st: Evaluating likelihood ratio tests at specified parameter values

[David Kling <klingd@reed.edu>]

I'm trying to implement a method for conducting likelihood ratio tests 
on models fit using data produced via multiple imputation.  The method 
is described in detail in Meng and Rubin (1992) and reviewed in Schafer 
(1997).  The only difficult part (from my point of view) of implementing 
the procedure is evaluating a likelihood ratio test at the mean values 
of the model parameters (the mean of the m parameter vectors, one from 
each of the m data sets generated during multiple imputation) for each 
of the m data sets.  I'm interested in learning whether it is possible 
to evaluate the log-likelihood of a model (specifically, a glm with a 
gaussian link) at user-specified values of the parameters without 
resorting to writing my own program to do the job.

Thanks!

David Kling

References:

Meng, X.L. and Donald Rubin, "Performing Likelihood Ratio Tests with 
Multiply-Imputed Data Sets" Biometrika 79 pp 103-111 (1992).

Schafer, J.L., Analysis of Incomplete Multivariate Data.  New York, NY: 
Chapman & Hall (1997)

*
*   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/