WASHINGTON — A Census Bureau memorandum on creating a state-by-state estimate of people illegally in the country is raising new fears of a politicized census — this time involving the population totals that will be used to reapportion the House of Representatives next year.
The memo, issued Monday, orders an internal task force to explore statistical methods of compiling an accurate estimate of noncitizens. It says the aim is to carry out President Trump’s July mandate to exclude undocumented residents from population totals used to determine how many House seats each state is entitled to. The directive, which is being challenged in court, is widely seen as an effort to shift some House seats to Republicans during reapportionment next spring.
Some experts inside and outside the bureau see the memorandum as the precursor to an effort to manipulate population figures to give Republicans an even greater edge in reapportionment. In essence, they say, they fear that the methods the bureau is being asked to devise will create a road map for the Trump administration to tailor the numbers for maximum political benefit.
“If all that information were to go to the Department of Commerce or the White House, that would be highly irregular,” said John Thompson, a 31-year Census Bureau veteran who ran the agency from 2013 to 2017. “In previous censuses, even the managers running the census did not see the apportionment figures until the whole process had been finished.”
Mr. Thompson was quick to note that the memorandum doesn’t explicitly suggest any impropriety. “It’s pretty carefully written, and it doesn’t say what’s going to happen with the information,” he said. “And it doesn’t say where the information is going.”
Nevertheless, some career Census Bureau employees say it’s hard to see an innocent reason for the request. A cover letter seeking “thoughts, questions and concerns” was signed by the bureau’s respected deputy director, Ronald S. Jarmin. But the experts are convinced that the memorandum was written elsewhere, by political appointees recently added to the bureau or by the Commerce Department, which oversees the agency.
“All of this is unprecedented,” one bureau expert familiar with the memorandum said in an interview this week. “The worry is that they’ll be used for apportionment in a way that will destroy the credibility of the census and the Census Bureau.” That person, like some others interviewed for this article, refused to be quoted by name for fear of retaliation.
In a written statement, the Census Bureau spokesman, Michael C. Cook, said the memorandum was unremarkable. “As always, when a working group is tasked, questions are posed to guide their initial review,” the statement said. “We do not comment on the specifics of internal deliberative documents.”
The statement did not address questions about whether political appointees wrote the memo or whether the Commerce Department would be involved in work on reapportionment.
The storm over the memorandum underscores how deeply the Census Bureau, a historically nonpartisan agency, has been racked by continuing controversies about whether the Trump administration seeks to use, and skew, its figures for political ends.
A group of 24 Democratic senators signed a letter on Thursday urging the bureau to reverse a decision, announced on Monday, to cut the duration of efforts to count the hardest-to-reach households down to six weeks from 10. About 37 percent of households remain uncounted.
“This appears to be yet another effort to sabotage a successful census, which include the administration’s earlier attempts to add an unnecessary and divisive citizenship question and the recent issuance of a memorandum seeking to exclude undocumented immigrants for apportionment purposes,” the letter said.
The administration took an extraordinary step in June by creating a second deputy director’s post at the bureau and filling it with a political appointee, Nathaniel T. Cogley, along with an aide who had been a Republican political consultant.
Mr. Cogley and the aide, Adam Korzeniewski, were briefly on the staff of Commerce Secretary Wilbur L. Ross Jr. before moving to the census jobs. Both have become known within the bureau for skeptical questioning of statistical methods used to improve census accuracy, particularly with hard-to-count groups like minorities and undocumented immigrants.
Census employees say the two men have became known for something else as well: In a briefing for Mr. Ross by career Census Bureau staff members, they mused openly about the political impact of statistical methods used to improve census results on some solidly Democratic states.
The Constitution stipulates that the census counts all residents, not just citizens. By seeking to exclude undocumented residents from the reapportionment count, some experts say, Mr. Trump opens the door to tinkering with the census totals. In effect, a state’s population for reapportionment would depend for the first time on how many noncitizens were removed from the tally.
That makes the methods used to compile an estimate of noncitizens crucial. A method that overestimates undocumented immigrants, for example, would hit states with large immigrant populations especially hard. That would be compounded if, as many predict, this year’s rushed census produces big undercounts of states’ total populations.
The bureau’s task force is studying how to compile an estimate of noncitizens using administrative records and other data. The memorandum asks the group to find ways to measure an array of characteristics of noncitizens that might be used to verify records, including whether they face deportation, have overstayed visas or face immigration hearings.
It also asks the bureau to study whether statistical methods could help determine a person’s status when records are of no help. One principal method, imputation, uses an algorithm to make an educated guess about the occupants of a household based on its neighbors.
The bureau imputes the characteristics of a small number of households in every census. Depending on the quality of records — and the method of imputation — a sizable share of the noncitizen estimate could be educated guesses.
Calculating reapportionment is a sensitive task for the bureau. It is generally given to a team of experts, insulated from senior managers and political appointees, who calculate the distribution of House seats on several computers run by different staff members. Only if all the results match are the figures sent to the White House.
The bureau does not have to calculate apportionment; it is required only to produce an accurate population count. But the agency has long done the apportionment calculation as a courtesy. “There’s a tradition of transparency when the numbers are handed off from the bureau,” one expert said. “It’s a norm, not a law.”
In a norm-busting administration, that is another thing for census experts to fret about.
“Having watched censuses for 50 years, there is no doubt in my mind that this is the most politicized census in my lifetime,” Bill O’Hare, a demographer and author of two books examining the accuracy of censuses, said on Wednesday. “There is a whole series of things that suggest this administration doesn’t see an accurate census count as very important.”