People vote at a polling place in the heavily Latino East L.A. area during the U.S. presidential election on November 6, 2012 in Los Angeles, California. (Photo/Getty Images)

Why some polls got the Latino vote wrong

Some pollsters saw it coming. For months, Latino polling and research organizations as well as some national polling organizations, including NBC News/WSJ/Telemundo, had Hispanic voters favoring President Obama by very large margins of about 50 points.  So the election results were no surprise, and the Latino vote was a key, decisive factor in Obama’s victory, especially in important battlegrounds such as Colorado, Nevada and from the way it is looking, Florida.   Yet before the election, some state and national polls had Republican candidate Mitt Romney more evenly matched among Latinos, thus tipping predictions in states like Florida and Colorado more in Romney’s direction.

“They completely missed the Latino vote; they got it all wrong,” says Latino Decisions’ and University of Washington political scientist Matt Barreto. “It  makes me mad when people who don’t know anything about Latinos question our understanding about the Latino electorate,” he says.

RELATED: Analysis: Pollsters missing the mark on Latinos

A Tampa Bay poll  from a few weeks ago, for example, had 44 percent of Latino voters favoring Obama and 46 percent favoring Romney, whereas recent NBC/WSJ/Marist polls as well as Latino Decisions polls had Obama support among Florida Hispanics around the higher 59 percent mark. It was the same things with national polls, as pointed out recently by the statistician Nate Silver in his New York Times blog.  He quoted interviews with Barreto on how polls have underestimated Latino populations in several states by not using Spanish-language pollsters, among other factors.

Pew Hispanic’s Associate Director Mark Hugo Lopez says that to adequately poll Latinos in states such as Florida, you have to  make sure you have pollsters who can switch from English to Spanish if need be, as well as make sure you go by census figures to adequately sample Latinos of different national origin.  This is crucial in a state like Florida, where recent polls seemed to oversample Cuban-American voters, who are more traditionally Republican, but who represent less of the electorate as the share of other Hispanics, such as Puerto Ricans and Central Americans, have gone up in the state.

“It is expensive to have fully bilingual people polling,” says Lopez, “but it is an important component in Pew Hispanic,” he explains.  Another issue is land lines versus cell phones; younger and less affluent Latinos might have cell phones but no land lines.  Lopez says Pew Hispanic uses 50 percent cell phones in their surveys.

RELATED: Beware of how pollsters call the Latino vote 

In Florida, Barreto thinks some national opinion polls missed the growing Hispanic population in Hillsborough and Orange counties.  In  Nevada, he says, a recent poll had Latinos favoring Obama over Romney 52 by 44 percent, even though Latino Decisions exit polls has the Obama Hispanic vote at 80 percent.  In fact, says Barreto, he still thinks the national exit polls for Latinos in Ohio and Virginia have undercounted the Hispanic vote for Obama.  “In states with smaller Latino populations, the polls tend to only be conducted in English, and they also tend to oversample Latino college-educated voters,” he explains.

So what now?  University of Miami’s Casey Klofstad says it was a “disservice to the public and misleading” to get the Latino vote so wrong.  Stanford University political scientist Gary Segura agrees and adds this has happened a few times before – and the question is if it will continue  happening again.

%d bloggers like this: