st: -mi impute chained (logit)- goes through many iterations without change in log likelihood...

[Anand Segar <anand.segar@chch.ox.ac.uk>]

Hi,

Please excuse me, I am new and this is the first time using the Statlist. 

I’m using Stata/IC 13.1 for Mac (64-bit Intel) Revision 15 Jan 2014

I am running an imputation model using -mi impute chained-

	mi impute chained (regress) age_surv weight height (ologit) sport work car_cat ///
	(logit) ldh_main gender famhx  (mlogit) smoke_simple= unit ra oa op, ///
	add(50) rseed(123456789) noisily augment burnin(25) ///
	savetrace(extrace_ldh1_pre.dta, replace) 

but it keeps getting stuck on the (logit) part. Within (logit), the model goes through many iterations and the log likelihood does not change. I have let it go up to about 5000 iterations before quitting.

This is what I see and it can go up to iteration 6000
	Iteration 213: log likelihood = -824.46062  (backed up)
	Iteration 214: log likelihood = -824.46062  (backed up)
	Iteration 215: log likelihood = -824.46062  (backed up)
	Iteration 216: log likelihood = -824.46062  (backed up)

It usually happens on the 8th iteration of m=1 but if I move the logit around within the model I can get to about m=3 before this happens.

Just from trial and error, I have dropped off individual variables and it seems maybe gender is causing it. Without gender I have let it get up to m=10 without this issue.

I have run a previous MI model with more variables for another question without this problem.

I am pulling my hair out and would greatly appreciate any help. 

Let me know if I can provide any further information.  

Thanks so much!

Anand


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/