Hi, Alex.
For clustered standard errors, see demo(robust). However, clustered
standard errors and multiple imputation don't work together by design.
Multiple imputation and robust standard errors do some correction to the
variance covariance matrix to obtain the "correct" standard errors, and
there hasn't been a theoretical/statistical examination as to which should
go first, or whether they are compatible at all.
You might want to try GEE logit or a multi-level logit (use a group-level
intercept) without the robust option with MI.
Best,
Olivia
On Sat, Aug 21, 2010 at 2:55 AM, Alexander Lee <amlee(a)stanford.edu> wrote:
Hey All,
I was wondering if anyone had some sample code for running
clustered standard errors for a logit in zelig. The manual says this
is possible, but every time I try to specify the cluster variable the
arguement comes back as unused, and I get the same model as if I
hadn't specifed anything. Right now I'm running something along the
lines of:
model1 <- zelig (DV ~ IVs , model="logit", robust=TRUE", data=
mi(datasets))
How do you indicate a cluster?
Alex
-
Zelig Mailing List, served by Harvard-MIT Data Center
Send messages: zelig(a)lists.gking.harvard.edu
[un]subscribe Options:
http://lists.gking.harvard.edu/?info=zelig
Zelig program information:
http://gking.harvard.edu/zelig/