This is a repeat posting. The original
got no replies. One possible reason for
that is that it is near impossible to
say what is the "right answer", absent
any idea of your data or your scientific
problem. Different families might be
appropriate depending on the data and
on your ideas about generative processes.
Nick
[email protected]
Rijo John
> I am estimating a Fractional logit model. Stata help says
> using
>
> glm y x1 x2...xk, family(binomial) link(logit)
>
> is same as "logistic" command and the later is infact
> preferred. Because
> family(binomial) means that y has the Bernoully distribution
> with values 0
> and 1 only. But my y variable is a continuous variable (a
> fraction) taking
> any values between 0 and 1. In this case which family and
> link should I be
> using?
>
> Is "family(gaussian) link(logit)" the right one?
> Also what is the criteria of deciding which scale parameter
> to be used?
> I mean whether scale(1) or scale(x2) etc.
*
* 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/