mixed effect modeling with imputed data set

  For convenience, you might want to consider using imputation with
the 'mice' package in order to
stay within R for your analysis (I don't know Stata's capabilities in
this area; I'd expect them to be pretty good,
but it is my impression that 'mice' is also quite
well-written/full-featured)

On Tue, Jun 16, 2015 at 12:26 PM, Daniel Fulop <dfulop.ucd at gmail.com>
wrote:

Use one of the apply functions to iterate over your imputed datasets.

If your imputed datasets are in columns 5 through n+4 of "mydata" (i.e.
assuming that x1, x2, x3, and regioid are in columns 1:4), the you could

do

something like:

model.list <- lapply(1:n, function(i)

glmer(mydata[,i+4] ~ x1+x2 +x3+(1|regiogid),family= binomial("logit"),
data=mydata) )

The output will then be a list of model objects (i.e. model fits).  You

can

then iterated through this results list in order to calculate mean

parameter

values from all your imputed data fits.

Or, likewise:

model.list <- lapply(5:ncol(mydata), function(i) ...

Hope this helps,
Dan.


ali via R-sig-mixed-models wrote:

Hi all
I imputed my data using multiple imputation procedure in STATA. I would
like
to conduct mixed effect modeling on the imputed data set in R. I do not
know
how to write the code over the imputed data set.
My code is:
fit<- glmer(y ~ x1+x2 +x3+(1|regiogid),family= binomial("logit"), data
=mydata)

Best Regards

--
Daniel Fulop, Ph.D.
Postdoctoral Scholar
Dept. Plant Biology, UC Davis
Maloof Lab, Rm. 2220
Life Sciences Addition, One Shields Ave.
Davis, CA 95616

510-253-7462
dfulop at ucdavis.edu