President Trump tried to deter Florida Hispanics from voting
Tirso Luis Páez helped write Canción de Trump, the catchy salsa tune that is now President Donald Trump’s official Spanish-language theme song for 2020.
Páez and Los 3 de la Haba
As it tried to deter unpersuadable voters from the polls, Donald Trump’s 2016 campaign had a crude understanding of South Florida’s Hispanic electorate. But in 2020, his campaign has a better understanding.
na — a South Florida band made up of him, his mother and his father — even sang it this month with a reticent but smiling Eric Trump during a boat parade off the coast of North Miami.
But four years earlier the Trump campaign had so little clue about Páez’s political
beliefs — and about South Florida’s crucial Hispanic voters in general — that its internal data classified him as a wavering Hillary Clinton supporter who should be bombarded with negative ads about the Democratic candidate. The idea was not to lure his vote but to “deter” him from voting at all since the campaign believed he would never cast a ballot for Trump.
“They got it wrong,” Páez said.
The Cuban-born singer was a supporter who proudly voted for Trump.
The Trump campaign data, exclusively obtained by the U.K.’s Channel 4 News and shared with the Miami Herald, shows just how little the
future president’s team seemed to understand South Florida’s Hispanic community, which includes a crucial GOP voting bloc of Cuban Americans.
A Herald analysis found that the campaign, which gathered terabytes of data on potential voters nationwide, listed an incorrect birthplace for half of all Hispanic voters in MiamiDade County who indicated they were born in Latin America when they registered to vote.
For instance, not a single Venezuelan American was identified correctly, according to voter registration records compiled by University of Florida political scientist Dan Smith. Most of the Venezuelans were listed as Cuban Americans in the Trump data. And 25 percent were miscategorized as being of Mexican descent.
Overall, the Trump campaign identified more people as Mexican American in Miami-Dade than the actual number of Mexican Americans registered to vote in the entire state.
After Trump’s shocking win, his campaign bragged about the precision of its data effort, which was designed by the controversial and now-bankrupt British firm Cambridge Analytica.
The campaign’s machine-learning algorithm divided voters into categories based on their likeliness to support either Trump or Clinton — and then went one step further than a campaign is ever known to have gone: Trump’s team identified uncommitted Clinton supporters who could be “deterred” from showing up to the polls with negative ads and disinformation.
“We didn’t go into details saying we need to be super-villains with this thing,” Leo Reyes, who worked on Trump’s digital campaign in 2016, told Channel 4 News. “What we were being told is that we have to do our best to win. And if we have to deter your vote, then that’s what we’re going to do.”
There’s disagreement about how successful microtargeting is in political campaigns, but experts agree that any such effort requires good data to work.
Ryan Enos, a professor of government at Harvard University, said the Trump campaign’s faulty birthplace data could have been a “real liability” in efforts to target Hispanics.
“If you want to build a predictive model of how people are going to vote, you need to get the country of origin right for people of Hispanic descent, especially in Florida,”
Enos said. “The things that predict a lot of human behavior, including voting, are very basic: our social identity, our gender, our race, our party.”
While a Herald analysis found some evidence that Trump’s efforts to deter voters worked in Black communities, that wasn’t true for Hispanic voters. Hispanic voters targeted for deterrence went to the polls at similar rates as those who weren’t.
Overall, Trump did worse with South Florida Hispanic voters than previous Republican presidential candidates.
It could be a question of language. In 2016, the Trump campaign didn’t run a single television ad in Spanish, even as proTrump super PACs ran ads for Black voters filled with disinformation about Clinton. The campaign didn’t even translate its website into Spanish, although it did run Spanish-language ads on South Florida radio stations and deployed Spanish-speaking door knockers.
This time around, polls show the president is doing significantly better with Hispanic voters and is spending millions on messaging tailored specifically for Miami’s major Spanish-speaking immigrant groups.
“President Trump’s Hispanic support is growing because, unlike Joe Biden, he has stood up to socialism and achieved the greatest economy ever for Hispanics before the global pandemic,” Trump campaign senior advisor Mercedes Schlapp, a South Florida native, said in an email.
Schlapp did not answer questions about Trump’s data operation. The campaign has said the idea that it deterred voters is “nonsense,” although the internal data analyzed by the Herald offers an X-ray view into Trump’s unprecedented effort to keep his opponent’s supporters away from the polls.
Now the campaign is using Cancion de Trump a Spanish-language ad playing not just in Florida but in other swing states with large Hispanic populations.
“In 2016, there were many people who were afraid to say [they supported Trump],” said Tirso Páez. “This song has helped many people to come out of the closet politically.”
