Re: st: Fitting a model when the outcome is a proportion - glm versus logistic command

[Richard Goldstein <richgold@ix.netcom.com>]

-logistic- is not doing what you think it is; logistic automatically
forces everything that is not 0 to be non-zero so that the outcome is
binary; -glm- (make sure to use "vce(robust)" also) does not do this;
here is a quote from the help file for -logistic-: "logistic fits a
logistic regression model of depvar on indepvars, where depvar is a 0/1
variable (or, more precisely, a 0/non-0 variable)."

Rich

On 3/21/14, 10:05 AM, anny fenton wrote:
> Dear All,
> 
> 
> When fitting an outcome that is a proportion, I know the typical
> approach is to follow Papke and Wooldridge (1996) and use glm with
> family(binomial), link(logic). However, I don't understand why one
> would use glm instead of the logistic command why the two commands
> would produce different fit statistics and coefficients (as they have
> with my own results).
> 
> 
> Thank you for any insight in advance,
> 
> 
> Anny
> 
> 
> Reference
> 
> Papke, L. E. and J. Wooldridge. Econometric methods for fractional
> response variables with an application to 401(k) plan participation
> rates. Journal of Applied Econometrics 11: 619-632.
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/