Because the model is underidentified. See [R] logit, model identification,
pages 252 - 257.
Scott Merryman
----- Original Message -----
From: "Benjamin Highton" <bhighton@ucdavis.edu>
To: <statalist@hsphsun2.harvard.edu>
Sent: Monday, July 15, 2002 2:53 PM
Subject: st: Why does STATA drop observations with MLE (logit/probit)?
> Hi all,
>
> I have a binary outcome variable (y) that I am modeling as a function of
> two independent variables (x1 and x2). x2 is also binary (0/1). The
> question I have relates to the fact that for all observations where the
> value of x2=1, the value of y=1. (When x2=0, y=0 for some observations
> and y=1 for others). When I attempt to estimate a logit model (logit y x1
x2),
> STATA drops the observations where x2=1 and estimates the model for the
> other observations. Specifically, the output says the following:
> note: x2~=0 predicts success perfectly
> x2 dropped and 36 obs not used
> The same thing happens with a probit model (probit y x1 x2). If willing
> to violate assumptions, I can use OLS (regress y x1 x2) and keep all the
> observations. Can someone explain to me why STATA drops the observations
> when I use logit or probit? Thanks, in advance. --Ben
>
> *
> * 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:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/