Assertions of bald political skullduggery on one side, and protecting voting rights on the other, are obscuring a core consequence of the Trump administration’s decision to include a citizenship question on the next census.

Posing such a query will likely reshuffle the allocation of congressional seats for the coming decade.

Unless the question is blocked by the courts or by Congress itself, neither of which is close to a sure thing, a handful of reliably blue states across the country are severely threatened with a decline in guaranteed clout on Capitol Hill during the 2020s, population studies suggest. And another handful of purple states in the Sun Belt could see their increasingly loud legislative voices quieted a bit.

This is entirely because the federal government asking such a question — especially a government perceived as so anti-immigrant — is certain to scare many thousands, if not millions, of noncitizens away from completing their forms, producing a significant undercount of the American population, disproportionately in neighborhoods filled with the undocumented.

It is not because President Donald Trump or anyone else in the government has proposed a change to assign House seats based on the states’ population of citizens, not human beings as has been the case since the beginning. No one in power has suggested that, and just two years ago, the Supreme Court ruled unanimously that the “one person, one vote” principle means all residents, regardless of eligibility to vote, should be counted in drawing election boundaries.

The Justice Department says the data collected from the question — the last census that included it was in 1950 — will improve its fights against fraudulent ballot casting and voting rights violations. Congressional Democrats and civil rights groups describe the question itself as a scare tactic designed to deny aid to and diminish the political clout of Hispanics.

Watch: The Many Ways to Draw a Gerrymander

Gains and losses

Census findings have long been the coin of the realm when it comes to allocating tens of billions of dollars in federal aid annually to almost every community and demographic group. But the principal purpose of the decennial counting, as dictated by the Constitution, is apportioning voting strength in the House.

It’s been a zero-sum game for more than a century, since Congress itself decided to limit the number of fully voting members to 435. So taking away House seats from some states and giving them to others, based on relative population changes, creates pins-and-needles anticipation at the start of every decade for incumbent lawmakers and their would-be colleagues from coast to coast.

Sometimes, the difference between a state gaining, losing or holding a seat can be the number of folks needed to populate a one-stoplight town. After the 2000 census, famously, Utah missed getting to send a fourth person to the House by just 856 people; the seat went to North Carolina instead.

And so the success rate for the largest single data-collection undertaking in the world — an old-fashioned 18th century head count even in this 21st century time of big data extrapolations — has a consequentially tiny margin for error.

Surely complicating the work will be trying to persuade all respondents to truthfully say if they are United States citizens — an especially provocative question while immigration status and nationalistic anxieties are white-hot ingredients in political debate.

Hanging on

“One can safely presume this will depress participation,” said Kimball W. Brace, president of the political demographics company Election Data Services. “I’m not happy with that. It could have an important impact.”

The firm has become perhaps the nation’s foremost authority on reapportionment, in part because it annually takes population growth and projections, and funnels them through the complex mathematical formula the government uses for allocating House seats to the 50 states. Brace identifies three of them, all reliably Democratic, as in particularly acute danger of reapportionment disappointment if there’s even a modest undercount.

California, the state which has by far the largest population of undocumented immigrants, was hanging onto its 53rd House seat with fewer than 76,000 people to spare at the end of last year, the EDS number-crunching concluded. If only that relatively small number of Californians — akin to the population of Menifee in Riverside County, east of Los Angeles — avoid the census, it would mean an actual reduction in the largest congressional delegation in the country, the first time California would see its representation on the Hill decline in 170 years as a state .

Lethargic population growth in Illinois already means it’s guaranteed to see its roster of 18 House seats decline by at least one. But current projections show it’s also hanging on to its 17th seat by just 104,000 people (a bit more than the population of Waukegan), and Chicago and its environs have significant undocumented populations of both Latinos and Eastern Europeans.

Finally, Rhode Island, with a sizeable Hispanic population north of Providence, is hanging on to its second House seat by fewer than 200 people — so it would almost certainly join the states with “at large” House members if there’s even the slightest undercount.

If the census participation rate plummets even more dramatically, the Election Data Services numbers point to setbacks for three warm and sunny states where populations are booming and the new arrivals — many thousands of them Hispanics — are creating more elections that Democrats have chances to win.

An undercount of more than 66,000 people in Arizona (the size of central Flagstaff) would jeopardize its claim to an additional, 10th House seat.

If the official census population for Texas falls just 60,000 (the number of people in Harlingen) below current projections that could limit the state to adding two House seats, not the anticipated three, to its already-second-largest allocation of 36.

And a Florida undercount of 120,000 (the size of Coral Springs) would limit the state, which now elects the third most House members at 27, to being able to choose just one more House member instead of the projected two.

Uphill battles

While civil rights groups and the attorneys general of several of the largest states said they would focus on lawsuits alleging the Trump administration’s move is both unconstitutional and in excess of its regulatory authority, senior Democratic lawmakers said they would move on several uphill legislative fronts to try and reverse the decision.

They are unlikely to have much luck advancing either of two bills they have already put forward. One, by Sen. Robert Menendez of New Jersey, would prohibit census questions regarding citizenship and immigration status. The other, by Rep. Carolyn B. Maloney of New York, would effectively block Trump’s move by requiring that any substantive census changes be “researched, studied, and tested” at least three years before the count .

Congress could have better luck wielding its power of the purse, denying any appropriations to pay for preparing a census with such a question. The issue therefore rockets to the top of the early list of policy riders Democrats will seek in the fiscal 2019 appropriations process, which is just getting started.

And if they are unsuccessful in advancing such a measure this fall, but are successful at winning control of at least the House, their leverage a year from now — over the spending bill governing the actual conduct of the census — would be increased substantially.

“This fight can potentially be a very long fight, right up to when the census is taken and maybe even afterward,” Vanita Gupta, head of the Leadership Conference on Civil and Human Rights and before that head of the Obama Justice Department’s civil rights office, vowed in a conference call with reporters Tuesday.