Statalist


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

st: Re: xtmepoisson


From   Simon Moore <scmoore.lists@googlemail.com>
To   statalist@hsphsun2.harvard.edu
Subject   st: Re: xtmepoisson
Date   Fri, 14 Aug 2009 17:34:27 +0100

Martin,

Many thanks for the stata.com presentations - I had not seen these

Michael E Roettger wrote:

Have you examined the distribution of the dependent variable?

If xtmepoisson fails to converge while xtpoisson converges, you make
have a distribution which does not fit the standard Poisson
distribution.   Rabe-Hesketh and Skrondal (2008) note that xtpoisson
adjusts for overdispersion due to the way the individual-level random
effect is specified;  xtpoisson also assumes that the random effect
has a log-gamma distribution.

If you have overdispersion, you might be able to use the xtgee
command, specifying that the dependent variable is a negative
binomial distribution.  Dummy variables may also be added to control
for year effects.

I remain at the "Refining starting values:" so the delay does not seem to be at convergence. Although the depvar is a little zero inflated si I'll take a look at xtgee.

Many thanks
Simon

*
*   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