New York Times
By Nate Cohn
February 22, 2016
The debate over who really won the Hispanic vote in Nevada, Hillary Clinton or Bernie Sanders, continues today. As I wrote Sunday, I think the balance of evidence points to Mrs. Clinton: Her strength in the heavily Hispanic areas of Las Vegas and among Hispanic voters in most national polls is, to my mind, much stronger evidence than an entrance/exit poll sample of 213 Hispanic respondents in 25 precincts.
The debate is important to both campaigns. The Sanders side is eager to promote that its message is connecting beyond white voters, where it has already had impressive success. The Clinton campaign would like to be able to say that its nonwhite coalition is holding together.
This helps explains why the poll had such a pro-Sanders vote, but it doesn’t prove the published results were right. In fact, it’s an unrealistic number that helps explain how the poll could have been off.
Just 23 percent of registered Democrats who are Hispanics in Nevada are 18 to 29, according to data from L2, a nonpartisan voter file vendor. The number grows to 27 percent if one includes all nonpartisan voters — many of whom do not lean Democratic.
The poll data found that the share of young voters was higher than the share of Hispanic adult citizens — 36 percent of adult citizen Hispanics are 18 to 29, according to an Upshot analysis of microdata from the 2010-2014 American Community Survey. It implies that the overall turnout among young Hispanic voters was actually higher than among Hispanic voters over age 29.
That’s unlikely given the consistent Census Bureau finding that the turnout among young Hispanic voters is extremely low. In 2012, just 34 percent of 18-to-24-year-old Hispanic citizens voted, compared with the much higher turnout rate (no less than 47 percent) among every older group. The low registration rate implied by the L2 data is consistent with the census data.
Perhaps the youth Hispanic turnout really did surge. But it seems more likely to be statistical noise in a small sample. The entrance-exit polls found much lower turnout among young Hispanic voters in 2008, so there is not much reason to assume that caucus electorates are especially likely to draw young Hispanic voters. The entrance-exit polls do not weight the sample by Hispanic origin, so it cannot correct for the possibility that young Hispanic voters were particularly likely to respond — whether just by chance or because of an underlying bias.
The high number of young Hispanic voters is the second smoking gun in the data. On Sunday, I noted the other key piece of evidence: how the entrance-exit poll showed Mrs. Clinton leading by just five percentage points in Clark County — the county where the preponderance of Hispanic voters live. She actually won by 10 percent (although it should be noted that vote and delegate shares aren’t necessarily the same).
As we said Sunday, none of this precludes the possibility that Mr. Sanders won the Hispanic vote. The technique of focusing on heavily Hispanic precincts has plenty of shortcomings.
But an entrance-exit poll of some 1,000 respondents and 25 precincts is simply not a reliable way to measure a clustered group like Latino voters. At that size, it’s all luck: Whether you include one or two or zero “heavily Hispanic” precincts can easily flip the entire result of the survey.
And that’s on top of all of the normal error that accompanies a small sample with a large margin of error.
For more information, go to: www.beverlyhillsimmigrationlaw.com