Fact-checking Stacey Abrams’ comparison of census response rates

Stacey Abrams said President Donald Trump is trying to discourage immigrants from responding to the decennial census and said response rates so far were trailing rates from the 2010 census.

Trump directed the commerce secretary, who oversees the census, to not count immigrants illegally in the country in the data used for the apportionment of U.S. House seats. Abrams said Trump’s directives would likely fail legal challenges, yet "heighten the likelihood that the census will look whiter, which tends to mean the voting power is more Republican."

"Right now, we are about 4% behind the response rate where we were in 2010. But for Black and Latino and Native American communities, they’re behind by 10%," Abrams said in a July 23 interview with MSNBC’s Rachel Maddow.

We wondered if Abrams’ claim on the response rate was accurate. Her staff said the point she was making was that the national self-response rate was 4 percentage points behind the final self-response rate in 2010; and that during the 2020 census, Latino, Native American, and Black communities’ self-response rates are 10 percentage points lower than in white communities.

We learned that it is not easy to compare this year’s census to the one in 2010. The U.S. Census Bureau told PolitiFact that shifts in timelines due to the coronavirus pandemic and different response methods prevent an accurate comparison of the 2010 and 2020 censuses. Households this year can answer the census questionnaire online, an option not available in 2010. The Census Bureau also does not break down response rates by demographics.

Abrams’ claim is based on an analysis of self-response rates from the Center for Urban Research at the City University of New York’s Graduate Center, said Lauren Zehyoue, communications director for Fair Count, a group created by Abrams to encourage census participation.

There’s some data that can be used to gauge response rates in 2010 and 2020, but Abrams was imprecise in her wording, said Steven Romalewski, director of the CUNY Mapping Service, a project of the Center for Urban Research.

Zehyoue told PolitiFact that Abrams was comparing the national response rate as of July 23, 2020, with the final response rate in 2010. 

Using the final rate for comparison (and not the same date from 2010) is reasonable, Romalewski said, as "that’s what all the census stakeholders are using as their reference point." In 2010, the self-response operation ended April 30, and if there was no pandemic in 2020, it would have been more appropriate to compare exact dates in 2010 and 2020, he said. 

Census Bureau data shows the 2010 response rate reached its highest point, 66.5%, on June 22, 2010. This year, on July 23, the national self-response rate was 62.4%. By this metric, Abrams’ point holds up, although she would have been more precise by saying we are behind "about 4 percentage points," not 4%, Romalewski said.

Self-response rates refer to households, not individuals.

After the self-response phase ended in 2010, census takers went out knocking on doors and collecting information from households that had not responded. At this point, households could still mail their responses and not respond in person. The final 66.5% rate includes responses mailed back throughout that period. 

The Census Bureau flagged a different number when we asked about the final response rate — 63.5% — referring to responses mailed back before census-takers went out.

Overall, both metrics still support the argument that the rate in 2020 so far is behind 2010’s final rate, although it’s hard to do an apples-to-apples comparison, said Steve Jost, who was associate director for communications at the U.S. Census Bureau during the 2000 and 2010 censuses. Jost serves as an adviser to the Census Project, a coalition of organizations that support inclusive and accurate population counts.

Another expert cautioned against making a comparison at all at this point. 

Not everyone has access to the internet to respond online, enumerators have not been sent out nationwide because of the coronavirus, and the counting of people in group dwellings has become confusing as prisoners are being furloughed and students are kept off campuses, said Cynthia Buckley, a sociology professor at the University of Illinois, Urbana-Champaign, and a Census enumerator. (The Census Bureau says college students should be counted where they live and sleep most of the time as of April 1. "For most students, that means in their college town, not back home with their parents.")

Featured Fact-check

Michele Bachmann

stated on June 17, 2009 in a Washington Times interview

Says the Constitution only requires her to tell the census "how many people are in our home."

Zehyoue said this part of the claim was about the 2020 census and compared with white (non-Hispanic) households, not a comparison of response rates in 2020 and 2010. (In a July 22 MSNBC interview, Abrams also mentioned the 10% and said it was in comparison with "white communities.")

Zehyoue again cited the CUNY center’s analysis.

The center examined patterns of response rates based on census-tract-level population characteristics, using the latest American Community Survey estimates (2014-18), to gauge how well communities with historically undercounted populations are self-responding in 2020 compared with other communities, Romalewski said. As of July 9, tracts with a plurality of white (non-Hispanic) households had a higher average response rate — at least 10 percentage points higher — compared with Black, Hispanic and American Indian/Alaska Native households, according to the center.

"It is pretty standard for minority communities to be classified as ‘hard to count,’ and they historically have had higher undercount rates than non-Hispanic whites," said Jennifer Van Hook, a professor of sociology and demography at Penn State University."

"The big question is whether coverage for these groups will be especially low this year," she said.

Abrams said, "Right now, we are about 4% behind the response rate where we were in 2010. But for Black and Latino and Native American communities, they’re behind by 10%."

Abrams’ staff said her point was that the national self-response rate was 4 percentage points behind the final self-response rate in 2010; and in 2020, Latino, Native American and Black communities’ self-response rates are 10 percentage points lower than in white communities.

There is one metric that supports Abrams’ point that the response rate this census is about 4 percentage points behind 2010’s and that the response from some communities trails those of mostly white (non-Hispanic) communities by at least 10 percentage points. 

Still, the Census Bureau and other experts said it’s difficult or impractical to compare response rates in 2010 and 2020 for several reasons, including different timelines, additional ways to answer the questionnaire, and an ongoing pandemic.

Abrams’ statement is partially accurate but leaves out important details or takes things out of context. We rate it Half True.