Family interrelationship codes and imputed values

I am wondering whether IPUMS creates parental linkage variables before or after making allocations/imputations to age and gender.

For a project I am working on, it is very important to match women to all their correct children in the household and have the children matched in the correct birth order. This is done using momloc combined with age and birthyear. However, we do not want to include children or mothers who have their age or gender imputed (qage and qsex !=0).

It occured to me that parental linkage variables may be created after there have been imputations to age and gender in the census data. Since the IPUMS provided rules for generating momloc/poploc uses the age and gender of household members, it could be that some of these variables depend on the way age and gender was imputed, which would be problematic for our analysis.

If the answer is different for IPUMS-I and NAPP, I would appreciate knowing that as well. Thanks in advance for any insight into this issue.

All IPUMS family interrelationship variables are generated after allocation. However, if you use the QAGE and QSEX characteristics (these are data quality flags) of the mother for every case, you can drop cases where QAGE and QSEX are != 0 and then drop cases were QAGE_MOM and QSEX_MOM are != 0. This should ensure that no imputed or allocated information is used. You can find more information on attaching characteristics to your data extra here.

1 Like