Identifying parents who are both noncitizens and uninsured of US born childrend

I am having trouble with identifying parents who are a) uninsured and b) non-citizens of U.S born children enrolled in Medicaid and not enrolled. I am unsure of what variables to use in IPUMS USA or if I should look into other IPUMS data sets such as CPS or NHGIS. If so, what variables should I use?

The variables I have used are “PUMA”,“SERIALNO” “AGEP”, “HINS4”, “HICOV”,“CIT”,“NOP”, and "FPARC”.

In the example below I used “NOP, AGEP, and CIT” to identify US-citizen children with non-native parents. However I don’t think this approach would be correct to identify parents who are both uninsured and noncitizens with US born insured children.
as_survey_design(weights = PWGTP)%>%#set weights for estimation (I don’t know why but if I move this line after group_by, 3 groups don’t show)
filter(NOP ==6|8)%>% #filter for children living with two parents where either the mother or father is foreign born
filter(AGEP <= 18)%>% #filter for only children under 18
filter(CIT >=1 & CIT <=4)%>% #filtering for children who are citizens
mutate(age_group = case_when(AGEP == 0 & AGEP <=1 ~ “Infant”, #creating age groups for children
AGEP >1 & AGEP <=5 ~ “Children 0-5”,
AGEP >=6 & AGEP<= 18 ~ “Children 6-18”))%>%
group_by(age_group)%>% #grouping by age group
summarize(health_child = survey_mean(HICOV==1, na.rm =TRUE),# % of children who have health insurance
nohealth_child = survey_mean(HICOV==2, na.rm =TRUE),# % of children who do not have health insurance
medicaid_child = survey_mean(HINS4==1, na.rm =TRUE),# % of children who are enrolled in Medicaid
n = unweighted(n()))#gives me raw count

I am currently trying this out with one state before analyzing for all states. It also helps with analyzing a smaller data set for one set vs. for all.

Any guidance is appreciated!

My understanding is that you wish to use IPUMS data to identify individuals that fit three characteristics: 1. they do not have health insurance coverage, 2. they are not U.S. citizens, and 3. they have a child or children born in the U.S. who has/have Medicaid insurance. This should be possible with IPUMS USA data. I will provide you with some information about how you might accomplish this.

First, it looks like you are using the original PUMS files from the Census Bureau, not IPUMS data files. I would recommend working with the harmonized data available from IPUMS USA unless there is a particular reason you are interested in the original data, as IPUMS’s harmonization process makes variable codes, universes, and other data attributes comparable across time and space. I will provide information here on IPUMS’s harmonized data.

You have not specified the year(s) for which you are hoping to conduct this analysis, but note that health insurance variables are only available in the ACS from 2008 onward.

In IPUMS USA, HCOVANY is available for all persons in the ACS from 2008 to 2021, and reports whether the person has any health insurance coverage. HINSCADE has the same availability and universe and reports whether the person is covered by Medicaid. BPL is available for all persons and reports the person’s birthplace. CITIZEN reports the citizenship status of all foreign-born persons from 1950 to 2021, and for other universes of persons from 1870 to 1940.

When you go to create your data extract, you will have the option to attach characteristics of other household members. You can attach the CITIZEN and HCOVANY values of respondents’ mothers and fathers to their records. You should also attach mother and father PERNUM. This will allow you to identify children of parents who meet your criteria. From there, you can use SERIAL and PERNUM to identify the parents themselves.

You could take a similar approach to your task using IPUMS CPS data. Either survey seems suitable to me, but you may want to compare information about the CPS and the ACS, such as sample design, survey universe, and sample sizes to determine which is best for your project.

IPUMS NHGIS would not be suitable for this purpose, as it provides summary tables aggregated at various geographic levels, and not microdata, or person-level data.

1 Like


Apologies for the late reply! This was incredibly helpful and it solved the issue I was having.

1 Like