Hi Helen,
Unfortunately, you are correct that "miest" does not support the ivreg
command. The simplest thing for you to do is to estimate your model for
each of the imputed datasets and then combine the results following the
standard formulas (described for example on p. 53 of our article on
missing data, available at
http://pantheon.yale.edu/~ks298/index_files/Missing2001.pdf). Of course,
you could also use the code for "miest" as a template to write code for
the ivreg command in stata.
Best wishes,
Ken
Helen Brown wrote:
Thank you for the clarification. I now understand
what the output of
misum means.
I have another doubt, though. It seems that miest does not support
ivregress in Stata so I am not sure how to combine the imputed
datasets for IV-estimation.
Thanks.
Regards,
Helen
On Tue, Dec 23, 2008 at 4:41 PM, Ken Scheve <kenneth.scheve(a)yale.edu
<mailto:kenneth.scheve@yale.edu>> wrote:
Helen,
"misum" gives you the multiple imputation estimate of the mean and
the standard error of this estimate. It does not give you the
standard deviation of the variable. I suspect you may be comparing
these two things and they are not the same thing. Of course, if
upon further investigation you see a problem, let me know.
Thanks,
Ken
Helen Brown wrote:
hello,
a very basic question: why is it that the standard errors
(calculated with misum) over the 10 imputed datasets with
Amelia II are much smaller than the ones from my original
dataset (even for variables that did not have so many missing
obs. or at all in the latter)? Am I doing sth. wrong with the
imputation process?
thanks so much
Helen
-
Amelia mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive:
http://lists.gking.harvard.edu/?info=amelia
More info about Amelia:
http://gking.harvard.edu/amelia