Re: st: xtprobit clustered standard errors or gllamm

[Gordon Hughes <G.A.Hughes@ed.ac.uk>]

-xtprobit- does not offer the vce(robust) or vce(cluster) 
options.  This is unusual for a Stata xt... procedure and you might 
take that up with Stata Corp.  However, for practical purposes the 
vce(bootstrap) option should be a perfectly satisfactory alternative 
provided that you declare an appropriate cluster() variable in the 
vce specification.  Use a much larger number of replications than the 
default - Cameron & Trivedi recommend at least 400.  For further 
details, see Colin Cameron & Pravin Trivedi (2010), Microeconometrics 
Using Stata, College Station: Stata Press.

Gordon Hughes
g.a.hughes@ed.ac.uk

============================

Dear all,

I use Stata 11 SE and I am looking for a way to estimate a xtprobit model
with clustered standard errors.

I have data from 180 subjects who are divided into 60 independent groups but
the subjects within one group interact with each other. The dataset contains
five periods but each subject occurs only two or three times within the five
periods.
My dependent variable is binary and a referee has asked for a random effects
probit model with clustered standard errors on group level to control for
individual subject effects, time and  the potential correlation of errors
within a group.


*
*   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/