you could do it in the same way you do in a regression by adding a
separate dummy variable for the outliers. or if the 'outliers' are in
fact just miscodes, you could remove them and let Amelia impute them.
Gary King
On Mon, 13 Nov 2006, aleman(a)fordham.edu wrote:
I used Amelia II to impute missing observations for a variable with a few
outliers. I did not want to remove these outliers in an ad hoc way, since
the cross section is integral to the analysis. Nevertheless, I found many
of the imputed values to be unduly "influenced" by these outlier
observations.
Is there a non ad-hoc way to account for these outliers when performing
the imputations without having to remove the entire cross section?
Thank you,
Jose Aleman
Assistant Professor
Political Science Department
Fordham University
- Amelia mailing list served by Harvard-MIT Data Center
[Un]Subscribe/View Archive:
http://lists.gking.harvard.edu/?info=amelia