Why the 2020 American Community Survey Is Different and Why It Matters

The COVID pandemic impeded data collection for the United States’ premier survey of local communities. How can we measure a changing America?

September is typically an exciting time for census data users because it signals the release of the U.S. Census Bureau’s American Community Survey (ACS)—the nation’s premier source of social, economic, housing, and demographic statistics for local communities.

Not this year.

It turns out the coronavirus pandemic not only had unprecedented effects on Americans’ health and livelihoods, but it also disrupted America’s data infrastructure—including the ACS.

In July, the Census Bureau announced that it will not be releasing the standard 2020 ACS 1-year data products in September as planned due to the impact of the pandemic on data quality. Instead, in November, it will release experimental estimates for a limited set of tables and geographic areas.

There were early warning signs that this was coming. In a presentation at our recent 2021 ACS Data Users Conference, Mark Asiala, an assistant division chief in charge of ACS statistical design, reported that between April and September 2020, the Census Bureau was forced to suspend many of its data collection operations, including in-person visits to homes that didn’t return survey questionnaires. He said the Census Bureau conducted one-third fewer interviews for the 2020 ACS, compared with the 2019 ACS.

A smaller sample size means larger margins of error and less reliable data—especially for small geographic areas. But the problems don’t end there. We learned in a recent Census Bureau webinar that there is significant “non-response bias” in the 2020 ACS data. People who responded to the survey had higher average incomes, education levels, and homeownership rates than those who did not. This means that the 2020 ACS estimates may be less accurate for the population groups that were most affected by the pandemic, such as workers in low-wage service and support occupations.

I think most data users appreciate the Census Bureau’s commitment to data quality and understand the decision to modify plans for the 2020 ACS data release. But it’s still a major disappointment for those who rely on these data.

Every year since 2005, ACS data have shown us how the country is changing, almost in real time. And for many policymakers and planners, the 2020 data would have provided critical insights into the changing needs of their communities.

As a demographer interested in emerging social and demographic trends, I was hoping the 2020 ACS data could help answer a few questions:

  • How many people moved during the pandemic, who are they, and where did they go?
  • What was the impact of the pandemic on people’s marriage decisions and living arrangements?
  • How did the pandemic affect patterns of commuting and telework?
  • What were the effects of the pandemic on poverty and health insurance coverage?
  • How did these effects differ across population groups and geographic areas?

For states, metropolitan areas, congressional districts, and many large counties and cities, the 2020 ACS 1-year data could have provided our first comprehensive portrait of how American life changed during the pandemic.

What about alternative sources of data? There is no substitute for the ACS, but other data sources can help us fill in some of the gaps:

  • The Census Bureau’s experimental Household Pulse Survey has provided nearly real-time data on income loss, disruptions in schools, food sufficiency, household spending, housing security and physical and mental health during the recent health and economic crises. But without a pre-pandemic baseline, it’s hard to put these numbers into context.
  • The 2021 Current Population Survey (Annual Social and Economic Supplement) will provide valuable information on social and economic trends during the pandemic, but the sample of around 100,000 households is too small to provide reliable annual estimates for states or local areas.
  • ACS 5-year estimates are tentatively scheduled to be released in March, but using the 2015-2019 and 2016-2020 estimates to measure trends during the pandemic will be challenging because four of the five years overlap. And, the Census Bureau has just started to evaluate the quality of the 2020 5-year estimates, so we do not know whether those data will be available in the same form as previous years.
  • 2020 Decennial Census Redistricting data are now available, but the census questionnaire included just a handful of questions about basic demographic and housing characteristics. The ACS replaced the census long form beginning in 2010 by collecting long form-type information throughout the decade rather than only once every 10 years.

The good news is that 2021 ACS data should be released next year as planned. Because ACS data collection is ongoing, we will eventually be able to compare population and housing characteristics before, during, and after the pandemic. Will some parts of the country bounce back faster than others? What are the long-term effects on employment, poverty, and inequality for different population groups?

The effects of the coronavirus pandemic and related economic shocks are expected to last for years. In 2022 and beyond, ACS data will continue to be a critical resource to understand how America is changing.