st: Re: firthlogit

["Joseph Coveney" <jcoveney@bigplanet.com>]

Nyasha Tirivayi wrote:

I have a problem of quasi-complete separation in my logit model for
[ropensity score matching. I tried using firthlogit instead and I get
the following error "initial vector: matrix must be dimension 27".

In PSM is it worthwile to correct for separation if it occurs? Since I
am unable to solve this problem, should instead remove the problematic
variables?

--------------------------------------------------------------------------------

If you're having a problem with -firthlogit-, you can contact me privately with 
the details at the e-mail address given in the its help file.  It will be 
helpful to provide as much of the code and dataset as you can, so that I can 
reproduce the error and debug the command or improve the help file.

Even if you are eventually successful with -firthlogit-, I don't understand 
how you can hope to achieve balance when there is separation.

Joseph Coveney


*
*   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/