[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: offset/exposure term in poisson analysis
As a stata beginner I would recommend the book by Hardin and Hilbe titled
*Generalized Linear Models and Extensions* published by Stata Press.
Hardin and Hilbe write that if one uses the exposure subcommand one uses
the actual value for the exposure, which will be (natural) logged.
However, if one uses the offset subcommand one should take the natural log
of the exposure variable and put the new variable in the parentheses.
For example, if my exposure variable is pop, a count of the people in the
population, my code would be something like this:
offset(lnpop) (assuming I computed lnpop to equal ln(pop))
The Stata manuals also state that it is often easier to use the commands
for the exact model if they exist in Stata rather than using the glm
module. Hence, you might try the poisson command, or if you have
over-dispersion, nbreg (negative binomial regression model), rather than
glm. Not that I know anything about your substantive research question or
other factors that may have led you to glm.
Hope this helps.
On Mon, 24 Jun 2002, Anne Knol wrote:
> I'm trying to model the rate of a certain disease (amount of cases divided by population). Am I allowed to use an exposure/offset term for the population when modelling the rate?
> Furthermore, when executing the poisson command I can use an exposure term but I can't use offset (while the help file claims that these options should give the same result). Stata comments:
> exposure = exp(pop) overflows;
> could not estimate constant-only model
> use exposure() option for exposure = pop
> When I'm using the glm command (with family=poisson) in order to be able to allow for the overdispersion of my data, I'm not allowed to use an exposure term and when trying to use offset, I get the following error message:
> initial values not feasible r(1400)
> Any ideas how to solve this problem so I can include include an exposure or offset term when using glm?
> Thanks (and I know, I'm a stata beginner, so I'm sorry if these are stupid questions...)
> Anne Knol
> Research Fellow
> Dept. Public Health
> Wellington School of Medicine,
> PO BOX 7343, Mein St
> Wellington, New Zealand
> Sign-up for your own FREE Personalized E-mail at Mail.com
> Save up to $160 by signing up for NetZero Platinum Internet service.
> * 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/
* For searches and help try: