Dear MI experts,

I am new to the field of MI and just learned Amelia II. I like to know
if the imputed data sets, say 10, generated with Amelia II can be used
to study the distribution of variables that were included in the MI process.

Specifically, I like to impute party identification, which has about 30%
missing, and explore (and hopefully restore) the proportion of partisan
voters.

Is it proper (or it is too naive) to could use imputed data sets in this
descriptive and predictive way? If multiply imputed data sets can be
used this way, what could be the best strategy to use these ten sets of
proportions? (e.g., averaging them? plotting them?)

Thank you very much.

Best,

Frank