Re: st: xtmixed with vce robust or cluster robust

[John Antonakis <john.antonakis@unil.ch>]

Hi Stas:

Thanks for the information, though I am not quite following (see Mark's 
post too).

I also ran this question through the Stata folks, who mentioned that 
they will probably add these options in future versions, though the SEs 
in xtmixed (or xtreg with mle) are consistent if the usual assumptions 
for mle hold.

For the time being one could use the user-written command -gllamm- 
(available through SSC and written by Sophia Rabe-Hesketh) which can 
handle robust and cluster robust vce in a multilevel model.

Regards,
J.

____________________________________________________

Prof. John Antonakis
Associate Dean Faculty of Business and Economics
University of Lausanne
Internef #618
CH-1015 Lausanne-Dorigny
Switzerland

Tel ++41 (0)21 692-3438
Fax ++41 (0)21 692-3305

Faculty page:
http://www.hec.unil.ch/people/jantonakis&cl=en

Personal page:
http://www.hec.unil.ch/jantonakis
____________________________________________________



On 31.08.2009 17:30, Stas Kolenikov wrote:
> The deep theoretical reason for lack of -robust- for panel data is
> that meaningful contributions to the objective function/likelihood are
> due to panels, rather than individual observations. Arguably one could
> wrap -vce(cluster)- on top of that if panels are nested in these
> clusters, but from Stata's internal perspective of -_robust-
> calculations, that's a very odd thing to do. Bobby G would probably
> give a more meaningful answer as the main author of -xtmixed-. I've
> heard the argument that "SAS does PROC MIXED with unequal panel
> weights" -- the lack of such estimation procedure is essentially the
> same panel level argument; see also
> http://www.stata.com/support/faqs/stat/xtweight.html. Bobby had a good
> answer to that which I don't quite remember. I think it went along the
> lines of James Hardin's FAQ that it is not quite clear what the
> interpretation of those weights is.
>
> On Sat, Aug 29, 2009 at 10:47 AM, John 
Antonakis<john.antonakis@unil.ch> wrote:
>> Does anyone know why xtmixed does not allow a robust or cluster 
robust vce?
>> Ditto for xtreg with maximum likelihood estimation. Does the 
procedure that
>> Stata implements assures sandwich estimation?
>

*
*   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/