Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


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

Re: st: use xtnbreg, fe or xtpoisson, fe vce(r)?


From   "JVerkuilen (Gmail)" <jvverkuilen@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: use xtnbreg, fe or xtpoisson, fe vce(r)?
Date   Thu, 18 Oct 2012 00:03:55 -0400

On Wed, Oct 17, 2012 at 5:28 PM, Vestal, Alex - COB
<Alex.Vestal@bus.oregonstate.edu> wrote:
> Using unbalanced panel data, I am modeling a count DV with overdispersion (patent counts).  I recently submitted a study using xtnbreg and a reviewer commented: "the best practice is to use xtpoisson with robust SEs that resolves the usual reasons to favor xtnbreg."
> I had been using xtnbreg because of its ability to handle overdispersion.  Is xtpoisson vce(robust) robust to overdispersion?  I'm looking for rationale as to which approach I should be using.  Any guidance would be greatly appreciated!
>

It's not obvious to me which one is best in this case. -xtnbreg- is
using a parameter whereas vce(robust) is avoiding using a parameter.
But I suspect the resulting inferences will end up being quite
similar. So perhaps throw the reviewer a bone by doing what he wants
and footnoting the other way, or vice versa.

As to the issue of the sources of overdispersion, I've often
encountered in practice the sad fact that any reasonable model will be
overdispersed due to lack of reasonable covariates so often you simply
have to fit what you have.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


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