Re: st: lrtest of nested logistic models with vce(cluster) specified

[Stas Kolenikov <skolenik@gmail.com>]

The likelihood ratio test has an odd distribution (a sum of weighted
chi-squares). You should use Wald test (and that was the recommendation
given in the other thread you started looking at):

test ind_var2

Of course this will be the same test that is shown in the default output.

On Fri, Mar 5, 2010 at 5:02 PM, Lily Yor <lilyyor1@gmail.com> wrote:

> Hi, I have two nested binary logistic regression models, both of which have
> vce(cluster) specified to account for intraclass clustering at the
> classroom
> level.  For example, my Stata commands look like this in simplified form:
>
> logit [dep var] [ind_var1], vce(cluster class_id)
> est store A
>
> logit [dep var] [ind_var1 ind_var2], vce(cluster class_id)
> est store B
>
> I would like to test whether model B fits the data significantly better
> than
> model A, so I ran the lrtest, but Stata gives me this error:
>
> . lrtest A
> LR test likely invalid for models with robust vce
> How do I test the fit of these nested models if I cannot use lrtest?  I
> found a similar question posted by another user earlier in the archives but
> it was never answered.  I would much appreciate any and all tips.
>
> Best,
> Lily
>
> *
> *   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/




-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.

*
*   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/