Does anyone know how to obtain Type 3 Tests of Fixed Effects estimates when using multiple imputation? We are using PROC MI, MIXED, and MIANALYZE which has worked well when the iterative estimates and associated standard errors are available. We are having difficulty using MIANALYZE to obtain the F statistics and degrees of freedoms from the MIXED Type 3 table. All calculations I've found about combining results across imputed datasets depend on the availability of standard errors, and we do not know where the standard error for the F value is located. Please advise.