In the lead-up to the election, there was a lot of talk about how Latinos would turn out in record numbers to stop Donald Trump. But not only did Trump end up winning, exit polls also indicated that he won 29 percent of the Latino vote — better than Mitt Romney had done four years earlier. That has sparked a heated debate among pollsters and advocacy groups over whether the exit polls can be trusted.
The most prominent skeptics of the exit polls’ findings have been Gary Segura and Matt Barreto, the co-founders of Latino Decisions, a polling firm that focuses on the U.S. Latino population. Segura and Barreto, who both worked for Hillary Clinton’s campaign, have argued vigorously over the past week that the exit polls were wrong and that Clinton performed better among Latinos than President Obama did in 2012. Based on their own polling, they think Clinton outperformed Obama and got nearly 80 percent of the Latino vote, not the 65 percent indicated by exit polls.
|MARGIN OF VICTORY AMONG LATINOS|
|Edison Research exit polls||+36||+44||-8|
My analysis of the available evidence, however, suggests that Segura and Barreto are taking their argument too far. While Clinton may have done better than the exit polls suggest, a detailed look at pre-election polls and actual election results from counties with large Latino populations suggests that Trump probably did no worse than Romney among Latinos, and probably did better. Different sources tell somewhat different stories, but the weight of the evidence does not back up the conclusions of Latino Decisions, particularly the actual election results from many of the country’s most heavily Latino areas, where Clinton underperformed Obama. Some of our best evidence that Latino Decisions’ analysis is flawed comes from the very counties that the organization chose to highlight in a post-election presentation arguing that Clinton performed well among Latinos.
Barreto, in an interview, said he stands by Latino Decisions’ analysis, which he said is based on a methodology that more accurately captures the voting behavior of the Latino population. “We have our survey that we have 125 percent confidence in,” Barreto said.
No one doubts that Clinton won handily among Latinos. But the scale of that victory matters. Latino leaders — most of whom supported Clinton — want to argue that their constituents punished Trump for his policies on immigration and other issues, more so than they did Mitt Romney in 2012. Trump wants to be able to claim as broad a mandate as possible. And in the longer term, Latinos are one of the fastest-growing segments of the electorate, and Democrats are counting on their support to retake the presidency in 2020. But if Trump, who said many Mexican immigrants are “rapists” and questioned the impartiality of a Mexican-American judge, basically equalled Romney’s performance among Latinos — Trump won 29 percent according to exit polls, and Romney won 27 percent — that suggests that Republicans might be able to make gains in the future, or at least prevent Democrats from winning over a larger share of that voting population.
The main basis for Latino Decisions’ argument that exit polls are wrong is that their own pre-election polling — conducted with live phone interviews among only Latinos — had Clinton winning Latinos by a margin of 61 percentage points (79 percent to 18 percent), whereas exit polls conducted by Edison Research showed Clinton winning by just 36 points (65 to 29 percent). The disagreement is the latest chapter in a long-running dispute between Edison and Latino-focused pollsters such as Latino Decisions, which have for more than a decade argued that Latinos vote for Democrats by wider margins than other sources claim.
Other polls that surveyed only Latino voters, such as the New Latino Voice poll, generally reported a Clinton margin that was larger than that reported by exit polls, but smaller than the one in the Latino Decisions survey. Polls that surveyed the entire electorate generally agreed with the exit polls. An average of the seven live-interview national surveys conducted in the final weeks of the campaign1 indicates that Clinton led Trump by 33 percentage points among Latinos. And a post-election online poll of the entire electorate from SurveyMonkey had Clinton ahead among Latinos by 39 percentage points. Both are similar to the 36-point lead found by Edison in its exit poll.
The various polls differ methodologically, sometimes in important ways. Latino Decisions argues that it captures Latinos’ views more accurately because the firm’s interviewers are bilingual2 and because it targets areas where a high percentage of the population is Latino, unlike exit polls, which are conducted in random sample precincts that often have few Latino residents. Edison Research argues that exit polls are superior because they don’t need to make assumptions about whether respondents actually voted — an issue that has emerged as a key question this year after pre-election polls overstated support for Clinton. Moreover, Latino Decisions finds people to interview in part by identifying people with common Spanish surnames; that means they may miss Latino voters who don’t have Latino-sounding names, who tend to be more Republican-leaning. Exit polls, because they interview people at random at polling places, don’t have that problem. In the end, there is probably no way to know for sure what share of Latinos voted for Clinton versus Trump, just as we can’t be certain how any group voted in previous presidential matchups. As Nate Cohn of The New York Times has noted, different methodologies for judging how various demographics vote can produce different results.
