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/