Latinos turned out in record numbers this election, and supported Clinton over Trump by an historic margin (79% to 18%). Unfortunately, a faulty exit poll number is undermining Latinos’ strong performance this cycle.
The exit poll is just one survey, it is not actual results It is not fact, it is a survey providing an estimate.
The national exit polls, conducted by Edison Research and sponsored by media organizations, said that Donald Trump won 29% of the Latino vote. That number doesn’t pass the laugh test. For example, how could Democrats have won big in Nevada if nearly a third of Latinos supported Donald Trump? It doesn’t make sense, and it certainly doesn’t match the experiences of canvassers on the ground.
Latino Decisions analysis of actual election data shows that turnout was up in Latino-dense precincts across the nation. And precinct level analysis shows Trump trailed Romney in an almost all majority Latino precincts, a fact county-level analysis glosses over. A fuller report analyzing this data in over a dozen states will be published soon.
When the political behavior of a minority group is unfairly and inaccurately diminished, it’s akin to the disenfranchisement of these same voters. Unfortunately, the 29% figure is being used in many places. Even NPR’s Code Switch, which is supposed to engage real talk on issues like these, bought the mainstream media myth. We cannot allow the media to use bad data that misrepresents the Latino community.
The media-sponsored exit polls are notoriously flawed at capturing “sub groups” like Latinos. Even Edison Research has admitted as much. Watch Pilar Marrero of La Opinion explain the exit poll flaws on the “Tavis Smiley Show” here, and read more from Paul Waldman in the Washington Post here.
There is also a lack of transparency in the media-sponsored exit polls. Edison has been silent on key methodological questions, and that silence raises suspicion. We do know that Edison does not poll a random sample of Latinos statewide. But we do not know if they have fully bilingual staff to recruit respondents. We do not know the demographic characteristics of their Latino sample, so there’s no way to determine whether it is anywhere near representative of the population. How can we trust a poll’s findings if they route they took to get to them is shrouded in secrecy?
Latino Decisions, in contrast, specializes in Latino polling and uses a robust set of tools to ensure their data is representative of the community. Their Election Eve poll had 18% of Latinos voting for Trump, which much more closely matches what canvassers saw on the ground. Read more about how Latino Decisions does their research here.
Many media organizations ignore the problems with Edison’s Latino numbers, and continue to use them, because they paid for the polls. But the fact remains that the number is wrong, and repeated use is helping to turn this myth into an accepted “fact” if we don’t push back.