There’s been some head-scratching about the entrance poll result; The New York Times suggested that Sanders’ 8-point win among Hispanics was an unreliable finding, perhaps distorted by the vagaries of cluster sampling.
Entrance and exit polls aren’t perfect, for sure; extrapolating from precinct populations to caucus-goers is pretty fraught in itself. And in fact there’s a good reason for Sanders to have done well among Hispanics: They’re young.
It’s well-established that Sanders has been tearing up the house with young voters. He won 84 percent of caucus-goers under the age of 30 in Iowa, 83 percent of under-30s in New Hampshire and 82 percent of millennials in Nevada. In a word, wow. Compare that to Barack Obama, who famously whomped John McCain among young voters in 2008, managing 66 percent of their votes.
So now let’s look at Hispanics. There were inadequate numbers of racial and ethnic minorities to analyze in Iowa and New Hampshire, but not so in Nevada. Hispanics accounted for 19 percent of voters – 213 respondents among the total sample of 1,024. That’s enough to evaluate given a probability-based sample.
What do we see? Per the entrance poll, Hispanics participating in the Nevada caucuses were nearly three times likelier than other caucus-goers to be younger than 30, and less than half as likely to be 60 or older.
Age matters. Hispanics younger than 45 voted 70-27 percent for Sanders over Clinton in Nevada, while non-Hispanics under 45 voted almost exactly the same, 73-24 percent. There simply were proportionately more of the former. (Further, while the sample size of Hispanics age 45 and older is small, their vote was more than 2 to 1 for Clinton – again, similar to her result among non-Hispanics in that age group.)
Internal validity takes us only so far, but there’s also external validity for the age-by-ethnicity differences in Nevada. Per the U.S. Census Bureau, the median age of Hispanics in the state is 27.5, while the median age of non-Hispanics is 42.4. The fact that Hispanics in the state are younger than non-Hispanics would seem to support the notion that Hispanic caucus-goers were younger, too, and thus more apt to be Sanders supporters.
In all, then, there’s decent evidence that the estimate is a good one – and that Sanders did well among Hispanics not on the basis of their ethnicity, but because of their age.
The best argument for questioning whether Sanders in fact won Hispanics is that his 8-point advantage is less than the entrance poll’s margin of sampling error for this group, 10 points. Regardless, a little reality check is in order.
Even if Sanders gets bragging rights for his results among Hispanics in Nevada, Hillary Clinton’s 54-point win among blacks seems substantially more significant. The vastly wider margin is one reason. There likely will be more blacks than Hispanics voting in the primaries to come. Across the 2008 Democratic primaries for which we had exit polls, blacks accounted for 19 percent of voters, Hispanics for 12 percent. Blacks soared to 55 percent of voters in South Carolina, the next state on the Democratic calendar.
Also, blacks are more widely distributed across the country, and so can make their votes felt in more states. Hispanic Democratic primary voters in 2008 were concentrated in just nine states – New Mexico, Texas, California, Arizona, Illinois, Nevada, New Jersey, Florida and New York. (It's important to note that Clinton is a native of one of those states and a former senator from another.)
So Clinton can let Sanders have his 8-point lead among Hispanics, so long as she can keep her large margin of 54 points among blacks.