Ana Páez, Tirso’s mother, is a registered Democrat who sat out 2016.
But earlier this year, at a private party hosted by a Trump supporter, Páez spoofed one of Los 3 de Habana’s songs with proTrump lyrics (“Yo voy a votar por Donald Trump”) — and it went viral. Trump even shared a version of the salsa on his YouTube page.
Now Páez is voting for the president.
So is her husband, German Pinelli, who wasn’t registered to vote in 2016, and so is Tirso Páez’s wife, Magela Crespo, who voted for Clinton.
Despite being a Democrat, Ana Páez was categorized in 2016 as a swing voter who could be convinced, or “persuaded,” in the campaign’s terminology, to vote for Trump. Crespo and Tirso Páez were selected for deterrence.
Voters under 35 like
Páez were more likely to be labeled deterrence than older ones, an analysis of the data shows.
It’s not clear how the Trump campaign’s algorithm came up with its determinations. Trump’s data scientists fed in millions of data points — including voter data that was gathered by the Republican National Committee, personal information purchased from commercial providers, and political donor lists — to categorize voters.
The model might have been sophisticated, but the outcome was crude: Black voters and voters living in primarily Black communities were far more likely to be categorized for deterrence, while Hispanics were generally underrepresented in the deterrence category, especially Cu
in
bans.
But Cubans who lived in majority-Black communities were almost twice as likely to be marked for deterrence as those who lived in majority-Cuban areas.
Dominicans and Colombians were almost as likely as Black voters to be labeled for deterrence, although the Trump campaign’s data show it didn’t know the real birthplace of those voters and largely believed them to be Cuban, Mexican or Puerto Ricans.
Miami-Dade had one of the highest levels of Hispanic deterrence in the state. Broward County, which has a large Venezuelan population, and Orange County, where many Puerto Ricans live, also had high levels of deterrence for Hispanic voters.
Trump’s deterrence operation took place not just in Florida but in other swing states too. And the president’s re-election campaign is relying on some of the same staffers who led the effort in 2016.
MISREPRESENTED AND MISUNDERSTOOD
There are more than 826,000 Hispanic voters in the Trump campaign’s internal data for MiamiDade.
It’s clear the campaign didn’t know much about them.
The Herald analyzed nearly 156,000 voters in the Trump data who had clearly marked their place of origin in state voter registration records that were compiled by Smith, the UF professor.
Smith’s cleaner data allowed the Herald to determine the voters’ real birthplaces.
The comparison showed that the campaign’s “country of origin” data, which it purchased from commercial providers, was entirely inaccurate for nearly 77,000 voters.
Of those voters, the Trump data mislabeled 100% of people from Argentina, Bolivia, Chile, Costa Rica, Ecuador, El Salvador, Panama, Paraguay, Peru, Spain, Uruguay, and Venezuela.
More than 90% of voters from Puerto Rico, Honduras, Colombia, Guatemala, Nicaragua and
the Dominican Republic were mislabeled. And 70% of Mexicans were mislabeled, compared to just 14% of Cubans.
The campaign included Puerto Rico, a U.S. territory, as a “country” of origin.
Alex Melendez, 51, was one mistake among many. Trump’s data listed him as Cuban.
Wrong.
Melendez, a retired pilot with no party affiliation, was born in Puerto Rico.
“For Republicans, maybe Cubans and Puerto Ricans are the same,” he said.
Melendez remembers his West Kendall neighborhood being deluged in 2016 with Spanish-speaking campaign volunteers wearing Make America Great Again hats and knocking on doors. The campaign listed him as a target for “persuasion.” Wrong again.
He saw the red hats coming and never answered the door.
While Trump won Florida, his tactics didn’t work with Hispanic voters in Miami-Dade.
Unlike Black voters, Hispanic voters selected for deterrence in 2016 voted at higher rates than they did four years earlier, a Herald analysis found. And Hispanic voters that the campaign labeled as extremely likely to support Trump voted in lower numbers in 2016 than they did in 2012.
“The Cuban vote was at an all-time low for Republicans,” said Eduardo Gamarra, a professor of politics and international relations at Florida International University.
Clinton did surprisingly well in Cuban-American precincts, research and exit polling data show.
“In 2016, [microtargeting] was a relatively new science and the data was untested,” said Karen Giorno, who worked on Trump’s campaign both in Florida and at the national level. “There was an understanding of how important it was to understand the voter correctly and in order to do that the profile of the voter became the priority for the GOP in 2020: Understanding the voter: what they believe in, what they want, where they come from.”
