Wiji Arulampalam wrote that he was having difficulty getting results from -xtprobit-
to agree with those produced by Limdep. Weihua Guan identified the problem as
having to do with the inability of Hermite quadrature to give trustworthy estimates of
the integral, and this would apply to the results of both -xtprobit- and those of
Limdep, since they both use similar algorithms.
Would adaptive quadrature prove more reliable in this case? For example,
-gllamm sue lague, i(ind1) family(binomial) link(probit) adapt-. Although the
samples size is relatively large (10326 observations in 2139 clusters, with a range of
one to six observations per cluster), there is only a single random effect (in addition
to the residual) to estimate, so -gllamm- shouldn't take too long to converge.
Another method of quadrature, e.g., -gllamm . . . ip(f)-, might also be considered.
Joseph Coveney
*
* 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/
