st: Re: xt regression with large n and small t (Was "How to deal with r(909) "op. sys. refuses to provide memory???")

["Seed, Paul" <paul.seed@kcl.ac.uk>]

Dear Statalist, 
Dear Gordon Hughes, 

As someone who often has panel data with many more cases than time points
(anything from 10 times as many to 1,000 times as many), 
I would be very glad of some pointers to the "warnings" and "much better ways".


Many thanks 

Paul T Seed, Senior Lecturer in Medical Statistics, 
Division of Women's Health, King's College London
Women's Health Academic Centre, King's Health Partners 
(+44) (0) 20 7188 3642.

> From: Gordon Hughes <G.A.Hughes@ed.ac.uk>
> To: statalist@hsphsun2.harvard.edu
> Cc: 
> Sent: Sunday, 9 December 2012, 12:08
> Subject: Re: st: How to deal with r(909) "op. sys. refuses to provide  memory"??? xtgls
>
> You should also think very hard about whether it makes sense to use xtgls for this purpose.  There are all sorts of warnings, in Stata documentation and the  
> literature, about the reliability of xtgls in these circumstances.  Technical problems of the kind that you encountered are all too often an indication of a 
> poorly conceived model.
>
> With very large N and rather small T you cannot possibly obtain sensible estimates of either within panel autocorrelation or across panel correlation.  As a 
> consequence all you can estimate is a model with heteroskedastic panel errors.  There are much better ways in Stata of dealing with this than by using xtgls.
>
> Regards
>
> Gordon Hughes
> g.a.hughes@ed.ac.uk


*
*   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/