As the battle over the when the 2020 census could stop counting unfolded in her court, U.S. District Judge Lucy Koh received screenshots and testimonials from . The messages were about orders the workers were given that concerned them about how well the Bureau was actually counting their communities.
One of those messages was sent by a Southern Colorado census enumerator named Filemon, on Oct. 2. That was after Judge Koh had issued her second order for the Census Bureau to keep counting nationwide until Oct. 31. (KUNC spoke to Filemon and confirmed his identity. He asked that his full name not be provided due to concern about retaliation from the Census Bureau.)
In his email to the court, Filemon had taken a picture of a text he claims was sent by a supervisor to his Bureau-issued phone that read: âWe need to start closing as many cases as possible. So, regardless of what happens on any of your cases make one of these selections so we can get rid of cases.â
"Cases" are what the census calls housing units that workers still have to visit and try to get a count of the people who live inside, usually because they haven't filled out the census questionnaire for their address, online or otherwise. Attached to the text was a list of 21 possible ways a case can be closed, including complete, complete by proxy (which is when a census worker asks neighbors for information about the residents of a unit that isnât responding to them), âMax Attemptsâ or vacant.
Writing that she was âdisturbedâ by the text, Judge Koh ordered the Census Bureau to respond to the allegation by 8 p.m. that night. The text and order national
That all surprised Filemon. He didnât expect the court to even read his message, let alone respond so strongly.
âI wanted the facts to be considered and I wanted the judge to have more perfect information,â he said.
While he sent the email to the court after Judge Koh extended the deadline to Oct. 31 the second time, the original text was sent in the days before the order â during a time when, as he and everyone working on the ground might have assumed, the census was bringing counting to an end on Oct. 5, despite Koh's first order which also told them to keep going until the end of the month.
The Census Bureau didnât provide the response given to the court about the message at KUNC's request, citing an inability to comment on ongoing litigation. The Bureau also declined to detail exactly what âget rid of cases'' and âregardless of what happens on any of your casesâ is supposed to mean or how the text message was intended to impact operations.
âIt sounded like it was just gonna be just cutting corners and letâs just make the numbers look better,â Filemon said, explaining why he thought it was worth sharing the message.
While thatâs his impression, Filemon admits heâs not 100% sure what was being asked of him and if â or how â it differed from his regular expectations. And he never asked for that clarification.
He emailed the court because he was worried about Southern Coloradoâs (and by extension his home countyâs) count, which had been lagging behind the rest of the state. In the beginning of September, the area census offices serving Denver, Aurora and Northern Colorado were more than 70% done with the housing units they needed to enumerate. Southern Colorado, which had the largest surface area and most low-self responding counties to cover, was just over halfway done.
By the time Filemon sent that email, Southern Coloradoâs completion rate was nearly caught up with the other three regions. When counting ended, all four were 99.9% done,
Nationally (rather than a resident filling out a form themselves, which made up the rest of the 99.9% response rate.) The majority were enumerated by a census employee directly interacting with a household member.
Of the total units covered by these operations, 24.1% were enumerated by proxy and 13.9% were counted with "high-quality administrative records," like state identification records or tax documents. It's possible that those last two percentages could go down as the census discovers, for example, that some units enumerated by proxy also self-responded.
When the deadline was shortened, first to Sept. 30, and then again to Oct. 5, Filemon was worried. The Census Bureau has not responded to KUNCâs questions about how the shifting deadlines impacted its employees on the ground.
âIt felt like a scramble coming up to the 5th (of October),â he said. âThe sense of urgency started to dissipate after the ruling (to extend counting).â
Filemon worked for the Census Bureau for a little more than two months. When the Supreme Court decided in favor of allowing counting to end, he said it seemed like it was about time. On the last day of counting, he only tried to count three housing units because the cases that were still left open were all so spread out. "I didn't count anyone the last day," he said.
"Either we had the opportunity to do a thorough job or someone above my head was artificially closing cases to make everything look better," he said.
He's not the first person to express concern about how well the "Non-Response Follow Up" operation actually counted homes in Colorado and nationally.
"So the fact that the Census Bureau touts a 99% completion rate of having counted all households is simply one that not only is it hard to believe, it is really not true," Arturo Vargas said in a press conference about the census this week. He's chief executive officer of the , a national "nonpartisan" nonprofit dedicated to promoting Latinx civic participation.
"The Census Bureau may have touched 99% of the households on its address list, but we have no idea if everybody who lives in those households were included in the census," Vargas said. "We have no idea if those census forms are actually complete with all the information that was supposed to be submitted and we don't even know if that address list at the Census Bureau compiled included all addresses."
However, the Bureau has time to make up for any missed people during its data review operations, where they essentially double-check their work using some complicated math and existing records. There is also concern about that part, now that the Bureau has a severely shortened timetable to complete the operation compared to its original plan and previous decade's counts. The Bureau has said advancements in technology and operational planning will ensure the data review will be done well even in the shortened timetable, but doubts remain.
In a Bureau press conference this week, Albert Fontenot Jr., the associate director for Decennial Census Programs, also noted that they were able to watch for anyone fudging the numbers on the ground.
"We could tell on the phone, how long an enumerator took on every question, we could tell how long they took when they did the interview. We could tell where they were when they took the interview so we were sure that they were not sitting at McDonald's, but they were in front of the house they were enumerating," said Fontenot.
