Thank you very much in advance for reading this post, and for any help that you can extend.

I am trying to impute some variables in my data using MICE via the mi impute chained syntax.

I do not have any missing data on: race, location (urban/rural), marital status, housing type
I have missing data on income, educational attainment, number of children. Suppose I were to try to run a model such as the following:

Code:
mi impute chained ///
(reg, include(c.nkids)) income ///
(poisson) education ///
(pmm, knn(10)) nkids) ///
= race location maritalstatus housingtype, ///
 dryrun rseed(20190930)
Ignore the fact that I should not be using only number of children (the variable nkids) as a predictor variable for income. This is just an example.

When I run this, I get an error:

Code:
nkids: imputation variable cannot be also specified as independent variable
 -- above applies to specification (reg, include(c.nkids)) income = race location maritalstatus housingtype
How can this be, and perhaps more fundamentally, what does this even mean?

Surely the whole idea of MICE is that I can impute a variable with missing data, and the subsequent model can use that variables with those imputed values in its own prediction equation.

Oddly enough I am able to "omit" nkids in which case nkids is not included as a predictor for income and all the other variables are. But I am trying to do the reverse - include only nkids. The code I have given above is an example, and the actual code is considerably longer. It is so much easier to specify the list of variables to "include" rather than exclude.

Any help? Thanks a ton.