We have a better chance of answering a different question: How did Clinton’s margin of victory among Latinos compare to Obama’s in 2012? Here, too, Latino Decisions disagrees with other pollsters. The firm’s data shows that Clinton did better among Latinos than Obama did four years earlier, when the firm showed him beating Romney by 52 points. Other firms’ polls of the entire electorate, however, show that she did worse than Obama, according to both exit polls and pre-election polls. Obama won by an average of 38 percentage points among Latinos in 2012, according to pre-election polls analyzed by the New York Times’ The Upshot blog, compared to Clinton’s average 33-point lead in the polls leading up to Election Day this year. Keep in mind too that Clinton did a little worse overall than the national polls suggested, so she may have done a little worse with Latinos, too.
Latino Decisions also points to another source of data: actual election results from counties across the country. A slideshow that Latino Decisions put together after the election shows that heavily Latino counties voted for Clinton by wide margins — she often won those counties by 40 or more percentage points.3
The trouble for Clinton is that although she won Latino-heavy districts, she lost ground in many of them compared to how Obama performed there four years earlier. There are 24 U.S. counties in which Latinos made up at least three-quarters of the voting-age population in 2015; Clinton’s margin of victory was smaller than Obama’s in 18 of them, by an average of nearly 10 percentage points.4 Clinton also underperformed Obama in five of the six counties where Latinos make up at least 90 percent of the voting-age population.
|COUNTY||HISPANIC SHARE||2016 VOTES||OBAMA||CLINTON||CHANGE|
|Jim Hogg, Texas||91.1||2,119||+56.7||+56.9||+0.2|
Clinton didn’t underperform everywhere. Latino Decisions points to a number of heavily Latino precincts in Chicago, Los Angeles County and Miami-Dade County (which has a heavily Cuban population) where Clinton outperformed Obama. She also did better in El Paso and San Antonio, Texas, which were not specifically highlighted by Latino Decisions. Those are all heavily populated areas, unlike some of the rural Texas counties where she underperformed. It’s possible that Clinton’s strength in those larger counties was enough to make her nationwide margin among Latinos wider than Obama’s. But even in the few areas where Clinton outperformed Obama, she rarely did so by as much as we would expect if Clinton improved her margin among Latinos by as much as Latino Decisions found in their survey.
Also, in a number of heavily populated areas, Clinton did worse than Obama in the precincts and districts where a lot of Latinos live. Latino Decisions laid out a long list of precincts and districts in which Clinton performed well, including parts of Las Cruces, New Mexico; Milwaukee; New York City; Cleveland; and Kissimmee, Florida. In nearly all of them, Clinton’s margin of victory was smaller than Obama’s. Her margin was also smaller in some Latino-heavy areas of Chicago and Los Angeles that weren’t mentioned by Latino Decisions. (Latino Decisions also highlighted Clinton’s performance in Arizona, where her margin was roughly the same as Obama’s in 2012.)
Barreto said Latino Decisions is primarily concerned with determining the share of the Latino vote that Clinton and Trump won, not how this year’s outcome compares to 2012’s. “We’re not necessarily interested in comparisons to Obama, although that will eventually be part of our story,” Barreto said. “We’re just trying to estimate what Clinton’s vote share was and what Trump’s vote share was in 2016 on Election Day.”
The county-level data also points to a larger issue: Clinton did significantly worse than Obama overall, both nationally and in most individual counties. That means that to have won Latinos by a larger margin than Obama — and especially to have won by 9 points more, as Latino Decisions’ data implies — Clinton would have to be finding much more support among that group than Obama did even as evidence suggests she got less support from every other racial and ethnic demographic. County-level election results suggest Clinton lost less ground among Latinos than among other demographic groups, especially non-Hispanic whites, but she still seems to have lost ground, not gained it.
Voting results don’t prove that Clinton did worse than Obama among Latinos, or that Trump did better than Romney. But the results do suggest that if nearly 80 percent of Latinos voted for Clinton, as Latino Decisions argues, then Latino turnout must have been down in many counties, or Clinton must have done much worse than Obama among non-Latinos in those counties. Otherwise, the overwhelming pro-Clinton Latino vote would have swung heavily Latino counties more dramatically toward Clinton. The evidence, then, suggests that Clinton fell short among Latinos in one of two ways: Either she didn’t win as large a share of them as Obama, or she didn’t convince as many of them to turn out to vote. Since both the exit polls and Latino Decisions agree that turnout among Latinos was up, the latter explanation doesn’t seem likely.
Still, even if Clinton’s margin with Latinos did shrink, the full story of how Latinos voted in this election may never be fully known. We will eventually get detailed data on who voted, which will allow us to make some inferences about turnout among various demographic groups. But even then, we won’t know how individual people voted — Americans, after all, cast secret ballots. Indeed, experts still argue over how Latinos voted in 2004, when some exit polls showed George W. Bush winning 44 percent of the Latino vote. Then, as now, Barreto argued the true share was much lower.
Farai Chideya contributed reporting.