Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


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

st: cannot use cluster or robust option of gllamm on a server


From   "Dai, Hengchen" <[email protected]>
To   "[email protected]" <[email protected]>
Subject   st: cannot use cluster or robust option of gllamm on a server
Date   Mon, 20 Jan 2014 17:25:59 +0000

Hi Statalist members,

I am running a Stata/SE 12.1 on a Unix server.  I am using gllamm (version 2.3.20 SRH 7 Sept 2011) to estimate a two-level random intercept logistic regression model. I have a problem with using the cluster or robust option. I am looking forward to your suggestions on addressing this issue. 

Specifically, my data have three levels. Hospital site is the highest level, doctors are nested within hospitals, and each doctor has multiple observations. 

My code is: gllamm Y X, i(doctor) family(binom) link(logit) adapt cluster(site)
 
This code works very well when I tried it on a small subset of my data using Stata on my desktop. Since I have a very large data set, I have to analyze data on a server. But Stata on the server cannot estimate this model when I used the exactly same subset of my data. Stata reported the following errors:
option site not allowed
r(198);

I also tried "gllamm, cluster(site)" after I estimated the model as the manual suggested. However, Stata reported: 
invalid syntax
r(198);

I had a similar problem when using robust standard errors. I have made sure that the server has installed gllarob.ado file. My code is "gllamm Y X, i(doctors) family(binom) link(logit) adapt robust". Stata reported:
calling gllarob
invalid syntax
r(198);

Has any of you encountered a similar problem? Is there anything that I should note when I run gllamm on a server?  

I also tried a three-level random intercept model to take into account of intraclass correlation within hospitals. However, this three-level model takes a long time to run even for a small subset of my data. So I am not sure whether it would be feasible to use a three-level model as an alternative to the cluster option. 

I very much appreciate your insights!
Hengchen

--------
Hengchen Dai
Doctoral Candidate
Operations and Information Management Department
The Wharton School, University of Pennsylvania
https://opimweb.wharton.upenn.edu/profile/692/


*
*   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/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index