Such non-convergence in these models is often a sign that the
effect of your continuous variable is not linear in the log(risk).
One solution is to allow that variable to have a non-linear effect.
I like (linear) splines for that, as you can still easily interpret
the coefficients. So in the example below for the model that
converges: the basline risk of union membership for someone from
the north with 0 wage is
*--------- begin example ------------
sysuse nlsw88, clear
gen baseline = 1
*does not converge
glm union south wage baseline, ///
link(log) family(binomial) ///
iter(50) eform nocons
*converges
mkspline w1 5 w2 10 w3 =wage
glm union south w? baseline, ///
link(log) family(binomial) ///
eform nocons
*----------- end example ------------
--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany
http://www.maartenbuis.nl
--------------------------
--- On Fri, 8/1/10, Rosenstock, Summer E. <[email protected]> wrote:
> From: Rosenstock, Summer E. <[email protected]>
> Subject: st: RE: convergence problem
> To: "[email protected]" <[email protected]>
> Date: Friday, 8 January, 2010, 11:33
> Thanks for the input on this! It
> looks like my estimates are coming in near the boundaries,
> which has shown to be a problem with log-binomial regression
> with continuous independent variables. THere are a couple of
> articles that document this. So, looks like I'll stick to OR
> on that particular variable with a foot note about
> convergence.
>
> Thanks again!
> Summer Rosenstock, MHS
> PhD Candidate
> Johns Hopkins University
> Bloomberg School of Public Health
> Global Disease Epidemiology and Control
> 018-3316915
> (301) 760-7468
> [email protected]
> ________________________________________
> From: Rosenstock, Summer E.
> Sent: Tuesday, January 05, 2010 5:08 PM
> To: [email protected]
> Subject: convergence problem
>
> I am using xtgee to analyze cluster data. I was using
> binomial family, logit link, independent correlation, robust
> variance, which was working just fine and providing ORs.
> However, what I really want is RR, so I changed the link to
> log and this is working just fine when the independent
> variables are not continuous. When my independent variables
> include a continuous variable, the model is failing to
> converge. I was not having this problem when I was using
> logit. Any thoughts on why these models won't converge when
> using the log link and how to fix it?
>
> Thanks!
> Summer
>
> *
> * 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/
