In August, Tom Friedmann grew puzzled when his Census Bureau-issued iPhone labeled every resident in his 20-unit condo complex in Massachusetts as “NRFU.” That’s census shorthand for “nonresponse followup,” or people who had yet to fill out the decennial questionnaire. 

Several of his neighbors insisted they’d completed their forms. One even produced proof: a confirmation message that explicitly urged her to disregard subsequent materials she might later receive. Still, the app prompted Friedmann, who went door to door counting people for the bureau, to tally these residents again.

This pattern repeated itself. Over the next several weeks, a handful of other Census Bureau door knockers – known within the bureau as enumerators – continued showing up to count the same residents, Friedmann said. The app eventually indicated that he, too, was marked NRFU, even though he was certain he’d filled out his paper form and mailed it in the spring. He texted his supervisor to alert her of the problem. 

“Tom, that is happening with most, if not all E’s,” she wrote back, referring to fellow enumerators. “I completed mine on line, Enumerated myself and still had an E visit and completed it with her.”

Friedmann quit a publishing job in 2019 and started working for the census out of a sense of civic duty. The data he helped gather will be used to help reapportion seats in the House of Representatives and redraw congressional and legislative districts. It will dictate how $1.5 trillion per year in federal funding will be allocated for programs such as Head Start, Medicaid and Medicare over the next decade. 

Friedmann said he cares deeply about all these things. Yet his experience also left him exasperated. 

“The thing that bothered me about the duplication of efforts was that it indicated to me there was a problem of certainty – of knowing whether someone’s info had been counted or not,” he said. “When I was in the field, I wondered, if this was happening to enumerators, who else is not being counted?”

Frustrations like Friedmann’s were common among nearly 100 census employees across the U.S. who responded to Reveal from The Center for Investigative Reporting’s ongoing callout seeking their experiences. (Reveal verified the employment of everyone quoted in this article). Facing a cascade of shifting deadlines, workers said poor communication and training, pressure to end the count early, clunky technology and haphazard management practices produced chaos on the ground as the census came to a close Oct. 15. 


We’re staying on this story. Work for the Census Bureau? We want to hear about how it’s going.


This year’s count was different from in years past for a number of reasons: It was the first to have been conducted with smartphone technology. It was significantly delayed by the COVID-19 pandemic. And work to follow up with households, which was supposed to take 11 weeks, ultimately was done in nine: The Trump administration abruptly decided to speed up the process in an effort to deliver population totals to the president by the bureau’s statutory deadline of Dec. 31.

In August, Commerce Secretary Wilbur Ross, who supervises the census, insisted that the truncated operation “will meet or exceed the standard for data collection set in previous decennial censuses.” But the inspector general’s office for the Department of Commerce concluded in a report released a month later that a shortened schedule “increases the risks to the accuracy of the 2020 Census.” The same report pointed out that “Senior officials at the Bureau, including the Director, did not know who ultimately made the decision to accelerate the Census schedule.” 

On Nov. 3, the inspector general’s office took a new step: It notified the bureau that it was planning to “assess the adequacy” of its quality control processes to make sure data collected during the decennial count was “complete and accurate.”

Census Bureau leaders have claimed that 99.98% of U.S. households were accounted for. But experts, including a former Census Bureau director and a number of leading researchers, have pointed out that “accounted for” is not the same as “counted” and that the 99.98% figure means little until the bureau releases more information about its data quality. 

“The Census Bureau was able to meet and overcome many challenges because of our innovative design and use of new technology, but it could not be done without the unflinching resolve of our staff,” Bureau Director Steven Dillingham said in a press release.

Meanwhile, almost two dozen staff members told Reveal that they shared many of the inspector general’s concerns. Once idealistic about their mission, they are now raising doubts about the accuracy of the information they collected and expressing worry about how government funding and political representation may be distributed as a result of their efforts. 

Some have vowed to never work for the census again.

“I doubt the count is going to be as accurate as it could have been and should be,” said California enumerator Nancy Salerno. “I feel the current administration has been, to be blunt, fucking with the timeline according to an agenda that makes me nauseous.”

‘I don’t really think the system was fine-tuned as it should have been’

Experiences like Friedmann’s show how Census Bureau miscommunications snowballed into crippling confusion on the ground. Roughly two dozen workers told Reveal that the bureau’s technology sent them back to the same addresses again and again, even as residents insisted they’d completed their questionnaires. None knew why it was happening or what the proper protocols were. With their managers just as perplexed, some enumerators were forced to improvise.

Heather Candon, who worked as a door knocker in Cold Spring, New York, estimated that 60% of all the addresses she visited in September were classified as NRFU, even though residents claimed over and over that they’d already finished their census forms. Some of her neighbors became so annoyed by the repeat visits that they started placing notes on their doors, imploring workers to move on. 

“Even when they weren’t home, they were trying to communicate to us, ‘Stop it!’ ” Candon said.

Workers across the U.S. used a series of divergent strategies to deal with this issue: Candon’s manager attributed repeat visits to a “glitch in the system” and stopped sending enumerators out, she said. One Georgia-based door knocker, Lisa McGee-Shields, said her manager told her to indicate these addresses didn’t exist – an approach the bureau said should be used only for empty lots. After three weeks of pursuing that strategy, she received a new directive: “Enumerate everything.”

