The sampling interval for selecting households from the ACS for the PUMS

The PUMS product is a subsample of the ACS. The text below is from the "Accuracy of the Data" document, p. 5.

The number of households in the PUMS is a fraction of ACS given by the (inverse of) the sampling interval. What is the size of this sampling interval? I thought I read somewhere that it was about 1.3. Does anyone know?

Public Use Microdata Sample (PUMS)

Accuracy of the Data (2015-2019)

 

Housing Units

The sampling for HUs (and persons from HUs) is performed on the ACS samples of HUs for each of the years 2015, 2016, 2017, 2018, and 2019 as follows:

  1. Records of ACS HUs were sorted within each state by: PUMA, ACS weighting area, interview mode, type of vacant, tenure, building type, household type, householder demographics (race, Hispanic origin, sex and age), county, tract, and housing unit weight.
  2. Systematic sampling is applied to ACS HUs:
    1. Within each state, a random number is chosen between zero and the sampling interval. A counter is initialized with the random number.
    2. At each HU record, the value of the counter is incremented by one and compared to the sampling interval.
      1. If the counter’s new value is greater than the sampling interval, the HU record is selected for the PUMS and a flag is set to 1. The counter is decreased by the sampling interval with the new value passed to the next record.
      2. If the counter is less than the sampling interval, the HU record is not selected for the PUMS and the value of the counter is passed to the next record without altering its value.
  3. All HUs selected for PUMS were placed in the PUMS HU sample file.

 

The PUMS HU sample file is matched to the ACS sample of persons. All persons in selected HUs were placed in the PUMS person sample.

Parents
  • In general, the sampling interval is set to a value such that the PUMS file is an approximate 1% sampling of the full population. For 2021, we see about 1.95 million households were interviewed in the ACS representing 142 million households.  The PUMS file has about 1.56 million household records (a little more than 1% of households), so the sampling interval is about 1.25 (on average) (= 1.95 / 1.56). The specific number chosen will be based on how many households responded to the ACS in a particular year and the targeted sample size (based on the 1% of the total population calculation).

Reply
  • In general, the sampling interval is set to a value such that the PUMS file is an approximate 1% sampling of the full population. For 2021, we see about 1.95 million households were interviewed in the ACS representing 142 million households.  The PUMS file has about 1.56 million household records (a little more than 1% of households), so the sampling interval is about 1.25 (on average) (= 1.95 / 1.56). The specific number chosen will be based on how many households responded to the ACS in a particular year and the targeted sample size (based on the 1% of the total population calculation).

Children