The Bureau also had a team of data experts watching to make sure that none of the numbers coming in from the field had "any inconsistencies or outliers," he said. "In limited cases, there were. And we sent those cases back into the field for re-enumeration. I think that's an important note that in 2020 we have that capability, in 2010 we did not have that capability."
'Little investigations'
At first, Southern Colorado enumerator Filemon said, the job was pretty easy â a lot of his cases were close to home and the people he interacted with were cooperative for the most part.
In the last few weeks, most of the units he visited were farther apart and had already been attempted by himself or other workers several times.
âItâs a lot of door-knocking and trying to do little investigations really,â he said. âNow weâre at the point where itâs a lot of people who are never home or donât want to be counted.â
âIâve started leaving little notes encouraging people to be counted,â he added.
Sometimes people can be angry or threatening. One person even pulled what Filemon believes was a handgun when he came to count their home, though he wasnât certain as he only saw it from a distance. He was careful as he approached that particular unit because he saw a âno trespassingâ sign with the image of a shotgun on it.
âI backed my car up in case somebody did come out with a shotgun or something,â he said. âI stayed 50 feet away and was just calling out trying to get someoneâs attention, âhello, hello Iâm with the censusâ. I had my car running, with the door open, ready to jump in it and gun it out of there if I had to. And yeah, that was it.â
Census workers are allowed on private property to do their jobs and are even expected to go past "no trespassing" signs and closed gates (if they feel safe doing so),
Even after that incident, Filemon chose to continue enumerating the surrounding neighborhood â the rest of whom were friendly.
âItâs the money, obviously,â he said, adding that the Bureau's higher base pay was a big part of what drew him to the job to begin with. âIâm not going to be the guy to just up and quit. And yeah, it is the passion (for the importance of the count) too. And it wasnât that scary to me.â
The Census Bureau said it takes the safety of its enumerators seriously. In 2010, enumerators were the victims of âmore than 700 reported acts of violenceâ nationwide, per , and they believe there were likely even more incidents that went unreported.
âBecause we hire locally, census takers are more likely to be assigned to areas they are already familiar with. As part of their training, we teach safety precautions and that they can alert their supervisor if they feel unsafe visiting an address that they are assigned,â the Bureau said in a statement.
The Bureau also sent out a memo to local law enforcement, Additionally, Filemon said census workers can see each otherâs case notes from previous visits, allowing them to know if there is any reason to be concerned about approaching a unit again.
In Northern Colorado, census enumerator Noah Moyer said he has felt worried about his safety several times too, but it never rose to the level of anything directly threatening.
âThere (are) a lot of really aggressive (trespassing) signs,â he said. âIt's like OK, I don't really want to go up that driveway. But most of the people I talked to when they were home were fine with me being there, which was nice but there were definitely some houses that I didn't interact with anyone and I'm glad I didnât.â
Moyer only worked for the census for about six weeks ending in mid-September. He counted units in Golden, Denver and some surrounding mountain communities based on where the censusâ enumerator app (loaded onto a Bureau-provided phone) told him to go.
âWe kind of live under the app,â he said.
Unlike Filemon, he doesnât remember the shortened deadlines having much of an effect on his work. On some of his last days with the Bureau, he would put in availability to work hours but would not get any cases. He doesnât know why for sure, but suspects it was because there just wasnât enough work left in his immediate area.
But like Filemon, the cases Moyer was getting toward the end were people who already had been visited by an enumerator before. Some, he said, had as many as six or more attempts.
âWell towards the end, especially by the time we just got down to the people that generally didn't want to do the census, that's where it especially started to get emotionally taxing,â he said. âPeople would answer their door and just wave me off or slam the door in my face or tell me to stop coming and it's like, âOK, I don't think you really understand what's going on here. This is required by law to complete. I'm not just going to stop coming.'â
Around that time, Moyer also found himself often interacting with housing unit residents who only speak Spanish.
âI don't speak Spanish and so it was just frustrating to go to houses and have someone answer the door, but not be able to complete it just because of the language barrier,â he said.
Sometimes, he said, he was able to communicate in âbroken Spanglishâ just barely well enough to get a basic headcount of the residents inside. Other times, he was far less successful, especially when he was expected to get information about residents from their Spanish-only speaking neighbors, which he said the census case app told him to do after the third failed attempt to make contact.
âI just hope those people still got counted,â he said.
When asked by KUNC about enumerating non-English speakers in Colorado, the Bureau only provided the following statement: âIf the census taker does not speak the language of the household resident, they can connect them with someone who does, or the census taker can give them a phone number to call and respond in that language over the phone.â
They didn't answer questions about who census takers could âconnectâ residents with or how many multi-lingual and Spanish-speaking enumerators were employed in Colorado.
Wildfires have âminimalâ but unclear impact
Wildfires are a risk factor for an accurate count, noted in September.
Despite the presence of the Cameron Peak, Middle Fork and Mullen wildfires raging in the region and causing widespread air quality issues in the northern half of Colorado, the Census Bureau said it was it had to count.
Workers were still enumerating in many areas where air quality dipped, but the Bureau said counting of more than 1,000 housing units in the region was put âon holdâ due to the wildfires. Enumerators had âsince started counting (the units).â
âThe health and safety of census employees is always at the upmost importance,â the Bureau said in an emailed statement. They added that wildfires âminimallyâ impacted progress.
The Bureau did not provide requested details about where the 1,000-plus units were, when those units were put on hold and when counting was resumed. It also only responded to questions about how it has counted areas with mandatory and voluntary evacuations in Larimer County by pointing to the for that region.
