www.bloomberg.com /news/articles/2020-10-15/top-statistician-u-s-census-is-being-sabotaged

Top Statistician: U.S. Census Is Being ‘Sabotaged'

By Kriston Capps
15-19 minutes

Census 2020 employees help a New Yorker fill out a census form at Sylvia's Restaurant in Harlem. A host of setbacks in this year’s census count could result in a severe undercount of hard-to-reach populations, experts warn. 

Census 2020 employees help a New Yorker fill out a census form at Sylvia's Restaurant in Harlem. A host of setbacks in this year’s census count could result in a severe undercount of hard-to-reach populations, experts warn. 

Photographer: Lev Radin/Pacific Press/LightRocket via Getty Images

Photographer: Lev Radin/Pacific Press/LightRocket via Getty Images

Every ten years, as the U.S. Census Bureau completes its constitutionally mandated count of the population of the U.S., it faces new obstacles. But the 2020 census has weathered some extraordinary blows, from challenges rolling out new technology to political machinations over the questionnaire itself.

The coronavirus pandemic forced the Census Bureau to extend the window for going door-to-door to count households that had not yet responded. Instead of wrapping up operations in July, the Census Bureau — with the support of the White House — set a deadline of Oct. 31. The agency also asked Congress to move the date for submitting the final count from December to April 2021.

Then the ground shifted. In July, the White House decided it wanted the data sooner, announcing that the final count was due in December after all, and that nonresponse follow-ups should wrap up early to give the agency time to finalize the data — five months earlier than the Census Bureau requested.

“It was a bombshell. The White House determined that they wanted the count before the end of the year,” says Robert Santos, vice president and chief methodologist at the Urban Institute and the president-elect of the American Statistical Association.

The abrupt change of plans drew a legal challenge led by the National Urban League, which proceeded through the courts even as Congress weighed a bill to give the Census Bureau more time. On Oct. 13, the U.S. Supreme Court allowed the Trump administration to stop the Census Bureau’s count immediately, two weeks before the deadline.

Santos — who co-chaired an American Statistical Association task force on the issues facing the 2020 Census with former U.S. Chief Statistician Nancy Potok — says that the decision is just the latest disaster to befall the census. The population count is responsible for making decisions about trillions of dollars in federal funding. And Santos did not mince words about how distorted the undercount could be.

“I do not believe that a fair and accurate census can occur,” he says. “I expect it to be one of the most flawed censuses in history.”

The court decision alone is not responsible for this looming disaster, Santos says. Any undercount will be the result of a “perfect storm” of factors, starting with a call to add a citizenship question to the census that was rejected by the Supreme Court in 2019. In subsequent months, President Donald Trump asked the Census Bureau to use administrative data to determine the citizenship status of every American adult and issued an executive order to exclude undocumented immigrants from the apportionment of congressional districts. Plus, the White House installed several hand-picked appointees inside the Census Bureau, leading to internal fears of political interference. Critics said that the administration was fulfilling a secret plan by the late Thomas Hofeller, the GOP’s so-called “Michelangelo of gerrymandering,” to use the citizenship question to give a boost to Republicans and non-Hispanic whites.

Then came the pandemic.

At this point, only new legislation (or litigation) could undo some of the harm that has been caused, Santos says. His faith in the 2020 count was already shaky: Last year, the Urban Institute produced a state-by-state model for three different undercount scenarios: low risk, medium risk, and high risk. All three of those scenarios showed severe undercounts in Western and Southern states with large Black and Latinx populations.

That was last year. Santos spoke with Bloomberg CityLab  to explain why he’s lost all hope for the 2020 Census. The conversation has been edited for clarity and length.

What is your outlook on state of the 2020 Census?

All that has happened has been tragic and catastrophic. We’ve had repeated threats to the conduct of the census through things like the citizenship fracas and an edict to create a citizenship count. Then the Covid-19 pandemic hit. That caused such a disruption that the Census Bureau had to suspend operations for a long period of time. It requested Congress to extend the dates for submission of counts. The White House joined in that request at first.

