I’m running models on imputed datasets. We occasionally get this error once a model runs, and we can’t identify what is causing it. We’ve replicated it across several imputation methods (mice and Blimp) but can’t create a reproducible example.
Does anyone have insight into what this error is telling us?
We do some post-imputation processing because the outcome in this model is dependent on the value of another also imputed variable, and it must be missing if that variable has a value of zero. This causes differences in sample size across imputed datasets. Would this cause the issue?