Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Alternatives to XTLOGIT,RE?


From   "Man Woman" <lal789@hotmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   st: Alternatives to XTLOGIT,RE?
Date   Tue, 18 Nov 2003 17:59:29 +0000

Dear Statalist,

I have panel-data consisting of 3200 individuals with 4 observations each.
2 of these observations are taken before a treatment, and the other 2 after
the treatment.  The dependent variable is bin(0,1).
I wish to calculate the odds ratio before vs. after treatment and I fitted
the following random effects model:

iis idno                                                  /* idno=person
identifier (i=1....3200) */
tis obsno                                               /* obsno=observation
number (j=1...4) */
xtlogit depvar treat, re or                        /* treat=treatment before
and after (t=0,1) */

I used the quadchk command to check whether the quadrature approximation
used for xtlogit,re
was reliable, but for this data the approximation was poor.  The panel-level
variance component was high (rho=0.65) indicating that the 4 observations
were highly correlated within each individual, so I thought this may be the
reason why.

Can anyone suggest any alternative methods that I can use to fit this model
in STATA?

Laurence

_________________________________________________________________
Use MSN Messenger to send music and pics to your friends
http://www.msn.co.uk/messenger

*
*   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/



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