Congress was in the process of doing exactly that, but two other things happened. The first was the executive order or memorandum to create counts of undocumented immigrants so they could be removed from official census counts. Which to me, based on my reading of the Constitution, is unconstitutional, but I’ll leave that to the lawyers.

The second one is that the White House changed its mind and required that the counts to be submitted in September. This means [it eliminated] literally a third of the time allocated for the Census Bureau to knock on doors, to try to enumerate the hard-to-count individuals, which happen to be African Americans, Latinx populations, immigrants, people in heavily transitory situations, tribal lands, some Asians.

No one I talk to thinks that a fair and accurate census can occur. I do not believe that a fair and accurate census can occur. I expect it to be one of the most flawed censuses in history. It will have to, by necessity, because of time, rely on exceedingly tenuous imputations. There are going to be a lot of households where enumerators do not get anyone to answer the door. The immigrant population is just terrified — the public policy climate is such that it’s not conducive to immigrant populations feeling a civic duty to participate in the census, because they believe and they fear that the information will be used against them.

Last year, the Urban Institute released a model that showed outcomes for every state based on different risk scenarios for an undercount — low risk, medium risk and high risk. Which of those risk categories are we in now?

Those risk categories are on the Earth. Where we are now is Mars.

The high-risk category for us was based on a non-Covid world. No one knew a pandemic was coming. No one knew the degree to which additional threats or exacerbation of the policy climate was going to occur. The high risk that we provided in terms of potential undercounts doesn’t even make it to low risk in the current day because of all the threats to the census.

Even if the Census Bureau pulls a rabbit out of the hat and comes close, it is certain that there will be differential undercounts and overcounts. White suburbs and the elderly are going to be way overcounted. Minority communities and immigrants are going to be way undercounted.

The job of the Census Bureau got incredibly harder.

What effect will the Supreme Court decision have on the count?

The Supreme Court decision basically struck down the order by the lower court to extend data collection to the planned Oct. 31 deadline. What that does is force the census field operation to stop running on a dime. Thursday [Oct. 15] is the last day that people will be able to either return their forms on their own or answer a door knocker who comes to enumerate the house because they haven’t sent in a form yet.

The implications are pretty profound. Already there have been innumerable challenges to conducting the 2020 census, with the economy, Covid, racial injustice protests, upheaval — on top of the administration flip-flopping on when it wanted to have the counts. Now that the Supreme Court ruling has been rendered, the field work is stopping almost immediately. The people that are the hardest to count simply won’t get counted more severely.

On top of that, because of the mandate to submit the census counts to the president by the end of December, there simply will not be enough time to do adequate quality checks. The usual processing that should have taken five months is now [happening in] a matter of 10 weeks or so. That’s going to have a major impact on the quality of the counts. It’s going to suggest that the areas and neighborhoods and cities that had high very self -response, where people participated, those are going to be counted pretty adequately. Communities of color and immigrant communities that are historically harder to count, including people who are renters, tribal lands, and so forth — they’re basically not going to be counted.

My greatest fear is that not only is there going to be a severe undercount of specific peoples, it’s going to be very severe.

Do you have any sense for how many people might have been counted had the census efforts continued to Oct. 31?

I would not be in a position to quantify it. You could take a look at the self-response rates that the Census Bureau publishes and look at two-week intervals, going backward, to see what would have happened had they continued for another two weeks.

It’s not going to be much. If you’re hard to count, you’re hard to count. The people who are hardest to count, the ones we worry most about in terms of undercount, are the same folks whose families don’t have food at night or don’t have a job. They’re worried about getting tossed out because they can’t make the rent. Or they’re ill and don’t have health insurance. Those folks tend not to have participating in the census at the top of their minds because they have basic needs that have to be met — for themselves, their children and sometimes for their parents and grandparents.

Because of those things that those people have to deal with, you actually could have extended the time for another month on top of that and you’d only get marginal gains. The census was already in a tough position. It would definitely help to get as much time as possible, because as people become acquainted with how to deal with Covid, they can do the grassroots effort that’s needed to get people to participate. But it’s going to be a marginal increase. At the end of the day, the die has been cast for quite a while: There are going to be undercounts of communities of color. The only question is how large — and is it so large that action needs to be taken in the form of litigation or legislation.

