MI estimation proportion and factor variable

[Jack Miller]

I am working on dataset about hosehould wealth. Missing variables were imputated according to the guide written by data provider. I have worked in Stata before, but I do not have experience regarding MI. I am using Stata 15.

I would like to create a table containing distribution of wealth portfolio components(binary variable has/has not) over country using command "mi estimate: proportion [varnames] over [varname indicating country]. Variable indicating country is a factor variable so I get error: 'factor-variable and time-series operators not allowed".

How should I proceed? I feel that this question must be basic, but I did not find an answer.

[daniel klein]

This problem is not specific to mi. When you specify over() you specify that you want the respective variable be treated as categorical. There is no need for factor variable notation; it is not even allowed. You do not want

proportion varname , over(i.varname_indicating_country)

but instead

proportion varname , over(varname_indicating_conutry)

Note absence of factor variable notation in the second example.

Best
Daniel

[Jack Miller]

Thanks a lot Daniel!