[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Grouped Logit question
Why would you want to use -glogit- rather than -glm- with the family(binomial) and link(logit) and the proportional denominator listed as indicated in the help file for -glm-? The fit and residual statistics are appropriately adjusted, and you can use a variety of estimating alternatives.
> Revealing my ignorance of the mathematical proofs, I'm struggling to
> get a grip on exactly what the Grouped Logit function does and how it
> should be used appropriately. Specifically I wonder about the use of
> count vs. proportion variables on both the left and right hand side.
> the command is (of course)
> glogit depvar denominator list_of_covariates
> Also, I assume that what this does (in layman terms) is essentially
> weight each row by the relative size of that observation compared to
> the size of the other observations. In my case I have observations by
> county and running the regression for the full US... i.e. I
> have some
> 3200 observations representing all use households.
> Thanks in advance,
* For searches and help try: