Re: st: observations differ with lrtest because of interaction terms

[Richard Williams <richardwilliams.ndu@gmail.com>]

At 04:13 PM 5/18/2010, Meredith Carlson wrote:
> Hello Stata users,
>
> I'm a Stata rookie, so please bear with me... I'm having problems 
> when I run lrtest with logistic regression to test for interaction 
> between wealth index and region variables.  The number of 
> observations differ, not because of missing values, but because of 
> the interaction terms.  The model including interaction has fewer 
> observations because some have been dropped because of this note...
>
> note: _IsesXreg_3_9 != 0 predicts failure perfectly
>       _IsesXreg_3_9 dropped and 4 obs not used
>
>
> What exactly does this note mean?  What can I do to correct the 
> number of observations so lrtest will run?

This occurs when, for example, every time x equals 0 y also equals 
0.  You may, of course, want to reconsider your model, as you may be 
spreading the data too thin.  Sometimes -exlogistic- can address the 
problem, at least if the sample isn't too big.  There are some 
routines out there that offer different approaches but off the top of 
my head I can't remember what they are.


-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/