Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Predicted probabilities after xtlogit


From   Jeph Herrin <junk@spandrel.net>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Predicted probabilities after xtlogit
Date   Mon, 04 May 2009 10:27:59 -0400

What you are doing is correct. The predicted probabilities are going
to = 1 rarely, if ever. As you say, they are probabilities. You may
find the invlogit function handy here:

  predict paction
  gen paction2=invlogit(paction)

or even

  predict paction2, pr0

to get the probabilities directly. Unfortunately, whichever way you
go, -xtlogit- will only let you get probabilities conditioned on the
random effects being zero, ie, based on the fixed portion of the model.

hope this helps,
Jeph


bparsons wrote:
Dear Statalist,

How can I generate predicted probabilities after using the xtlogit
command? I ran my model with action as my DV then .predict paction .gen paction2 = exp(paction)/(1+exp(paction))

As I understand it, the paction2 column should be my predicted
probabilities. However, my results don't make sense with the data I'm
using. My action variable is dichotomous and the predicted values are
not equal to 1, even when the action variable is 1.

I wasn't able to make sense of the responses to this thread: http://www.stata.com/statalist/archive/2003-09/msg00543.html

... but if that is the correct way to do it and someone can explain
it in a different way, I would greatly appreciate the help.

Thank you, Barbara Fenton






* *   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/

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index