st: Importing multiply imputed panel data

["Kemeny, Thomas Ernest" <tkemeny@email.unc.edu>]

Dear Statalist,

I have a panel dataset that includes multiply-imputed information. I have converted this data into -ice- format, and now I am trying to figure out how to import it.

The original multiple imputation (performed earlier by Census Bureau economists) created 10 imputes for the missing variable. So there are ten possible values for each variable at a given time-point.

I thought it made sense to set the up like this, with _mj indexing imputation number (zero means original data); _mj indexing individuals, v is the substantive variable and yr indexes years :

_mj	_mi	 v	yr
0	1	3	0
0	2	.	0
1	1	3	0
1	2	8.5	0
2	1	3	0
2	2	9.5	0
0	1	5	1
0	2	.	1
1	1	5	1
1	2	9	1
2	1	5	1
2	2	11	1


You can see that in years 0 and 1 in the original data, ID#2 is missing the value of v. In imputes 1 and 2 this value is imputed.

If I try to import this into stata's MI suite (Stata 12.1) using -mi import ice, automatic-, I get the following error:
.variables (_mj _mi) do not uniquely identify the observations
    at least one (_mj _mi) value is duplicated
r(459);

This makes sense - the panel nature of the data is causing problems. Any idea how to import multiply imputed data like this into Stata's MI format? It need not be through -ice-, I'm open to any solution here. 

Thanks in advance for your help!
Tom Kemeny


--
Thomas Kemeny
Research Assistant Professor
Department of Public Policy
University of North Carolina at Chapel Hill
http://unc.edu/~tkemeny/





*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/