As a novice to multiple imputation methods, I'm looking for advice regarding appropriate approaches to handling missing data in the following situation:
I have 3 groups of patients with different brain lesion patterns. The patients have been given a battery of 6 cognitive tests assessing memory, reasoning, visuoperceptual functioning, etc. But not all patients were able to complete all 6 cognitive tests. There are supplemental data, eg, age, education, etc. I want to examine group differences on the cognitive tests. In the past, with complete case analysis, I have used ANOVA, MANOVA, or logistic regression. With this data set, I would like to use multiple imputation to handle the missing congitive test data. In the past, I have used SAS proc MI, SOLAS, or F Harrel's splus transcan program for multiple imputation. In this 3 group case, I would appreciate any advice regarding imputation and modeling strategies. For example, in SOLAS I am unable to combine ANOVA results from the individual imputed data set analyses. Thanks, SR Millis
