Hum my answer was not posed. I have used SPSS 22 and estimated a multiple imputed data set (20 dataset with about 9200 person in each). I have tried to run a GEE (binary logistic with both factor and covariates) with my MI data set. I should get an output with results from each imputed data set and pooled together (that reflect the variance between each MI data set). I have read in other forums that this might be a problem in SPSS, but would be very grateful if somebody have a solution to this.