Creating a new variable for whether or not anyone in HIU is on Medicaid/SCHIPs in past year

CPS
1

Hi,

I’m trying to create a dummy variable for whether or not anyone in the household (as measured by HIU if possible, instead of using FAMUNIT) has been on Medicaid/SCHIPs in the past year. It is straightforward to do this for just the respondents, but is there anyway to account for other household members and attach it to the individual? For example, Head of household is not on Medicaid, however, he has a child that is on SCHIPs. So, although he would respond “no” to HINSCAID because he himself is not on it, he would respond “yes” to “anyone in HIU on Medicaid/SCHIPs/public insurance.”

Thank you!

2

Computationally speaking, there are a number of ways to attach information from one member of a household to all others. The IPUMS page on Family Interrelationship Variables includes an Endnote that gives example syntax for linking information from children to parents by creating a temporary data file. However, the process is a bit easier since you are linking information based solely onSERIAL (the unique household identifier within a sample). You can simply create a new variable called HSCHIP which is either a 1 [“receives SCHIP”] or a 0 [“Does not receive SCHIP”]. You could then collapse the data to the household level using YEAR, DATANUM, and SERIAL to uniquely identify households, selecting the maximum value of HSCHIP to use for the household value. This would give you a dataset of households that identifies whether anyone in the house receives SCHIP (HSCHIP==1) or no one in the house receives SCHIP (HSCHIP==0). You could then merge this household level information onto the original person level data by using a many-to-one merge on YEAR, DATANUM, and SERIAL.

I hope this helps.

3

Thank you! That’s very helpful.

The other question, however, is whether or not the variables SCHIPS or KIDCAID picks up on kids that haven’t been surveyed (so those under 15). It seems to me that the variable is just asking if they respondant is a child and then if so, are on SCHIPs (versus asking the respondant if s/he has children on SCHIPs). Is this true? If so, is there another variable that accounts for children on SCHIPs that I am missing?