(Giorno denied that Trump engaged in a deterrence campaign in 2016, saying she had never heard that term used and that those tactics would amount to “voter suppression.”)
For his part, Tirso Páez says that his wife is now a stronger Trump supporter than he is.
“I think that [social media] has influenced her over the past three years,” Páez told the Herald. “The change that has happened on social media with YouTubers like Alexander Otaola has pushed her to realize many things and change her way of thinking.”
Otaola is a 41-year-old
Cuban social media “influencer” who talks about politics and culture on his You. (The Trump campaign is one of the sponsors listed on the channel.) Otaola says that although he voted for Clinton in 2016, this year he will vote for Trump.
Ana Páez said that although Biden seems like a good person, she would never vote for him because she doesn’t like his running mate, Sen. Kamala Harris, or the “others that support him” like
Rep. Alexandria OcasioCortez and Sen. Bernie Sanders.
RECALIBRATING THE CAMPAIGN
Make America Number One, a pro-Trump super PAC that worked closely with Cambridge Analytica in 2016, realized the GOP needed to do better with Hispanics in 2020.
In internal documents, the group said that it recommended investing “in Hispanic data collection through survey research, third-party data scraping, and digital targeting. The Hispanic portion of the electorate is only growing, and for Trump, or any Republican, to be successful in the future, understanding the messaging and targeting of Hispanic voters is paramount.”
Trump’s team was listening, knowing they would need to narrow their opponent’s margin of victory in heavily Democratic Miami-Dade to win Florida again.
The president is now leading Biden 61% to 35% with Cuban-American voters in Miami-Dade, according to an early October poll by Bendixen & Amandi International and the Miami Herald. In
2016, his support from that community hovered in the mid-to-low 50s. Trump is doing especially well with recent Cuban immigrants, who skew heavily Republican, while Biden has more support among American-born Cubans and still holds an overall lead with Hispanic voters in South Florida.
The improved poll numbers reflect the GOP’s new commitment to reaching South Florida Hispanics.
In the Miami market, Trump has spent $3 million on Spanish-language television ads so far this cycle, trailing Biden’s $5 million, but still a significant improvement over 2016, according to Advertising Analytics. ProTrump political action committees have spent an additional $1.7 million in Spanish-language ads in South Florida. (Groups supporting Biden have pumped in $6.7 million.)
The campaign has also developed messages targeting crucial immigrant groups.
For Cubans, one ad plays footage of Biden followed by a speech from Fidel Castro. For Venezuelans, Biden, then the vice president, is shown meeting with Venezuelan Presi
dent Nicolas Maduro; a socialist; another shows Maduro ally Diosdado Cabello, who has been indicted in the United States on narco-terrorism charges, suggesting he supports Biden. For Colombians, one spot features Colombia’s left-wing politician and economist Gustavo Petro, once a member of a guerrilla group, saying he would vote for Biden if he could cast a ballot. That same ad also accuses Biden of “betraying Nicaraguans.”
All are in Spanish.
Some of Trump’s newfound success is the advantage of incumbency — the president has a list of policy decisions he touts, such as reimposing sanctions on Cuba, issuing an executive order targeting the finances of Nicaraguan President Daniel Ortega, and seeking to unseat Venezuela’s Maduro. He has more money, too, raising roughly $600 million in 2020, nearly double what he did four years ago.
The campaign’s state director and regional director for South Florida are both from the Miami area.
“They’ve hired a lot of people down here who know and understand the community,” Nelson Diaz, chairman of the MiamiDade County GOP, said of the Trump campaign. “They listen to advice from people like me and locally who know the community.”
Carlos Cabral, an attorney at a law firm in Miami Lakes, is one of the president’s new supporters.
A former Obama voter targeted for deterrence in 2016, he said he had no confidence in Clinton and didn’t cast a ballot.
Now Cabral, who is half-Cuban and half-Portuguese, is leaning toward Trump.
The president’s messaging that Biden will raise taxes on small businesses has him scared. So has the constant barrage of ads linking Biden to authoritarian socialist regimes in Latin America. And he’s a hunter who feels Trump’s views on gun rights match his own.
“I’ve got to choose the lesser of two evils,” Cabral said.
Miami Herald staff writers David Smiley, Christina Saint Louis, Isaiah Smalls and McClatchy DC staff writer Shirsho Dasgupta contributed to this report.
This story was researched and written using a 2016 Trump campaign/Republican National Committee voter dataset obtained by Channel 4 News in Great Britain and shared with the Miami Herald. The two news organizations consulted with each other and shared information but produced their own reports.