Re: st: Weighting on Sub-samples of Complex Survey Data and Specifying Correlation for PA Models

[Stas Kolenikov <skolenik@gmail.com>]

On Thu, May 6, 2010 at 10:37 AM, Ryan McCann
<rmccann@keybridgeresearch.com> wrote:
> Secondly, the random effects model would seem more appropriate than fixed
> effects because most of the variation in the sample is between as opposed to
> within  (the panel is not that wide to begin with (average time series for
> an individual is only 2.5 periods).  STATA does not allow for the use of
> weights with RE

You can try -gllamm- with that. Actually I would believe it would work
with the whole data set of 24000. It does not care about missing data
in any particular wave. But the only "correlation" it implicitly
supports is "exchangeable".

I will have to second Austin in his concerns regarding endogeneity
issues -- if you don't resolve them, you get biases estimates that are
not meaningful and interpretable. In -gllamm-, you could try to model
some of these issues by specifying simultaneous models for all
endogenous variables -- I have been working with a model of panel
attrition, for instance (although it takes at least a week for each
specification to converge, so I am not making that much progress). Of
course these models are only good for something if you correctly
modeled these causal dependencies.

-- 
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/