Hi Statalisters,
Is there any kind of likelihood ratio (or equivalent) test that
assesses the overall fit of an IV model (2sls, liml...)? I had figured
no because an IV's overall fit can be strange, have negative R^2 etc.,
but I couldn't find a paper saying so and figured I'd confirm.
If such a test existed, it might compare models like:
i: ivreg y (w1 = z1) x
ii: ivreg y (w1 w2 = z1 z2) x
and say, e.g., that model ii is probably a better overall fit so I
should include w2 and z2 in my regressions.
Thanks for the help!
Cory
*
* 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/
