The U.S. Census Bureau today released undercount and overcount rates for housing units from the 2020 Post-Enumeration Survey (PES). These results provide insight into the quality of the 2020 Census counts for housing units for the nation, regions, states, the District of Columbia, and Puerto Rico, and by selected characteristics and census operations for the nation.
“This release completes the set of results from the 2020 Post-Enumeration Survey,” Census Bureau Director Robert L. Santos said. “This is one of many self-assessments we’re doing that allow us to think critically about how we transform our operations and plan for the 2030 Census.”
The PES estimates help show how well the 2020 Census covered housing units in the nation, excluding Remote Alaska areas. To do so, the PES created an independent estimate of the number of housing units in the United States and Puerto Rico on April 1, 2020.
Key findings:
- National coverage
- The PES estimate of 140.4 million housing units in the nation was not statistically different from the 2020 Census housing count of 140.5 million. In other words, the estimate of net coverage error of 0.04% (or 60,000 housing units) was not statistically different from zero. The estimate of percent net coverage error of housing units in the nation for 2020 was statistically different from the undercount estimated in 2010.
- In previous decades, the PES estimated the census undercounted housing units by 0.60% in 2010; 0.61% in 2000; and 0.96% in 1990.
- In 2020, the estimated rate of erroneous enumerations of housing units was 3.1% or 4.4 million.
- Subnational coverage
- The PES estimated a statistically significant overcount (1.9%) of housing units in the Northeast Region.
- Census counts of housing units in the Midwest, South and West did not have an estimated statistically significant undercount or overcount.
- Housing units for the following states had a statistically significant undercount: South Carolina (2.5%) and Vermont (4.1%).
- Housing units for the following states had a statistically significant overcount: Alabama (2.7%), Massachusetts (1.4%), New Jersey (2.5%), New York (3.7%), Ohio (1.2%), Rhode Island (1.6%) and Utah (0.8%).
- Housing units in Puerto Rico did not have a statistically significant undercount or overcount in the 2020 Census. (The Census Bureau today also released coverage results for Puerto Rico population counts.)
- Coverage estimates by occupancy and tenure (owner, renter)
- Occupied housing units had a statistically significant overcount in the 2020 Census of 0.33%.
- Rented housing units had a statistically significant overcount in the 2020 Census of 0.85%.
- The estimated net coverage error rate of owned housing units was not statistically significant. Vacant housing units had a statistically significant undercount of 2.6%.
- Coverage estimates by type of structure
- Small multiunit housing units (structures with two to nine units at the basic street address) had a statistically significant overcount in the 2020 Census of 5.1%.
- Mobile homes and other units had a statistically significant undercount of 4.3%.
- Coverage estimates by race alone or in combination and Hispanic origin of the householder
- Housing units with a householder who was Black or African American (0.87%), Asian (1.37%), Native Hawaiian or Other Pacific Islander (2.64%), or Some Other Race (0.58%) had statistically significant overcounts.
- The estimated coverage rate of housing units with a householder who was White, American Indian or Alaska Native, or Hispanic or Latino were not statistically different from zero.
- Coverage estimates by census operation
- Housing units in the Update Leave and Update Enumerate Type of Enumeration Areas had a statistically significant undercount of 4.2%.
The Census Bureau released the first results from the PES along with additional results from the 2020 Demographic Analysis (DA) earlier this year. They showed the strength of the total U.S. population count and that coverage varied by demographic groups such as race and Hispanic origin, as well as by age groups and sex.
The Census Bureau also released earlier this year person coverage estimates for each state and the District of Columbia from the PES.
Difference between coverage of housing units and population counts
The way the Census Bureau accounts for all housing units in the U.S. is slightly different than how it accounts for the population. The key difference is that the Census Bureau works throughout the decade to build a list of all the housing units in the nation. This address list, called the Master Address File, is then used to invite people living at those addresses to respond to the census and to follow up with them if they do not.
The list is also used to ensure the population is tabulated to the right location. This means that housing unit coverage relies more on an accurate address list than on obtaining responses required for counting people.
“These results provide valuable insights that inform and motivate numerous innovations and operations related to building and refining the address list from the 2020 Census. Furthermore, work remains on the 2020 Census. We’re not done yet. There are still large data products forthcoming that show the rich diversity of the American people,” Santos said. “As part of the transformation process, we’re taking a fresh look at who we are and looking at ways to engage with the public and stakeholders to get input on designing the 2030 Census – one of many efforts underway to learn how we, the Census Bureau, can contribute more relevant data to the nation.”
In addition to releasing the Census Coverage Estimates for Housing Units in the United States and Census Coverage Estimates for Puerto Rico reports, the Census Bureau also released the Source and Accuracy of 2020 Post-Enumeration Survey Housing Unit Estimates.
Other reports are available, including a treatise on imputing demographic characteristics for the PES and a technical report describing methods and results used to reduce nonresponse errors in the PES for person estimates.
Original source can be found here.
