Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


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

Re: st: xtnbreg - robusteness check and model relevance


From   "JVerkuilen (Gmail)" <jvverkuilen@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: xtnbreg - robusteness check and model relevance
Date   Fri, 18 Jan 2013 09:19:10 -0500

On Fri, Jan 18, 2013 at 5:16 AM, Mário Marques <mmarques@eeg.uminho.pt> wrote:
>
> Thank you for your advising lr test, I have performed that,as also
> Andre suggested, and it works.

Yes it does.

> Regarding the observations dropped because zero outcomes, it was not a
> huge concern however it represents about 30% of country-pairs. If I
> perform xtnbreg, re, an excerpt of the output is as follows. Do you
> think it is plausible and convincing use a random effects instead?

Well I'm not an econometrician so I don't get the imperative to use
fixed effects. ;) I don't believe the super-population/sampling
argument behind random effects models and instead view them as
penalized likelihood estimators so I would generally use fixed effects
as a way to check the random effects model. FE estimators are pretty
costly in many circumstances, though, as is clearly happening to you.
I wonder whether you really have zero inflation going on, too.

But that aside, the answer is I don't really know without having a
clear idea of your data.



> I take the oportunity to ask you whether there is any way, under
> xtnbreg, fe,re model, to obtain clustered/robust standard errors?

Clustering is built into the model. As to whether it's possible to
obtain robust, the answer is I don't know. If it's not supported

*
*   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 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index