Americans are sharply divided on the NFL’s decision to feature Puerto Rican-born artist Bad Bunny at the 2026 Super Bowl halftime show, according to a new survey by Quinnipiac University.

Overall, nearly half of respondents said they approve of the decision, with 29 percent disapproving and another 24 percent offering no opinion. While nearly 3 in 4 Democrats said they approve of the NFL’s decision, more than 3 in 5 Republicans oppose the move.

Opinions were also largely divided by race and age: Bad Bunny enjoyed significantly more support among Black and Hispanic adults than among white respondents, and his biggest contingent of support came from those aged 18-34.

The NFL, which has repeatedly drawn the ire of the MAGA movement in recent years, once again faced attacks from President Donald Trump and his supporters after announcing last month that Bad Bunny would take the stage at the most-watched live event in the country.

The artist, who was one of the top three most-streamed artists in the world on Spotify last year, has been vocal in his criticism of the Trump administration’s immigration crackdown, declining to schedule dates in the mainland U.S. during his upcoming world tour due to concerns about heightened ICE presence. The music video for his song “NUEVAYoL” also features a parody of Trump’s voice purporting to “apologize to the immigrants of America.”

MAGA firebrands quickly seized on the announcement to call the singer a “Trump hater,” and Turning Point USA announced a counterprogram halftime show, although the conservative organizing group has yet to announce featured performers.

Trump called the decision “absolutely ridiculous” in an interview with Newsmax’s Greg Kelly earlier this month, and Homeland Security Secretary Kristi Noem has warned that ICE agents would be “all over” the Super Bowl.

Still, the NFL has stood by its decision, with Commissioner Roger Goodell reiterating last week that the league would move forward with the show despite blowback from conservatives.

The Quinnipiac poll was conducted by telephone Oct. 16-20, with a random sample of 1,519 adults. The margin of sampling error was plus or minus 3.3 percentage points.