The FAQ request that you show the commands you used to produce your
models and results from those statements. Please do so. Otherwise we
can only waste our time by speculating on what is happening.
The log-likelihood for the "null model" (no covariates) is missing.
What happens if you fit that?
e.g.
sysuse auto
logistic foreign
-S.
On Mon, Dec 21, 2009 at 10:23 AM, Ekrem Kalkan <[email protected]> wrote:
> Dear Stata users,
>
> I estimate restricted and unrestricted 3SLS models. I compare these
> models with LR test. However, the loglikelihood value of the
> restricted model is higher than that of the unrestricted model. To my
> knowledge, this is not expected in econometric theory. As the result,
> LR-statistics becomes a negative value. Please see my results below.
> Can you please comment on this? Thank you.
>
> . lrtest rSLS3_85 uSLS3_85, stats
>
> Likelihood-ratio test
> LR chi2(65) = -1311.29
> (Assumption: rSLS3_85 nested in uSLS3_85) Prob > chi2 = 1.0000
>
> -----------------------------------------------------------------------------
> Model | Obs ll(null) ll(model) df AIC BIC
> -------------+---------------------------------------------------------------
> rSLS3_85 | 864 . 34853.08 541 -68624.17 -66048.16
> uSLS3_85 | 864 . 34197.44 606 -67182.87 -64297.36
>
>
> Ekrem Kalkan
> *
> * 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/
>
--
Steven Samuels
[email protected]
18 Cantine's Island
Saugerties NY 12477
USA
845-246-0774
*
* 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/
