Mailbag: Questions about ACS, Census, and population estimates

There are three key data sets from the U.S. Census Bureau that we use to understand population trends: the decennial census, the American Community Survey (ACS), and population and housing unit estimates. (Background: a deep dive into each of these data sources). These sources are similar, but not quite the same, which can lead to questions from data users.

Here is a summary of some of the questions we’ve received from data users and/or seen on our favorite ACS data forum—the Population Reference Bureau’s ACS Data Users Group.

How do I interpret 5-year American Community Survey data?

Five-year estimates are based on surveys collected over a five-year period. Taking five years of surveys gives a larger sample of responses, so it’s possible to develop estimates for smaller geographic areas with a reliable margin of error, which makes them reliable as estimates.

5-year estimates are period estimates. Here’s what that means: You could have responded to the survey sent in the first year of the five-year period, and I could have responded to the survey in the fourth year of the same five-year period, and both of our responses would be used as samples to come up with estimates for that five-year period. In the following year, your survey answers would not be used – because they’re now six years old – but mine would still be part of the next five year period. (Response adapted from JamiRae.)

It’s important not to compare 5-year estimates with overlapping dates. To compare 2016-2020, for example, you’d want to use 2010-2014 and not 2015-2019.

There are two different numbers for the 2020 population for my county, one from the Census Bureau’s 2020 redistricting data file and one from the 2020 population estimates. Which is the correct one to use?

Both of these numbers are correct, but we generally recommend using the 2020 Census numbers. The Census numbers released in August 2021 are the official decennial census total population and reflect the population as of April 1, 2020. The population estimates are also from the Census Bureau and are estimates from July 1, 2020. The new estimates for July 1, 2021 were released on December 21, 2021 but are only available at the national and state levels right now. County-level estimates will be released this spring. (Response adapted from Chuck Purvis.)

Why does the race variable data differ between the 2020 ACS 1-year Public Use Microdata (PUMS) with experimental data and earlier years, when race is one of the dimensions controlled for in the weighing?

These changes are likely due to the changes to the race and ethnicity in the 2020 ACS, which were similar to the changes made in the 2020 Census. One big change is that the number of characters allowed in the write-in area was increased from 30 to 200 characters, allowing up to six distinct groups to be coded and tabulated from each line.  (Response derived from Karen Louie.)

What is a housing unit? What are group quarters?

An explainer from Metropolitan Council Research, which was updated to reflect 2020 Census definitions and written for city planners, researchers, and demographers is available in their downloadable local planning handbook.

Is the 2020 Census redistricting population the official final population for states and counties?

This is actually tricky: there’s no such thing as a final file – and states differ in terms of what is considered “official.” Here are the various 2020 Census products:

• The Redistricting Data file. This could have errata corrections (through end of 2023) out of the Count Question Resolution process, but the hurdles for that CQR process are high, and corrections affecting county and state totals will be exceedingly rare.
• There will also be subsequent 2020 Census data products, but not until the end of 2022. The Census Bureau is saying those products should have the same state population totals as Redistricting Data file. But there could be miniscule differences in county population totals – due to the new Disclosure Avoidance System.

Check the state laws where you live: our advice is to follow their guidance. (Response derived from Todd Graham.)

Why are ACS estimates for some places much different for subpopulations?

ACS data are benchmarked to the Census Bureau’s population estimates for a given year, and those estimates are benchmarked to the most recent Census, so the last year of a decade can often look very different – and those differences may differ across age, race/ethnicity, gender, etc. Beth Jarosz of the Population Reference Bureau recommends two ways to deal with this in the short-term:

Option 1: Use ACS, and note the break-in-series at 2020. (This is not a preferred approach, but is often used in the period between when the decennial is released and when revised historical estimates are available, see Option 2 below.)

Option 2: Create an independent set of revised intercensal estimates (or work with a known set). This is the preferred approach BUT note that the ACS data itself will not be re-benchmarked. So if you go with revised estimates, you should not report count estimates from ACS side-by-side with those revised estimates. (It is ok to report revised population counts from a new set of independent estimates alongside *rates* or *percentages* from ACS.) (Response derived from Beth Jarosz.)

Republish our content for free under a Creative Commons license.

Need help understanding population change and its impacts on your community or business? Carolina Demography offers demographic research tailored to your needs.

Contact us today for a free initial consultation.

Categories: Census 2020