I was trying to use multiple imputation for a random effects model where a factor variable has missing data. I get a warning message that says "Invalid factor level, NA ge