Wolfgang Meyer <[email protected]> asks:
> I'm using Stata 8.0 and I'm trying to estimate a cross-section Probit model
> over several countries.
> I've encountered some problems using the command 'glm' with the options
> family(binomial) link(probit) nwest(nw).
> I get an error message saying 'insufficient observations'.
It's hard to say what is going wrong here without having knowing more about
the data. Newey-West s.e.'s require the specification of a time -t()-
variable. What do the values in that variable look like?
In any case, if Wolfgang wishes to email me privately with the data and his
.do file, I can take a closer look.
> The command works fine with
> glm ... , family(binomial) link(probit) nwest()
> which gives the var-cov matrix using the OIM. However, I'd prefer the
> Newey-West correction.
The above is the result of incomplete error trapping on our part.
The -nwest()- specification is ignored since it defines local macro -nwest- to
be empty. Since the -nwest()- option requires a specification of the type of
weights, we'll fix -glm- so that the above, rather than just ignore the
option, gives the appropriate error message instead. The fix will be made
available in a future ado update.
> The other way to estimate the model using the 'binreg' command doesn't use
> all available observations. What's going wrong?
Again, I'll be glad to take a closer look if Wolfgang emails me privately
with his data and the command he is issuing.
--Bobby
[email protected]
*
* 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/
