Factor variable returning error code in MI logit model

[Tristen Clifton]

Hello all,

I am using Stata 19. I have imputed my dataset and I am running logistic regression models.
I am getting error 452

kidscat: factor variables may not contain noninteger values

This is telling me that this categorical variable named kidscat has non-integer values. The issue is that this variable does not, in fact, have any non-integer values. All cases are coded as 1, 2, or 3.

When I run the code below (found on another statalist thread), it shows me that the non-integer values that are causing this issue are missing values. However, this variable is imputed, so missing values should not be a problem.

list kidscat if mod(kidscat, 1) !=0

I reran my imputation and the issue remains.
Can anyone tell me what the issue may be here, and how I can go about fixing it?
Thank you.


Note: I work with secured data so I am not able to show extended code and results.

[daniel klein]

Show us your imputation model, please. In fact, show all mi commands, exactly as typed. Chances are, you've imputed non-integers.