Hispanic votes are easy to quantify when analyzing votes in 100-percent Hispanic precincts and comparing those percentages to actual exit polling on Election Day. The further away from 100-percent Hispanic precincts, the less confidence in numbers.


It is also disingenuous to suggest that Hispanics think alike on politics and elections.

First, this problem: Hispanics with non-Spanish names that result from intermarriages of Hispanic women and non-Hispanic named spouses under-influence poll and election results.

Pew Research's Hispanic Trends has studied Hispanic intermarriage and concludes that one in four Hispanic women marry "out" of their ethnic group, regardless of national origin. Literally thousands of Hispanics with non-Spanish names proliferate throughout the country because their Hispanic mothers marry non-Spanish-named men every year.

Those studies also indicate that these women and their spouses have higher incomes than the community norm; higher incomes generally mean better education and better neighborhoods than, say, the poverty-ridden Rio Grande Valley in Texas.

The situation creates a rift among political observers and pollsters.

The "rift" includes these problems:

Better-off Hispanics vote in greater percentages and frequency than those in poverty areas. Combining those attributes with non-Spanish names, there is a built-in under-sampling — bias — of these better-off Hispanics regardless, of the methodology used by pollster.

Some pollsters declare that the existing national sample — exit polls — aren't accurate for several reasons: (1) the sample is too small; (2) interviews are not done in Spanish; and (3) poverty areas are ignored (such as Spanish Harlem in New York and the Rio Grande Valley).

This rift has become public by a University of Washington professor, Matt Barreto, who disputes the figures on Hispanic voting done by Edison Research on Election Day. Barreto is a co-founder of Latino Decisions, another polling firm.

Edison does Election Day exit polling for a conglomerate media group consisting of the Associated Press, NBC, Fox, ABC and CNN. It is the national standard.

Barreto objects to Election Day exit polling results; samples are too small and no Spanish-language polling is done by Edison. However, size of samples are always disputed. Sampling in Spanish is not significant, nor even interesting, because even naturalized citizens are required to read English. In Spanish or English, respondents know who they voted for. How does one say "Sam Brownback" in Spanish?

Edison reported that Kansas Gov. Sam Brownback (R) received 47 percent of the Hispanic vote and in Georgia, Gov. Nathan Deal (R) also received 47 percent of the Hispanic vote.

Barreto disputes those percentages. He says that his organization interviewed 4,200 Hispanics in English or Spanish for five days leading up to the election, not a few hundred like Edison did on Election Day after they voted — thus his numbers are more accurate.

He claims his organization found that before the election, Brownback only had 31 percent support, 16 points below Edison's Election Day exit polling; In the Georgia gubernatorial race, Barreto found only 27 percent Hispanic support, 20 points below Edison's findings on Election Day.

Barreto publically declares that "educated" Hispanics are over-sampled by Edison and that Edison did not survey in the Rio Grande Valley. He offered no proof of his assertion.

Back to Georgia, Atlanta’s 11Alive TV station sponsored surveys by SurveyUSA (a professional polling company) for the gubernatorial and Senate races that started in April.

The April poll showed Deal with 38 percent Hispanic support. The August poll showed him with 43 percent Hispanic support. The Sept. 19 poll showed Deal with 40 percent Hispanic support. September's second poll showed Deal with 40 percent Hispanic support and U.S. Sen.-elect David Perdue (R) with 44 percent Hispanic support, while his Democratic opponent had 32 percent support.

Barreto and colleague Gary Segura challenged 11Alive's survey report in a letter as "wildly incorrect speculation."

According to Election Day interviews conducted by Edison, Deal scored 47 percent Hispanic support, substantially higher than he polled in 11Alive's polling over six months. Perdue scored 42 percent Hispanic support.

Edison's numbers and percentages are based on Election Day interviews after people voted. That is more trustworthy in these precincts than polling before Election Day.

Now, about 2016.

The problem is that national origins profoundly influence views of and votes for candidates; that difference totally skews surveys of Hispanics.

Dominicans now outnumber Puerto Ricans in New York, but they vote the same; the same for Puerto Ricans in New Jersey, Florida and Chicago. They are and vote Democratic.

Is the reported 2012 Hispanic vote really more about die-hard Dominican and Puerto Rican Democratic voters on the East Coast voting for Obama than how he did with Western Mexican- Americans? We don't know.

But the Census Bureau reports that two-thirds of the "Hispanic/Latino" population is not Puerto Rican/Dominican or Cuban, but Mexican-American and that population dominates Hispanic populations in the South, the Midwest, the Southwest and the West. They are not left-leaning as a rule; surveys by groups such as Pew Hispanic Trends report that Mexican-origin voters tend to be moderate and conservative. Historically, they vote in higher numbers for Republicans than their Dominican and Puerto Rican cousins in the East.

The only Latino/Hispanic governors in the United States are Mexican-American and Republican: Brian Sandoval of Nevada and Susana Martinez of New Mexico.

Pollsters and survey takers, including Election Day exit pollsters, should separate results by doing separate polling between these two Hispanic groups — Mexican and non-Mexican. That would give us a realistic look at the "Hispanic" vote — a regional and state specific look that would be more realistic than presently done by anyone.

This piece has been corrected. A sentence erroneously stated that there has never been a Hispanic Democratic governor; Bill Richardson and Raul Castro were Hispanic Democratic governors of New Mexico and Arizona, respectively.

Contreras formerly wrote for the New America News Service of The New York Times.