Are you saying that it was actions and decisions by the White House, not the Supreme Court, that could threaten the census count?

It’s not just the White House. It’s everything combined. It’s the perfect storm of obstacles and challenges, when you start putting together everything from Covid to the economy to the toxic immigrant policies of the year before with the citizenship question fracas to the insertion of political appointees at the highest level. My goodness, the optics of that are terrible for an organization that’s supposed to have scientific independence and integrity.

I can go on — the flip-flop in the schedules. One minute you’re going for another month. The next minute, your field work stops immediately. Regardless of how superb they are as scientists and field operations folks, people can’t actually function properly if you’re getting such mixed orders about what to do one day to the next.

Another threat to the census was the insertion of political appointees. No one knew this was going to happen. Then they started to ask nosy questions about census operations after the Census Bureau officially said they would not be involved in operations.

Is there any remedy to an unfair census?

My suspicion is that the congressional allocation is going to be a done deal if the actual whole person counts are used, regardless of the flaws that everyone knows are going to be baked into the census.

If the administration tries to take out undocumented folks, that’s going to lead to lot of litigation. That goes directly against the Constitution.

Population counts are used as the basis for the next 10 years of population projections, and those projections are used for really important studies and to issue federal funding to the tune of $1.5 trillion per year. It’s possible that one could do some independent research — first to identify the areas of problems geographically in terms of undercount or overcount (college students, duplicates and such), and second, to come up with an estimate for what that was — so that a correction could be made to the population projections. Rather than starting with a severely flawed census and then baking those errors in for the next 10 years of population projections, one could start with those and make adjustments for known deficiencies.

That would take legislation. It’s entirely possible, but a lot of that depends on the composition of the Congress.

Do you view the Trump administration’s efforts to get a citizenship count as a work-around to the failure to put a citizenship question on the census?

Because the Supreme Court ruling did not allow the citizenship question on the decennial census, the White House basically took up the Census Bureau on its earlier offer to estimate the number of citizens using administrative records. The CVAP program, Citizens of Voting Age, has been around, and it’s always used the American Community Survey five-year collapsed counts, so that you can get down to the block group level. It gets down to 8 to 10 city blocks. What the White House wants is counts down to the block level.

There are potentially two reasons. Both to me are nefarious. The first is that it allows you to do hyper-gerrymandering. Before, block groups were so big that you could have a block group where part of it covered a minority neighborhood and part of it covered a non-minority neighborhood. There was no way of splitting that up. By getting down to the block level, it will allow state legislatures to decide for themselves if they want to use citizenship counts rather than total population counts to create their state legislative boundaries. If they decide to use citizenship, they’ll be able to do that down to the block level. Two blocks next to each other — one of them goes to one district, the other goes to the other district.

What’s the second potentially nefarious purpose behind the White House’s plan?

The first is hyper-gerrymandering. The second is enforcement. The Census Bureau collects information for statistical purposes, not for enforcement purposes. The notion of these counts ever being used for enforcement — basically targeting high population, low citizenship-count blocks for ICE raids or whatever — that would be really, really problematic. That would almost surely result in lawsuits. The Census Bureau doesn’t want that. Most of the public doesn’t want that.

What can officials at the Census Bureau do to improve the census now?

I’m a great supporter of the Census Bureau. I love that place. I think they’re the best in the world. I have to call it the way I see it. I honestly think folks need to look in the mirror and say: What does it mean to support the Constitution? Do we really submit data that we believe to be inadequate? They’re going to have to come to terms with that in their processing of the data. Is this good enough to send?

Are you saying that officials at the Census Bureau should not submit the census count?

No, no, absolutely not. They have to submit something. That doesn’t mean they’re going to feel good about it.

Do you believe that the census is being rigged to produce an outcome that is favorable to Republicans?

No. I wouldn’t say it’s being rigged. It’s being sabotaged.