Michelle R. Smith / AP

It isn’t hard to discern a pattern in the way the Trump administration is planning to conduct the 2020 Census, in the same way that it’s not hard to discern the racial animus against Hispanics that undergirds the president’s moves on immigration. Yet even if the motives are apparent, the effects are not so easy to predict.

Most notably, the Commerce Department has announced plans to add a question asking respondents about their citizenship, a move stoutly opposed by immigrant advocates. The Office of Management and Budget also put the kibosh on a plan to combine the Census’s race and ethnicity questions into a single question, which the Census Bureau had concluded would produce a more accurate count, especially of nonwhite people. The Census Bureau is also planning to draw more heavily on existing federal records and digital tools.

“It seems like Latinos are in the crosshairs for Census 2020,” Arturo Vargas, executive director of the National Association of Latino Elected and Appointed Officials, said Monday at the Aspen Ideas Festival, which is hosted by the Aspen Institute and The Atlantic.

The Census is constitutionally required to count everyone in the country every 10 years. That includes both citizens and noncitizens, though Alabama is currently suing to try to have unauthorized immigrants excluded from the tally. The implications of the tally are wide-ranging: The numbers determine not only how seats in the U.S. House are allocated and reallocated among states, but also federal funding levels for programs like Medicaid, the Children’s Health Insurance Program, highway funding, and more.

Needless to say, counting everyone in the country is an extremely difficult task. Unlike surveys, which sample the population and then extrapolate data, the Census must try to count everyone, but to do so it has to grapple with finding every address in the United States, people who fail to return forms, and other challenges. College students are often overcounted, listed both in dorms and at their parents’ houses. African Americans are chronically undercounted—Vargas said the Census only realized how badly off its tally of black men was when more enlisted in World War II than existed, according to Census figures.

The introduction of new digital tools, including the use of federal records to fill in gaps in returned forms, is a positive development. After all, the federal government already has extensive records. The question is whether the Trump administration is likely to execute the shift competently and without political motivation.

The signs so far are not encouraging. The end-to-end test, a critical dress rehearsal for 2020, has been plagued by problems, and can’t follow the exact questions of the real Census because they aren’t set yet.

Or consider the citizenship question. The desire for an accurate count of how many citizens there are makes sense, even if it’s not the central aim of the Census. In fact, the American Community Survey, another Census Bureau initiative, has done such a count for years, without major problems. Besides that, Commerce Secretary Wilbur Ross—who is also in hot water for less-than-honest financial disclosures—has offered contradictory accounts for the genesis of and motivation for the citizenship question.

Vargas said that focus groups with Hispanics indicated that the question would depress willingness to return the Census, which could result in undercounting of the population. Not returning a Census form is punishable by a fine, as is lying on the form. Even before the new question, the Census Bureau concluded it was undercounting Hispanics, though Vargas said NALEO believes the undercount is greater than the 1.5 percent the Census Bureau calculated.

Undercounting can have important effects for the decade that follows the Census.

“If people are not counted, the communities where they live will get that much less political representation and the resources they need,” Vargas said. “The irony is, typically these programs are for vulnerable populations. These are the very same people the Census Bureau does a poor job of counting.”

That doesn’t mean that only blue states will be harmed. After 2010, the only red state to see its congressional delegation shrink was Louisiana, which lost residents after Hurricane Katrina. If the 2020 Census undercounts Hispanics, however, conservative and progressive states alike could suffer.

“It’s going to hurt Texas, and it’s going to hurt California. It’s going to hurt New York, and Arizona,” Vargas said. “The South is the region of the country that has experienced the fastest and largest increase in the immigrant population. Those states stand to lose money over the next decade if not all their immigrants are helped.”

The Census Bureau also pulled back on a plan that would have more extensively canvassed rural areas, meaning that rural residents could be undercounted as well.

The likely winners from the Trump changes, Vargas said, are Midwestern states like Ohio and Wisconsin. Both voted for Trump in 2016 but, unlike immigrant-heavy Southern states, they’re more purple than red.

This makes the Census another example of policies pursued by the Trump administration that could hurt his base. The president’s tax plan helped his wealthy backers, but offered much less to the economically precarious white voters who bolstered his coalition. The trade wars he has begun are likely to hurt rural farmers and some manufacturing employees in Rust Belt areas that backed him in 2016.

Red-state governments have sometimes been willing to take those hits for ideological reasons, especially when the people who stand to benefit most from the services that are being lost are poor or minorities—witness, for example, the many Republican-led states that opted to decline federal funding to expand Medicaid under Obamacare. Nonetheless, the potential damage to red states, in both funding and representation, of Trump’s Census moves is an interesting lesson in the unintended consequences of political choices.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.