Estimated reading time: 5 minutes

Today, the U.S. Census Bureau released the standard 2016-2020 American Community Survey (ACS) 5-year estimates that are critical for government and business uses and understanding the social and economic characteristics of the U.S. population and economy. As explained in a previously released blog, the ACS data collection operations were significantly impacted during the COVID-19 pandemic. These disruptions prevented us from collecting information from certain segments of the population. This resulted in substantial nonresponse bias in the 2020 ACS data, which means the characteristics of people who responded to the survey were significantly different from the people who did not respond. Because of this bias, the resulting data would not have been representative of the U.S. population.

We delayed the release to refine our methodology to reduce the impact of the nonresponse bias on the estimates and to ensure the methodology performed appropriately at various levels of geography.

Over the last few months, we refined how we produce the estimates. We revised our estimation methodology to improve the 2020 portion of the 2016-2020 ACS 5-year data. Specifically, we:

• Used administrative, third-party, and decennial census data to formulate the statistical weights we used for tabulating the estimates.
• Expanded our use of entropy-balance weighting. This weighting technique is suited to handle numerous inputs into the weighting model and is described in the working paper Addressing Nonresponse Bias in the American Community Survey During the Pandemic Using Administrative Data.
• Together, these revisions in our methodology made the 2020 ACS data more representative of the U.S. population. The 2020 ACS data were then integrated with ACS data from 2016, 2017, 2018, and 2019, using the standard ACS estimation methodology, to produce the 2016-2020 ACS 5-year data products. More information about the methodology is available in the ACS Accuracy of the Data (5-year 2016-2020) document. 

The ACS 5-year estimates do not represent a specific point in time but a blending of the data collected across calendar years to produce the estimates. The Period Estimates in the American Community Survey blog provides more information on how to interpret the ACS 5-year estimates.

Subject matter experts with extensive knowledge of ACS operations, trends, and characteristics thoroughly reviewed the data at multiple levels of geography. This comprehensive review served not only to ensure the data were correctly processed, but also to identify anomalies in the data.

To illustrate the complexity of the subject matter expert review and evaluation, below, as examples, are two topics related to median household income and commuting.

• Median Household Income —The newly released 2016-2020 ACS 5-year data show that U.S. median household income increased to $64,994 when compared to the 2011-2015 ACS 5-year data adjusted for inflation. Between the two nonoverlapping periods, median household income increased in 48 states and the District of Columbia, decreased in 1 state (Alaska), and was not statistically different in 1 state (Wyoming) and Puerto Rico. Although the most recent estimates contain data that include the economic shock from the COVID-19 pandemic, they also contain data collected in the four final years (2016-2019) of the longest expansion in the history of U.S. business cycles. It is important to note that the ACS 5-year estimates are not designed to measure rapid change or economic shocks during short periods because the data come from a 5-year period.
• Travel to Work — Another example is how a respondent travels to work and how long it takes them to get there. Between the 2011-2015 and the 2016-2020 periods, the percentage of U.S. workers ages 16 and older who drove alone to work declined from 76.4 percent to 74.9 percent. Working from home became more common, increasing from 4.4 percent in 2011-2015 to 7.3 percent in 2016-2020. Amid the pandemic, large numbers of workers worked from home. However, those effects were contained within one year of the 2016-2020 ACS 5-year period. Among those who did not work from home, average travel time to work increased from 25.9 minutes in 2011-2015 to 26.9 minutes in 2016-2020.

These examples reflect changes occurring over the last five years rather than any specific change in 2020.

In this thorough review, we found that the 2016-2020 ACS 5-year data are fit for public release, government and business uses, and understanding the social and economic characteristics of the U.S. population and economy.

In general, data users can compare across geographies and population groups within the ACS 5-year estimates. We strongly recommend against comparing estimates in overlapping 5-year periods since much of the data in each estimate are the same. The difference between the overlapping 5-year estimates in essence is measuring the difference between the nonoverlapping portions. For a better picture of change over time, we recommend only comparing nonoverlapping periods. More information and guidance are available on the Comparing ACS Data page and in our recent blog Period Estimates in the American Community Survey.

In summary, the ACS is the nation’s most comprehensive source of population and housing information about the United States. Knowing the significance and impact of the data, we remain committed to the collection and dissemination of quality ACS data.

Over the last few months, we worked to refine the ACS methodology to provide data fit for use by our data users. We will continue to use a revised methodology for the 5-year estimates in future releases that include the 2020 data in order to continue to provide high quality data about our nation.

Additionally, the Census Bureau is here to support our data users. Please visit our ACS Resource Hub (census.gov) for more information on how to access and get assistance with using ACS data.