Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: I am not sure what is the null for the sargan-hansen statistic(xtoverid)


From   "Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: I am not sure what is the null for the sargan-hansen statistic(xtoverid)
Date   Wed, 22 Aug 2007 13:22:16 +0100

Jorge,

> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu 
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of 
> Jorge Rivera
> Sent: 22 August 2007 00:40
> To: statalist@hsphsun2.harvard.edu
> Subject: st: I am not sure what is the null for the 
> sargan-hansen statistic(xtoverid)
> 
> Prof. Mark E. Schaffer
> >
> > Thank you for your response to my question, it took me a while to 
> > install xtoverid. The problem for me was that we have restricted 
> > access to the computers in our lab and I had to call the people in 
> > charge of the lab to enable the changes to take place. I was able to

> > run the xtoverid but I am not sure what is the null for the 
> > sargan-hansen statistic. Thank you for your time

Have a look at the help file for xtoverid - there's a longish paragraph
discussing the interpretation of a fixed-vs-random effects test as an
overid test.

One way to anwer your question is the standard FE-vs-RE formulation: the
null is the same null as the standard Hausman fixed vs. random effects,
i.e., both estimators are consistent.  Rejection of the null is then
interpreted to mean that the random effects estimator is not consistent.

Another way to put it (and why it's an overid test) is that the random
effects estimator uses more orthogonality conditions than the fixed
effects estimator; namely, that the group means are uncorrelated with
the idiosyncratic error e_ij.  These extra orthogonality conditions are
responsible for the increased efficiency of the random effects estimator
vs the fixed effects estimator.  However, the extra orthogonality
conditions might not be valid.  The null is that they are valid, and
rejection of the null suggests that they are not (and hence the random
effects estimator that uses them is inconsistent).

Hope this helps.

Cheers,
Mark

> >
> > Kind Regards,
> >
> > Jorge Rivera
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
> 

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   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   |   What's new   |   Site index