Shortened Census Led to an Incomplete Count in Some Areas

This story originally appeared on Stateline.

The early end to the 2020 census has some areas complaining they needed more time to count residents in a chaotic environment of coronavirus shutdowns and storm evacuations.

Parts of Louisiana and tribal lands in Arizona, New Mexico and Utah face the biggest gaps in the pandemic-shortened count. But Manhattan in New York City and parts of New England, Mississippi and North Carolina also fell short of average completion rates. The count in those areas will have to be finished using estimates from administrative records and other techniques.

The Census Bureau stopped accepting responses on Oct. 15 after the U.S. Supreme Court ended court challenges seeking to extend the count another two weeks. The Trump administration, after initially asking Congress for more time because of the pandemic, did a turnabout in July and asked to speed up the count and exclude people living here illegally from congressional apportionment.

The Census Bureau claimed a 99.98% completion rate nationally when the count ended Oct. 15, but some areas did not reach that level. Local leaders in places where there was an undercount complain that they could lose federal funding and congressional representation, both of which depend on the census.

The census error rate has improved in recent decades, from a 1.6% net undercount in 1990 to a 0.49% net overcount in 2000, to a net overcount of one-tenth of 1% in 2010. But even in 2010, the Black population was undercounted by 2% and Hispanics by 1.5%, the bureau found. Native Americans living on reservations were the most undercounted, by 4.9%.

“It is disappointing to hear numbers are coming in lower than anticipated,” said Shreveport Mayor Adrian Perkins, a Democrat. “Considering all the challenges we have faced in 2020, any extra time for the census would have been helpful.”

Parts of Louisiana, raked by storms just as the count was wrapping up, took the biggest hit in follow-up visits needed to count people who didn’t respond voluntarily. The western part of the state, which the bureau calls the Shreveport Census Office and includes 32 counties north of Lake Charles, finished only 93.9% of its follow-up work, according to census statistics.

The Navajo Nation, which covers parts of Arizona, Utah and New Mexico, was the only other area where the percentage of follow-up work completed was less than 99%, though just barely. COVID-19 outbreaks forced a shutdown of census operations in the area from March to September, and the Nation joined the lawsuit seeking to stop federal officials from shortening the census.

Those suing for more time included cities and counties in California, Illinois, Texas and Washington, as well as the Navajo Nation and other tribes.

The lawsuit forced a two-week extension of the count, which had an especially big effect on the Navajo Nation and Louisiana, according to a Stateline analysis of census statistics. In those two weeks, the follow-up completion rate in Navajo Nation rose from 77.6% to 98.9%, while the Shreveport area improved from 75.3% to 93.9%, the biggest point changes in the country.

In 14 other areas — in Alabama, Colorado, Georgia, Mississippi, North Carolina and South Carolina — the follow-up completion rate improved by at least 10 points.

“It’s disheartening that the highest court of the land ignored the devastating impacts that COVID-19 has had on the Navajo Nation and the census count,” he said. “The Navajo Nation fought to the very end and never let up.”

The tribe now must work with the bureau to use administrative records to make up the counting gaps, which were highest in the Chinle area at the center of tribal lands in Arizona, Nez said in his joint address with Vice President Myron Lizer.  

Among the challenges in Navajo Nation: During the shutdown, residents were asked to respond by phone or internet, but many lack both phone and internet service, said Tiffany Bah Charley, a spokesperson for the tribe’s Human Rights Commission. Language barriers also made it hard to respond by mail without help, she said. 

At a press briefing on Oct. 21, census officials said they worked frantically to concentrate resources in lagging areas during the final weeks of the count, and said they were hampered by storms and other issues. But they also cited the advantages of new technology that sped up the process compared with 2010.

In late August, Hurricane Laura shut down the count in the Lake Charles area for five weeks, and the bureau spent the time tracking down evacuated people in other parts of Louisiana, using state-provided shelter lists, to ask if they needed to be counted, said Tim Olson, a Census Bureau associate director, in the briefing.

When the area opened for counting, thousands of census workers from around the country descended on the Lake Charles area for follow-up interviews.

“We had a solid three, three and a half days,” Olson said. “And then Hurricane Delta came in and shut things down.” At the last minute some administrative records were added for homes that couldn’t be reached, and that boosted the completion rate for Louisiana, associate director Albert Fontenot Jr. said at the briefing.

“Southwest Louisiana was certainly the most impacted in the nation in terms of those hurricanes and what we had to do to complete the workload,” said Olson.

Louisiana was the only state with less than 99.9% of households accounted for, either by self-response or follow-up work.

“Despite the 99.9% [national completion rate] number, we just don’t know yet how successful the census was,” said Steven Romalewski, director of the City University of New York’s mapping service, who tracked completion rates on an interactive map.

“We also don’t know whether an extra two weeks would have or would not have made a difference. But we do know that the Census Bureau staff issued some pretty strong warnings internally that they needed more time as they planned to finish the count,” Romalewski added.

An internal assessment by the bureau in August warned that the fast-track schedule “created risks for serious errors” that “may not be fixed due to lack of time to research and understand the root cause.” The bureau asked Congress for a time extension to turn in state totals, due by law at the end of the year, but did not get it.

The good news for the count is that internet technology worked smoothly and helped speed follow-up work, the bureau’s Fontenot stressed at the October briefing.

Unlike 2010, Fontenot said, “the computer figured out the most efficient route for cases and put them in that order at the best time of day to enumerate that particular household.” A new option for responding on the internet went smoothly too, avoiding more follow-up headaches.

“We heard from a lot of critics that our internet response options would fail and we would become Healthcare.gov II,” Fontenot said, referring to the widely criticized rollout of an Obamacare website in 2013. “Our internet self-response option successfully managed our highest traffic demand.”