The key principle is that -glm- (or GLMs) is (or
are) based on assumptions about the mean response, not the
response data. Thus some of the data can be
zeros even if you use a log link. Whether the link
and family chosen are good for the problem and the data
is naturally an important question, but one that is empirical,
not theoretical. Or so I understand the matter.
Nick
[email protected]
David Jacobs
> I thought that count estimators would only handle positive
> integers. At least that's what the Stata reference books say. Is
> GLM an exception?
*
* 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/
