Combining the results after multiple imputation

From: "Diane Mould" <DRMOULD@ATTGLOBAL.NET> Subject: Combining the results after multiple imputation Date: Wed, 27 Jun 2001 13:25:31 -0400 Dear NMUSERS I would appreciate some input on how to compute the interval estimates for parameter values following multiple imputation. I have simulated several data sets to replace missing data, and then fit them in a logistic regression model. The parameter estimates for all runs are quite similar, and have been tabulated. I wanted to use "Rubin's Rule" for scalar estimands to combine these parameters to get the interval estimates. I have been following the handout given by Joseph Schaffer at PAGE this year, which defines some of this on Handout 24. However, the handout does not define all the parameters, unfortunately. Can anybody give a quick summary of the procedure? I would really appreciate it. Best Regards Diane