As part of the Latino Studies Reunions Event on May 31, 2013 Professor Ali Valenzuela (Politics) will speak about the “Latino vote” in the 2012 election and its significance for the growing political power of Hispanics. More details about Valenzuela’s topic follows:
“Scientific Polling y el Voto Latino” — by
Ali A. Valenzuela (Assistant Professor of Politics)
During the 2012 election, scientific polling and polling aggregators such as Nate Silver at The New York Times, Simon Jackman at The Huffington Post, and our own Sam Wang at the Princeton Election Consortium entered the national consciousness like never before. Polling aggregators’ rise to prominence was due in part to the availability of massive amounts of state and national polling data, which the aggregators combined into sophisticated statistical models used to predict election outcomes in individual states as well as nationally. Several of the predictive models turned out to be highly accurate and immediately debunked claims of unfairly skewed polls favoring the Democrats that had been circulating in the weeks before Election Day. The polls did favor the Democrats, but that was because the Democrats were ahead in the presidential contest.
There are two important points in this conclusion. First, party identification – the level of support for the Democratic or Republican Party among the American public – changes from election-to-election, year-to-year and sometimes from poll-to-poll. Enthusiasm for the candidates, the state of the economy, and candidate positions on major policy issues like healthcare or U.S. military action abroad can all affect the proportion of Americans that identify as Democrats or as Republicans in a given poll. Just because a poll shows more Democrats or more Republicans than in the previous election does not mean that the poll is inaccurate. Second, scientific polling, and especially the combined average of many scientific polls, provides a very accurate snapshot of the level of support for a candidate in a national election. Scientific polling within states, and especially the average of many scientific polls within a state, provides a very accurate snapshot of state-levels of support for a candidate.
However, when it comes to polling Latino voters, many commercial firms do a poor job because of challenges and costs associated with interviewing voters in Spanish. For example, most polling firms employ a call back method in which, when they encounter a Spanish-speaker, they hang up and schedule a call-back with someone who can carry out the interview in Spanish. This has the effect of reducing response rates and the number of Spanish-only interviews that are successfully completed. Utilizing fully bilingual callers for Latino polling is expensive but necessary when upwards of 45% of the Latino electorate consistently prefers to interview in Spanish. This figure is among registered Latino voters who are all American citizens, to say nothing of the overall Latino population.
More critically, many polling firms, including the National Exit Poll, which interviews voters as they leave their voting places, do not interview in Spanish at all (more information). Exit Poll results are widely reported by the media, yet the effect of English-only interviewing is to bias results for Latinos towards those who are more acculturated: those who speak better English and live in wealthier parts of the country. Research consistently shows that more acculturated Latinos tend to hold more conservative policy attitudes and vote preferences more favorable to the Republican Party than among less acculturated Latinos. So polls that do not interview in Spanish, or that interview a lower-than-average proportion of Latinos who speak only Spanish, produce results that do not accurately reflect the Latino vote nationally or in states with a sizable Latino population.
The conservative bias of English-only Latino polling tends to produce results showing Republican candidates with more Latino support than in reality. In 2010, inaccurate Latino polling created a situation where Senator Harry Reid (D-NV) was predicted to lose his reelection bid for the U.S. Senate by a slight margin, but he instead won by 5.6 percentage points over his anti-immigrant challenger, Sharron Angle. Senator Reid won in large part because of the Latino vote, which supported his candidacy by a 90-8 margin, a huge gulf in Latino support between the two statewide candidates (more information). With Latinos representing about 12% of the Nevada electorate in 2010, their 82-point margin in favor of Reid translates into almost 10 additional percentage points for the Democratic win column, a figure greater than Reid’s margin of victory. Latino voters were pivotal to Harry Reid’s reelection and his return to the U.S. Senate as Majority Leader.
Fast-forward to 2012 and the crucial nature of the Latino vote repeated itself in at least four states: Colorado, Florida, Nevada and New Mexico. In these states, three of which were highly contested “swing” states (CO, FL and NV), Latinos’ level of support for President Obama was so much greater than their support for Governor Romney that they made the crucial difference in winning these states for the President. This outcome depended on two key factors: one, which I have already noted, is the large gap in Latino support between the Democratic and Republican candidates, an average difference of 54 percentage-points in these four states. Two, the Latino electorate in these four states—that part of the Latino population that was eligible and turned out to vote—was large enough to translate the gap in support between the two candidates into a substantial and pivotal vote contribution for a Democratic win. Without the Latino vote, and without such a wide margin of support for the Democrats among Latinos, it is unclear whether President Obama would have carried the day on November 6, 2012.
How did we get to such overwhelming Democratic support among the Latino electorate? As recently as 2004 approximately 40% of Latino voters supported George W. Bush, a vote margin with then-Senator Kerry of only 20 percentage points (more information). In my presentation, I will report more detailed Latino polling results from 2012 and discuss how we might understand such high levels of Democratic support among Latinos today.