I'm writing because I have a quick (na�ve) technical/statistical
question I was hoping someone could answer. I have an analyses using
xtlogit with a random effects specification. I know the test of rho at
the bottom of the output is for whether the contribution of the panel
variance is significantly greater than 0. It's been suggested that I
perform a breusch-pagan test on this, which appears to be much the same
thing, conceptually. It seems reasonable because xtlogit saves the panel
level standard deviation in e(sigma_u) which is what you would need to
calculate the breusch-pagan test. Breusch-Pagan in xttest0 is not
implemented after MLE commands such as xtlogit (or xtreg, mle) however,
just after the GLS version. My experience with STATA is that if it's not
there, there's generally a good reason. Could someone explain why it's
not there? Does it concern the variance estimation?
John Sandberg
Assistant Professor
Department of Sociology
McGill University
Stephen Leacock Building Room 712
855 Sherbrooke Street West
Montreal, Quebec, H3A 2T7, Canada
[email protected]
*
* 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/
