Trump census order can’t be done accurately, experts say

The man who designed the method Homeland Security uses to estimate the number of illegal immigrants says there’s just too much margin for error to use it to carry out President Trump’s new executive order to strip illegal immigrants out of the census count used to apportion congressional seats.

Most of the public debate over Mr. Trump’s order centers on whether it’s legal.

But Robert Warren, a demographer who used to work for the government and is now at the Center for Migration Studies in New York, told The Washington Times whatever the legality — which he very much doubts — it’s also unworkable.

Given the tools the government has right now, he can’t see how they could come up with a number accurate enough to rely on for something as important as doling out seats in the U.S. House for the next decade.

“It’s not possible to do it accurately,” Mr. Warren says. “The problem is that the estimates are not precise enough.”

Mr. Trump’s order does not affect who’s counted by the census. Illegal immigrants will be tallied like everyone else.

But when the administration produces the count to be used to assign the 435 seats in the House of Representatives to the various states, Mr. Trump wants to delete the number of illegal immigrants as much as practicable, arguing they distort political power by unfairly boosting the populations of some states.

Census Bureau Director Steven Dillingham told Congress last week his agency is just at the beginning of trying to figure out how to carry out the president’s order.

“We are going to analyze the data that we have and look at the methodologies that might be employed for that purpose,” he said.

The key to Mr. Trump’s order is getting good state counts of the population without illegal immigrants.

Mr. Warren says the system used by the government begins by trying to estimate the foreign-born population based on the American Community Survey.

That’s what used to be the census’s long form, a lengthy set of questions sent to a much smaller subset of people than the full 2020 census will reach. The ACS asks about place of birth — and, ironically, also asks about current citizenship, a question that was deeply controversial when Mr. Trump suggested including it in the full census.

Once the foreign born estimate is known, analysts subtract the known number of people admitted legally by year, and subtract that from the foreign born. They also factor in mortality and emigration, and what’s left over is an estimate of the unauthorized population.

That can also be done by state.

But one problem, Mr. Warren said, is the emigration calculation is based on calculations from 40 years ago.

More troubling for apportionment purposes, there’s no way to account for internal migration. So someone who applied for a green card from California in 1990 is assumed to still be living there, even though he may have moved many times since.

So even if the government could give a decent estimate of the unauthorized population nationally, accurately placing them state-by-state is not possible given the method the feds use.

“I do not think it can be done. On any data system that I’ve heard proposed, it cannot be done,” Mr. Warren says.

He says the methodology he now uses at CMS is more accurate. He starts with the universe of people who say they entered after 1982 — the effective date of the last major amnesty — then subtracts those who have a high probability of being legal, based on things like their use of government programs only open to legal residents.

He makes other adjustments, such as for undercounting in the census itself, and ends up with a universe of people in the country illegally — which he can then identify by the state they listed in the ACS.

The Times asked whether, if the government came calling, he would share his data. He said he hasn’t been asked, and seemed skeptical CMS would be willing to participate in Mr. Trump’s project anyway.

Mr. Trump’s executive order was vague on how the government should go about trying to delete illegal immigrants from the count.

The most obvious option would be to use statistical methods like the ones Mr. Warren created.

But the administration could also try to knock out individual people it figures are illegally present, by matching names, addresses and ages in the census data with information Homeland Security has in its own files.

That would involve taking lists of people known to be here illegally, such as people in deportation proceedings or who have pending asylum claims. The government probably knows specific identities of a few million illegal immigrants or other unauthorized residents based on their current efforts to adjust status or claim protections under humanitarian programs, said Steven A. Camarota, a demographer at the Center for Immigration Studies.

But that’s a fraction of the total population, which experts say is at least 10 million.

It also leaves tricky questions about whether people with DACA or Temporary Protected Status would be counted as legal or illegal residents.

Amy O’Hara, the former chief of administrative records at the Census Bureau and now a professor at Georgetown University, told Science magazine that the systems just don’t exist to sort out all immigrants by legal status.

“I don’t know what set of data sources the bureau could identify for that purpose,” she said. “And for the ones they have, it’s not clear how they would operationalize them.”

Whatever method Mr. Trump’s team does try, it’s going to be snared in the courts for years, with the questions raised by Mr. Warren likely to poke holes in whatever answers the government comes up with.

“As a practical matter, it’s very difficult to actually do,” Mr. Camarota said.

But he said it is an important conversation for the country to have, because it underscores some of the consequences of illegal immigration.

“If they’re included, American citizens lose political representation in some areas of the country to give to other areas of the country where a large fraction of the people are illegal immigrants,” he said.

In estimates he published last year, Mr. Camarota said the difference between counting or excluding illegal immigrants shifts three seats. With illegal immigrants included, California, Texas and New York each have an additional seat. Without them, those seats would go to Ohio, Alabama and Minnesota.

Republicans point out that some of the official census count is already a guess.

If a household doesn’t reply, statisticians “impute” data for that home by extrapolating from nearby households that did provide answers.

“There are times when we have reason to believe, evidence, that someone is living in that household but we are unable to communicate with them, that in fact we do have an imputation accepted method that has been accepted by the courts,” Mr. Dillingham told Congress.

Rep. Chip Roy, Texas Republican, was more blunt: “It means that we make stuff up,” he said.

Mr. Roy also pointed out that the American Community Survey, which is an estimate and not an actual count, is already used for enforcing the Voting Rights Act. He said it’s not a big leap from that to using ACS calculations to come up with reapportionment data.

Multiple lawsuits have been filed to block Mr. Trump’s order. All of them argue he’s violating the Constitution’s 14th Amendment, approved in the wake of the Civil War.

That amendment reads: “Representatives shall be apportioned among the several States according to their respective numbers, counting the whole number of persons in each State, excluding Indians not taxed.”

In a congressional hearing last week, four past directors of the Census Bureau were asked if that covers illegal immigrants. All four said it does.

But Attorney General William P. Barr, at a different hearing on Capitol Hill, said the “whole number of persons” section is usually read to mean inhabitants. Congress, he said, has the right to define who qualifies as an inhabitant, and it has delegated that to the Commerce Department within the executive branch — meaning the president can decide.

“So as the law stands now we think the Commerce secretary as the delegate of congressional power can define that term,” he said. “That’s a reasonable argument to make.”

Immigration — legal and illegal — plays a massive role in determining how much a single citizen’s vote counts.

In California’s 21st Congressional District, for example, nearly 30% of the adult population is not a citizen. By contrast, in the state’s 1st District just 3.4% are not citizens, according to Mr. Camarota’s data.

In 2018 a Democrat won a tight election in the 21st District with just 57,239 votes out of about 114,000 cast. California’s 1st District saw a Republican win more comfortably, with 160,046 votes out of more than 291,000 cast. That means each voter in the 1st District counted less than half what a voter in the 21st District counted when it came to selecting a representative.

In Texas, meanwhile, more than 300,000 votes were cast in the 25th Congressional District in 2018, and a Republican won with about 163,000. Up the state in the 33rd District, fewer than 120,000 votes were cast and a Democrat won with about 91,000.

Mr. Camarota says on average about half of non-citizens are in the country illegal, with variability among the states — the rate is a little lower in California, and a little higher in Texas.