Hi. For a project I am using plausible values (performance data) and hence use sem in combination with the imputation framework. This works fine. However, I would like to test for significant group difference afterwards, which can be done using estat ginvariant. However, this does not work. The command leaves chi2 and p-values behind. I wonder how I should best combine them over the imputed samples. Is there any advice? I have found this here, maybe StataCorp can comment on that. Thanks a lot.