A day after the government released the first results from the 2020 Census, some states and civic organizations were reeling from unexpected results, and wondered if the differences between projections and actual data might be an indicator of problems with the count.
Decennial census data is used to determine the apportionment of House seats, redistricting and $1.5 trillion a year in federal funding, so the release of its data is always closely scrutinized. But this time, perhaps more than ever, the count faced unprecedented hurdles. They included underfunding, attempts by the Trump administration to add a citizenship question and exclude undocumented immigrants from apportionment, the coronavirus pandemic, and natural disasters that struck just before the count ended.
Former attorney general Eric Holder, chairman of the National Democratic Redistricting Committee, raised the possibility that the lower-than-expected counts in Florida, Arizona and Texas could be due to an undercount of Hispanics resulting from the Trump administration’s “shameful” handling of the decennial census, including its attempts to include a citizenship question. “I just wonder if it had the impact of suppressing the count,” Holder said.
But until more-detailed 2020 Census data showing race, ethnicity and geography is released later in the year, it will be hard to confirm any suspicions of tallies of Hispanic residents being under par. The data released Monday shows no relationship between a state’s share of Hispanic residents and whether the state population tally turned out to be lower than projected.
“Nothing looked terribly outside of expectations or historical patterns,” said Chris Dick, founder of DA Advisors, an analytics consulting company, and former Census Bureau statistician and branch chief. Noting that the Arizona data in 2010 were also lower than estimates, he said, “I think we have to be careful. I don’t think we have enough information to say the census was flawed, but I don’t think we have enough information to say the census was a success.”
Not all states with a high percentage of Hispanics had an undercount, and many states that were undercounted have relatively small Hispanic populations. For instance, while Texas and Florida, which are 39 percent and 26 percent Hispanic, respectively, had undercounts, New Mexico, which is nearly half Hispanic, did not. Nor did California, which is 39 percent Hispanic, or New York, which is 19 percent Hispanic.
But Arturo Vargas, chief executive of NALEO Educational Fund, pointed out that New Mexico, California, and New York invested state funds in outreach efforts, while Florida and Arizona did not. Texas did outreach, too, but only at the last minute.
Vargas said that given all the obstacles the decennial count faced, “it’s too much of a coincidence for me for them to report that this is the lowest growth," he said. “It’s too many coincidences to not look at these numbers with some healthy skepticism.”
Besides Arizona, the largest undercounts were in places — D.C., South Carolina and North Carolina — where Hispanics make up a smaller share of residents: 11 percent, 6 percent and 9 percent, respectively.
Census counts for the majority of states came within 1 percent of what was projected. But in Arizona and the District, the final count was 3 percent less than projected. On the other end, New York and New Jersey counts came in more than 4 percent higher than what was estimated.
Discrepancies always exist between estimates and actual data.
In a preliminary report on quality metrics released Monday afternoon, Census Bureau demographers said the initial population counts from the 2020 Census were “generally aligned with benchmark data.” Nationally, the count came in 0.6 percent over, compared to 0.1 in 2010.
In 2010, slightly more states had counts closer to projection, but bigger outliers existed. The count was 3.9 percent under in Arizona and 4.9 over in Hawaii.
The Census Bureau stated that “degree of difference is largely consistent with what we have seen in the past” and preliminary analysis should not be taken as “an assessment of the accuracy or reasonableness of the 2020 Census results.” Further assessments are expected to follow.
Census watchers may be more cautious this time around with discrepancies, said Ken Hodges, a demographer with Claritas, a consumer information company.
“People have gone into this with legitimate concerns because of all of the challenges that the 2020 Census faced," he said. "We’ve always looked to the count as the standard as to how accurate our estimates have been, and we will still use the census count, but a lot of us are wondering if the census counts are still that standard of what we’ve come to expect.”
But he also noted that estimates are also subject to error, “and one wonders, ‘What was the problem with the count?’ whereas one could wonder, ‘What was the problem with the estimate?’ ”
George Hammond, director of the University of Arizona’s Economic and Business Research Center, said that “census population estimates tend not to be particularly close, even in percentage terms, to what the actual population count might be.”
In the District, city officials said they would take a closer look at the data. Andrew Trueblood, director of the D.C. Office of Planning, said Tuesday that it is too soon to say exactly what the city’s investigation will look like, although he said one question it would attempt to answer was whether a significant number of college students were not counted because they did not return to their campuses during the coronavirus pandemic.
The self-response rate for D.C. residents was 64 percent, according to data provided by the Census Bureau, lower than the national average of 67.
“We know we already had a population that was more likely to be undercounted, and with covid, even more likely. Some people maybe weren’t in the places they would normally be. Students is a great example. … we’ll need to look at all those subsets of populations,” Trueblood said.
Apportionment numbers have faced legal challenges in the past, though none have succeeded. In 2001, Utah sued the government over its refusal to count Mormon missionaries or other nongovernment American workers living overseas. The policy cost Utah a chance to gain a House seat after the 2000 census; it missed out by just 857 people.
The 2020 data had an even more dramatic miss: New York lost a seat by a margin of just 89 people. Gov. Andrew M. Cuomo (D) said Tuesday that his office is exploring whether it can contest the census’s figure for the state’s population.
“The distribution of congressional seats is really a zero-sum game,” said Terri Ann Lowenthal, former staff director of the House census oversight subcommittee. “It is sensitive to small shifts in population, whether it’s New York losing a district by a small number of people or Minnesota retaining its last seat by an even smaller margin. Any change in state numbers would reshuffle the outcome in unpredictable ways.”
Jason Torchinsky, the chief counsel of the National Republican Redistricting Trust, said when asked Monday about legal challenges to the topline apportionment numbers: “There is always the possibility of lawsuits over the census. The issue is going to be who is harmed and what data do you have and who has standing to bring these challenges. And frankly I think right now anybody who is looking at those doesn’t have enough data to figure out what is there.”
Brittany Renee Mayes, Michael Scherer and Julie Zauzmer contributed to this report.