American Community Survey (ACS) data are controlled so that the number of housing units and people in certain categories conform with the Census Bureau’s official estimates from the Population Estimates Program (PEP).

The ACS uses the postcensal estimates as controls for years following the last decennial census until the next decennial data are available. These estimates are population estimates produced for the years after a decennial census when only the beginning population is known.  The entire time series is extended and revised each year.  For example, the 2019 ACS 1-year and 2015-2019 ACS 5-year estimates used the Vintage 2019 postcensal estimates as controls which provide estimates covering the entire five-year period.

The ACS uses intercensal estimates as controls in decennial years and for part of the 5-year estimates following the decennial years.  Intercensal estimates are population estimates produced for the years between two decennial censuses (e.g., 2000 Census and 2010 Census) when both the beginning and ending populations are known.  They are produced once a decade by adjusting the existing time series of postcensal estimates for the entire decade to smooth the transition from one decennial census count to the next.  For years when both postcensal and intercensal estimates are available, intercensal estimates are preferred. Thus, for the ACS 5-year estimates, the intercensal estimates are used for the portion of the five-year period that falls prior to the last census and postcensal estimates are used for the portion that falls after the last census.

Controls Used After 2010 Census

To provide some examples of this in practice, the ACS program used the following controls in our multiyear estimates following the 2010 Census:

2006–2010 ACS 5-year:

• 2006, 2007, 2008, 2009: intercensal estimates
• 2010: extrapolation of the 2010 Census count from April 1 to our standard July 1, 2010 reference date using the intercensal estimates (these extrapolated estimates were also used as controls for the 2010 ACS 1-year estimates)

2007–2011 ACS 5-year:

• 2007, 2008, 2009: intercensal estimates
• 2010, 2011: Vintage 2011 postcensal estimates (the 2011 Vintage 2011 was used as controls for the 2011 ACS 1-year estimates)

This process continued until the 2010-2014 ACS 5-year release when all five years used the Vintage 2014 postcensal estimates.  The 2019 ACS data release would typically be the last year when we use the postcensal estimates based on the 2010 Census as controls.

Plan for Controls After 2020 Census

The ACS program intended to replicate what we did following the 2010 Census for the 2020 ACS estimates.  Due to delays in the 2020 Census processing, the 2010 to 2020 Intercensal Estimates have been delayed from 2021 to 2022. This impacts our plans for the controls used both in the 1-year as well as in the ACS 5-year estimates.

The ACS program now plans to use the Vintage 2020 postcensal estimates for the 2020 ACS 1-year and 2016–2020 ACS 5-year data releases.  If the 2010 to 2020 Intercensal Estimates are produced in time for 2021 ACS processing, we plan to use intercensal estimates for 2017, 2018, and 2019 with Vintage 2021 postcensal estimates for 2020 and 2021.