IPUMS Microdata Families with Children

Hey all,

I’m trying to construct a family variable in order to perform some migration analyses. But I’m getting tripped up on how exactly to make that work. Functionally, I’m trying to recreate the Massachusetts data within the S1101 table. Particularly, the “Household with own children of the householder under 18 years”. (Would link to the table but it doesn’t seem to work). I’m just not sure how to approach that, and any help would be appreciated.


Households and families are defined differently in the American Community Survey. Each household refers to a specific housing unit/address and contains at least one family of related individuals. Multi-family households are identified in the NFAMS variable.

The table from the Census Bureau that you referenced uses the original American Community Survey data to make the estimate you are interested in replicating for Massachusetts. IPUMS data is different from that data in two important ways. First, the Census Bureau uses the entirety of the ACS data available internally to make those estimates. The public use version of the data that IPUMS receives is a subset of that data. Second, the Census Bureau alters the data before releasing it to the public to protect respondents’ privacy. The ACS data you receive from IPUMS is not identical to the data used by the Census Bureau to create official estimates tables.Your best bet with IPUMS data is to use the Census Bureau household type identifier CBHHTYPE, which classifies households by household composition and presence of the householder’s own children.

Additionally, IPUMS creates our own family interrelationship pointer variables, which match parents with children on the basis of biological as well as non-biological (step and adoptive) relationships. If you are hoping to most closely replicate the Census Bureau’s estimates, you should use our source variables, rather than our IPUMS constructed pointer variables. If you would like to include non-biological parent-child relationships, you should use our pointer variables.

Using IPUMS-constructed variables to estimate the number of households with own children of the householder under 18 years old in Massachusetts, you should at minimum select:

  • RECTYPE, which labels records as either household or person records.
  • SERIAL, which is a unique household identifier.
  • HHWT, a household weight to be used with any household level analysis.
  • RELATE, which describes an individual’s relationship to the head of household. Individuals with RELATE=1 are the head of household.
  • NCHILD, which reports the number of own children residing with each individual.
  • YNGCH, which reports the age of the youngest own child residing with each individual.
  • STATEICP, state.

You can use these variables together to estimate the number of households with heads (RELATE=1) with own children in the household (NCHILD>0) under the age of 18 (YNGCH<18) in Massachusetts (STATEICP=3).