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 on April 23, and its replacement, statalist.org is already up and running.


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

Re: st: estat overid & sample weights


From   Stas Kolenikov <skolenik@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: estat overid & sample weights
Date   Thu, 30 Dec 2010 11:00:37 -0600

On Thu, Dec 30, 2010 at 10:31 AM, C.J. Krizan <c_j_krizan@yahoo.com> wrote:
> Can anyone tell me why the OVERID option on IVREGRESS needs the FORCEWEIGHTS option when using data with sample weights.  Also, how do I determine if my results are valid afterwards?
>
> There is a blurb in the documentation that the reported critical values may be inappropriate with weighted data and I'm trying to determine if I can trust it in my case.

To put in very bluntly, if a chi-square statistic is obtained as the
value of the objective function at the minimum (or is a scaled version
of that value) when the data are i.i.d., it will become a sum of
weighted chi-squares with complex survey data. So likelihood ratios
are affected, and the minima of GMM objective functions are affected.
The chi-square test statistic resulting from parameter estimates, such
as Wald tests, are OK, though. There's a long line of work in survey
statistics literature on these issues starting from Rao & Scott (1984)
paper (see full reference in -help svy: tabulate twoway-). The test
statistic is unlikely to be valid, I am afraid.

Generally, economists and econometricians does not give a damn about
complex survey data features, and I have not yet come across a good
reference discussing the properties of econometric estimators, even as
common as IV, in complex survey data context.

Using CAPITALS usually means you are SHOUTING. If you are used to Some
Alternative Software where it was customary to shout at it to get any
work done, then you probably have already figured out that Stata
understands more moderate tone, and is case sensitive. So strictly
speaking there are no such commands and options that you mentioned.

-- 
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/


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