I have been asked to come up with a number for adult dependents (i. e., not the householder or spouse, 18-65 years old, financially dependent on the head of household/spouse). I do not believe that ACS data lets me ask that question directly. Using PUMS (OPUMS in my case) it might be possible to infer a number by looking to groups such as persons who meet the criteria and attend college or graduate school or who are disabled and unemployed. I have also looked at IRS data but so far have not found any tables that would speak to the question.
Does anyone have suggestions for how to manipulate ACS or PUMS to get at this question or other data sources that might be on point?
Can’t imagine it in ACS. Maybe microdata from IRS Statistics of Income? They do deal with dependents www.irs.gov/.../soi-tax-stats-individual-public-use-microdata-files
It sounds like you're trying to come up with the definition and criteria?
For sure, look at the income of each individual person in a household (or more limitedly: a family). That's going to be in the "P" table of ACS PUMS.
College enrollment status sort of makes sense... but is the college enrolled person a relative (usu. child) of the householder? or is she an unrelated housemate/roommate? In that latter case, that does not seem like a dependent relationship (though she may be dependent on family living entirely elsewhere -- and you won't have those characteristics in ACS PUMS. Good luck!
If you look at the ACS (2021) survey form https://www2.census.gov/programs-surveys/acs/methodology/questionnaires/2022/quest22.pdf
There is question for persons 2, 3, etc (question #2) that asks for the relationship to Person 1 (the person completing the survey form). To be a dependent (in the IRS sense) you need to be related to another person in the household and make less than $4300 per year. Each person's income in on the survey form. Together with age you should be able to use these PUMS variables to get "an adult dependent." variable. Remember the PUMS data is only for PUMA geographies, which need to have over 100,000 persons and don't necessarily correspond to city/town or county boundaries.
Dave Dorer (you have my email and phone so email or call if you want to discuss)
Thanks for the suggestion. This is helpful.