“I don’t believe the Census Bureau was on top of what they were supposed to be doing,” said McGee-Shields, who told us she had done census work since 2000. “I don’t really think the system was fine-tuned as it should have been. Before they rolled it out, they should have worked on it all 10 of the years.”

A census worker at the Lincoln Center for the Performing Arts’ plaza in New York in September 2020. Credit: Alexi Rosenfeld/Getty Images Credit: Alexi Rosenfeld/Getty Images

In an Aug. 31 blog post, Dillingham offered three potential reasons why residents who have already responded might see enumerators at their doors for “two visits”: They may have filled out the questionnaire using their address, not their census ID; there may be two addresses referring to the same housing unit; or enumerators may conduct a second interview for quality assurance purposes. In a later press release, the bureau quietly changed its wording to “multiple visits” and acknowledged that formatting issues in its so-called “master address file” could be responsible for duplicates.  

Every census operation contains duplicate, incorrect and even fraudulent responses. And ordinarily, there’s ample time to catch some of those errors. In 2010, there were 8.5 million duplicates and 1.5 million erroneous inclusions in which people should not have been counted or were included in the wrong place. But this year, to meet Trump administration deadlines, that phase has been dramatically abbreviated: It’s now 12 weeks shorter than its average length over the last three decades, according to the Beeck Center for Social Impact and Innovation at Georgetown University.

The Census Bureau has claimed that this timeframe is possible thanks to new technology that has streamlined processing. But some experts disagree.

“Given the complexity of this count, it is unlikely they would be able to cut last decades’ time in half,” said Denice Ross, a fellow at the Beeck Center. “Sure, they have new tech, but that technology does not obviate the need for the quality control measures they’ve cut to meet the deadline.”

According to the Beeck Center for Social Impact and Innovation at Georgetown University, the data processing period for the 2020 census is 12 weeks shorter than its average length over the last three decades. Courtesy of the Beeck Center for Social Impact and Innovation Credit: Courtesy of the Beeck Center for Social Impact and Innovation

Other experts point to the long-term impact these cuts could have.

“Rushing the census data processing period will hurt everyone and could mean missing dangerous flaws in the data,” said Vanita Gupta, president and CEO of the Leadership Conference on Civil and Human Rights. “We have to get this census right, because the numbers will guide allocation of critical economic and pandemic recovery resources to states and localities, as well as private-sector investment to spur economic development, for years to come.”

At the mercy of ‘Bounty Hunters’

Brookes Giller was working as a field supervisor in the Los Angeles area when he received several lists of workers to terminate. On each list, there was a column naming which supervisor was responsible for the termination – under the title “Bounty Hunter.”

As he began making calls, workers told him they’d tried to complete their training but were unable to access the proper resources or take tests: Some had trouble with the bureau’s website; some said their login credentials never worked. Others told him they’d already finished the job and handed in their gear but were being called repeatedly by supervisors anyway. Some weren’t even enumerators; they had no idea how they’d ended up on the list. And in at least one case, Giller said, a fellow supervisor’s entire team was targeted for termination in error. He’s not sure whether they were ultimately let go.

The bureau has received 230 worker complaints related to terminations as of Sept. 30, according to Equal Employment Opportunity Commission data – more than for any other issue. Enumerators could be fired for accruing overtime (or, in some instances, refusing to work overtime), for working fewer than 20 hours per week, for completing too few cases per day or for refusing to work weekends, they said. The Census Bureau, for its part, maintains that there is no minimum hourly requirement for enumerators’ work.

Racking up terminations became a competitive pastime among his colleagues, Giller said. They talked gleefully about their pursuits and boasted about ignoring enumerators’ pleas to keep working – all while earning overtime that was ordinarily granted sparingly.

“People work for the census as a lifeline,” Giller said. “And then to hear other people talking about it like it’s kind of a game – like, ‘Who can bring in the most things?’ … It didn’t really feel up to any professional standards.”

The bureau was poised to begin a round of layoffs after it announced it was shutting down the count early. It agreed to pause those layoffs when a federal judge on Sept. 5 ordered it to stop winding down operations.

But soon thereafter, supervisors in at least one regional census office received explicit directions to process staff departures as resignations, not terminations. An internal document from this office, titled “NEW EXITING PROTOCOL,” instructed supervisors to intentionally refrain from filling out a section marked “type of personnel action.” It also contained clear instructions: “DO NOT DATE DOCUMENTATION;” “ALL ENUMERATORS MUST FILL OUT A RESIGNATION LETTER. (DO NOT DATE RESIGNATION LETTER).”

The Census Bureau affirmed that “Bounty Hunter” is not an official term and that the “exiting protocol” instructions, as written, are not official Census Bureau procedures. The bureau did not address why such directives might have been given.

‘Why did I have idle enumerators?’

Christopher Smith, a census field supervisor in Missouri, began to worry when the census hit its second phase of the follow-up operation. That’s when cases with multiple attempts are assigned to better-performing enumerators to mark those households as “closed” – meaning they successfully interviewed a resident and recorded the information of a housing unit. At first, he was told to keep his team of enumerators – the bureau might reopen 6,000 cases for addresses they’d been unable to reach. 

