Every few cycles, the American electorate submits itself to a natural experiment, and the results arrive with the tidy cruelty of a peer-reviewed finding. In 2024, Latino voters, a constituency long treated by Democratic strategists as a fixed asset rather than a persuadable one, broke for Donald Trump in historic numbers. The Republican Party, never an institution to let a confound go unexploited, gerrymandered Florida and Texas around the result and declared a realignment. Two years later, the new NBC News/Telemundo poll has returned the replication, and the replication does not go well for the hypothesis.
The numbers are, in the technical sense, bracing. Trump’s approval among Latino voters sits at 27 percent, with 69 percent disapproving. Democrats now lead by 24 points on the generic congressional ballot, 59 to 35, double their 2024 margin. Two-thirds of Latino voters say the administration’s deportation policies have gone too far. Seventy-eight percent disapprove of his handling of the cost of living. Even Micah Roberts, the Republican pollster who conducted the survey, called the reversal “stunning” and pronounced 2024 “the outlier,” which is roughly equivalent to a defense attorney standing up in court and advising the jury to convict.
What we are observing, of course, is not a sudden ideological conversion but the entirely predictable consequence of a campaign built on vibes colliding with a governing agenda built on cruelty and grocery prices. The literature on this is robust. Voters, it turns out, evaluate incumbents on material conditions, and 65 percent of Latino respondents say Trump’s policies have hurt the economy. Among Latino independents, 72 percent say so. One does not need a doctorate to detect the directional signal.
The qualitative data is where it becomes genuinely difficult to read without wincing. Luisa Santiago, a 59-year-old Florida retiree, voted for Trump in 2024 after a lifetime of voting Democratic and, by her own account, persuaded her friends to do the same. “I’m more embarrassed than my friends because I was the one who convinced them to vote for him,” she told NBC. She described the deportation regime as “brutal” and “cold.” A younger Florida voter, identified only as Maria, noted that Trump had promised to deport “the worst of the worst,” and yet, “I’ve had many friends, family members deported. They were supposed to be here. They had paperwork, and it didn’t matter.”
There it is, in one sentence. They had paperwork, and it didn’t matter. The entire Trump immigration architecture, the raids, the workplace sweeps, the 31 percent of Latino voters who now carry identification out of fear of ICE, rests on the fiction that it targets only the deserving. The poll’s respondents have performed the observational study the administration never bothered to run: 83 percent support deporting undocumented immigrants who committed serious crimes, and only 37 percent support deporting everyone. The gap between those two numbers is where the Republican coalition used to live, before it decided the gap was for losers.
Now, in the interest of scholarly honesty, one must complicate the complication. The Democratic Party’s own image has not improved in two years; the gains are anti-Trump, not pro-Democrat, as pollster Aileen Cardona-Arroyo conceded. Republicans are still pulling more Latino support than in any NBC poll before 2024, dating to 2012. The realignment was overstated, but the floor beneath it has genuinely risen. Democrats who mistake this poll for affection rather than recoil will write a very different paper in November.
Still. The party that wagered its House majority on the durability of one man’s charm with one demographic has now received its mid-course evaluation. The heavily Latino districts drawn deep red in Tallahassee and Austin are suddenly competitive. The swingiest of swing groups, as Roberts called them, has swung. And the professoriate’s conclusion, offered with the weary patience of someone who has graded this exam before: when your governing model requires voters to be poorer, more frightened, and more deported than they expected to be, your margin of error was never as large as you believed.