Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: bug in -xtmelogit-?


From   "Roberto G. Gutierrez" (StataCorp) <rgutierrez@stata.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: bug in -xtmelogit-?
Date   Fri, 04 Sep 2009 14:56:06 -0500

Jeph Herrin <junk@spandrel.net> asks:

> By accident I coded the following

>    xtmelogit depvar var1 || groupid: var2, cov(unstructured) var

> Note that var2 is not in the equation, but I have specified it as
> a random coefficient. After puzzling over my results for some time
> I discovered my error, but I am surprised that -xtmelogit- did not
> discover it for me. What exactly was it estimated for the variance
> of the coefficient on -var2-? Is this a bug?

Not a bug, but instead something you probably don't want to do statistically.
By leaving -var2- out of the first equation, you estimated the variance across
groups (-groupid-) of the coefficient on -var2-, but you imposed the
constraint that the mean across groups is exactly zero.  It is akin to leaving
main-effects terms out of a model with interactions.

--Bobby
rgutierrez@stata.com
*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index