It was welcome news for Smith, who believed his team should keep door-knocking up until the count’s final moments. Yet in mid-October, he received a new directive: Pick up your enumerators’ equipment; the count was over. He never received an explanation about what happened to those cases. He fears they were closed improperly, without the recommended number of attempts.

“We started letting people go before it was even clear when the end of the operation was going to be,” he said. “Why did I have idle enumerators who weren’t being assigned cases, who were sitting on their asses and resigning from their positions, when there was 6,000 cases?”

Aly, a Texas enumerator who requested to be identified only by her first name, recalled a similar sense of whiplash. As court deadlines zigzagged back and forth, she and fellow enumerators rushed to complete as many cases as possible. They were told to close those cases “no matter what,” even though they couldn’t reach many residents in time. When the deadline was extended, she said, these cases were never reopened. 

“Then, two days before the second (or third?!) extension ended in mid October, I was told suddenly there were 1500 cases that needed to be done ASAP,” she wrote. “They hadn’t been done correctly and needed to be redone. Not two days later, presumably still with hundreds of these cases not done, the final extension was overturned and we had to cease all operations.”

The Census Bureau requires census takers to wear a mask while conducting their work. Workers voiced worries about counting tactics as government-issued deadlines contracted. Credit: U.S. Census Bureau Credit: U.S. Census Bureau

On Oct. 27, a coalition of advocacy groups and local governments filed an amended suit over similar concerns. Enumerators claimed they were pressured by supervisors to cut corners while “clos(ing) down remaining cases by whatever means necessary,” according to the complaint. A field supervisor in Baltimore echoed Smith’s worries directly, claiming that “thousands of cases were manually marked completed without explanation” and removed from the local caseload after only one attempt. Another census field supervisor in San Francisco was told to take “accuracy reducing shortcuts to get the work completed ASAP.” And in North Carolina, enumerators were instructed to “guess how many people lived in a housing unit without any input from someone at the house or someone otherwise familiar with the housing unit.”  

Census workers who filled out Reveal’s questionnaire voiced other worries about counting tactics as government-issued deadlines contracted. Cases were closed based on estimates of cars or bikes parked nearby, one enumerator said. Multiple sources told us they were instructed to not interview homeless people, but rather count them from a distance. Another enumerator claimed that while performing a population count at night in rural Oregon, he was told to assume RVs and campers contained only one person, unless he saw clear evidence suggesting otherwise.

“Obviously, this likely produced wildly inaccurate results,” he wrote. “There is no reason this couldn’t have been done in the daylight.”

The Census Bureau said cases for addresses it deems occupied should be closed only in one of three ways: through an in-person interview with inhabitants, through an interview with another knowledgeable source or through administrative records.

Despite the problems and concerns that workers said they experienced on the job, many were clear that working on the census was important to them. They said they felt frustrated that they couldn’t do the best possible job.

“I think that people should understand that some of their poorest neighbors – the people that really don’t have a lot of time and resources available to help serve the community – just did their part to make sure that we all have proper representation in the government,” Christopher Smith said. “There were not very many wealthy, comfortable people who were doing this because they cared about the community – at least not on my team.”

This story was edited by Sumi Aggarwal and Andrew Donohue and copy edited by Nikki Frick.

Byard Duncan can be reached at bduncan@revealnews.org, and David Rodriguez can be reached at drodriguez@revealnews.org. Follow them on Twitter: @ByardDuncan and @DaveeJonesLock.

Byard Duncan is a reporter and producer for  engagement and collaborations for Reveal. He manages Reveal’s Reporting Networks, which provide more than 1,000 local journalists across the U.S. with resources and training to continue Reveal investigations in their communities. He also helps lead audience engagement initiatives around Reveal’s stories and assists local reporters in elevating their work to a national platform. In addition to Reveal, Duncan’s work has appeared in GQ, Esquire, The California Sunday Magazine and Columbia Journalism Review, among other outlets. He was part of Reveal’s Behind the Smiles project team, which was named a Pulitzer Prize finalist in 2019. He is the recipient of two Edward R. Murrow Awards, a National Headliner Award, an Al Neuharth Innovation in Investigative Journalism Award, and two first-place awards for feature storytelling from the Society of Professional Journalists and Best of the West. Duncan is based in Reveal's Emeryville, California, office.

David Rodriguez is a community engagement producer for Reveal. Before joining Reveal, Rodriguez's work as an engagement assistant producer at Southern California Public Radio helped develop a report on how newsrooms can improve their reporting on the 2020 Census, which won the 2019 Gather Award in Engaged Journalism. 

Rodriguez has reported stories on immigration at the Investigative Reporting Workshop in American University. He is an alum of NPR's Next Generation Radio and San Francisco State University. He previously completed internships with KPCC's podcast team, where he helped produce The Big One: Your Survival Guide, and with Reveal, where he created a database tracking how much money and time the United States government has spent buying land along the U.S.-Mexico border. He is based in Los Angeles.