I'm a bit confused about the use of mlvecsum to compute the gradient
in a method d1 or d2 evaluator. According to the manual and the Gould
book, the command -mlvecsum- requires that the model satsify the
linear restriction to properly compute the gradient. However, my
understanding is that d0, d1 and d2 are useful precisely when the
linear restriction doesn't apply, e.g. in a random effects model on
panel data.
So, if one's model does not meet the linear restriction and, for that
reason, one wishes to use d1 or d2, how would you use mlvecsum? The
random effects regression model example in Gould uses mlvecsum, but I
don't see how this would work given the requirement I stated above.
Thanks,
Rachel
*
* 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/