A data-driven controversy has erupted over a finding in the entrance poll of Nevada’s Democratic caucusgoers Saturday, with the results of the poll among Latinos seemingly at odds with widely held expectations that the group was a strong one for Hillary Clinton.
The results of the Nevada entrance poll Saturday showed black voters squarely in Clinton’s corner, as expected, while Latino voters broke for Sanders over Clinton by an 8-point margin, just outside the margin of sampling error for the group.
The latter finding prompted doubts in some corners, including from the Clinton campaign. Their criticisms mostly rest on an analysis of the vote by precinct, conducted by the New York Times. Looking at the vote by precinct in locations with a higher concentration of Hispanic voters, the analysis suggested Clinton won more county delegates in those precincts than those in other locations.
With the slate of upcoming primary and caucus states shifting from nearly all-white Iowa and New Hampshire to a more diverse batch of states including Nevada, South Carolina and many of the states set to vote on March 1, whether the racial and ethnic minorities in those states would prove a firewall for Clinton’s campaign against Vermont Sen. Bernie Sanders has been a sharp focus for both campaigns.
Polling issues
The preferences of Latino voters as reported in exit and entrance polling has been a source of controversy for more than a decade, since a 2004 national exit poll finding estimated that Republican President George W. Bush captured 44% of the Latino vote in his re-election bid. Since that time, some have argued that the exit poll often underestimates the Democratic share of the Latino vote, or that it misrepresents the preferences of Latinos more generally, as is the argument here.
Those arguments often rest on comparisons to data collected in telephone polling of Latinos that are either sampled from areas with a high concentration of Hispanic residents, or from lists of Latino voters identified by their Hispanic surname.
In both instances, these surveys miss a substantial share of the Hispanic population, either by not drawing samples from places where Latinos are not heavily concentrated, or by ignoring those who do not have a Hispanic surname, and that’s on top of the general difficulty of identifying likely voters in telephone surveys.
In this case, the comparison is to county delegate counts in precincts that appear to represent heavily Hispanic areas, again based on a surname analysis.
That analysis assumes proportional turnout by race within any given precinct, and perhaps more importantly, assumes that once caucusgoers arrived, the vote along racial and ethnic lines for those Latinos who voted in heavily Hispanic precincts was about the same as that seen in Hispanics voting in other precinct locations around the state, something that can’t be proven by either entrance polls or this style of precinct-level analysis.
Still, exit and entrance polls have their own methodological challenges when measuring the vote among subgroups of the population, such as Latinos, that are not evenly spread throughout a state. A sample of precincts that has three heavily Hispanic precincts in one election year and only one in the following year could suggest a drop in turnout among Hispanics, for example, but that finding would be better chalked up to sampling error than any real change in the electorate.
In 2016, age is key
But there are reasons to believe that the entrance poll findings reflect a race among Latinos that is actually tighter than was broadly expected before this first contest featuring a substantial minority population, and the key is age.
Latino voters in Nevada are generally younger than whites or blacks in the state, and as we’ve seen in both national polls and exit polls so far, younger voters are perhaps more apt to back Sanders than any other group in the Democratic electorate.
According to the Census Bureau, the median age of all Latinos in Nevada is 26 years old, compared with a median age of 44 for whites and 33 for blacks. And looking at past votes in the state, the general election exit poll there in 2012 found that 34% of Latino voters were under age 30 compared with 9% of whites and 27% of blacks.
In Saturday’s entrance poll, 38% of Latinos who showed up to caucus were under age 30, similar to the share in the 2012 presidential election exit poll, but almost double the share of Latino caucusgoers who were under age 30 in 2008, when Clinton narrowly carried the group with 20% support, just ahead of John Edwards and Barack Obama. In Saturday’s data, 14% of blacks and 12% of whites were under age 30.
Those younger Latinos broke sharply for Sanders over Clinton in this entrance poll, while Latino voters over age 30 broke in Clinton’s favor by about a 2-to-1 margin.
The real wildcard is whether that sharp age divide combined with heavier turnout among younger Latinos carries over into any other states where Latino voters make up a significant share of the population, and that assessment will likely have to wait until after South Carolina’s Democrats go to the polls.
This week: South Carolina, Nevada – Take two
Looking ahead to the Democratic side of South Carolina’s primaries, polling conducted mostly last week has found Clinton holding a broad lead over Sanders in the run-up to that contest, set to be held Saturday. A new CNN Poll of Polls covering live-interviewer telephone surveys conducted between February 10 and 17 finds Clinton averages 57% support while Sanders holds 32%.
Most of the polling conducted in South Carolina shows Clinton’s candidacy boosted by strong support among black voters, who are expected to make up a majority of Saturday’s primary voters. It may be the first opportunity to look in-depth at the demographic divides among black voters, to see whether Sanders’ appeal to young voters extends to African Americans in their 20s.
Another element to watch for Saturday that could boost Clinton’s numbers: Turnout.
Every Democratic contest held so far has seen turnout drop compared with 2008 levels, while on the Republican side, turnout has climbed.
South Carolina is the first state with a truly open primary. There is no party registration, so any voter can choose to participate in either party’s event, but they can only choose one.
With the Republican primary having drawn record turnout, it’s possible the numbers participating on the Democratic side will be lower and more concentrated among the party’s faithful. Those who consider themselves Democrats have been more firmly in Clinton’s corner across contests this year, and the CNN/ORC Poll in South Carolina showed Clinton’s margin even larger among self-identified Democrats.
For the Nevada Republican caucuses, little methodologically-sound polling has been conducted, but a CNN/ORC Poll conducted before the South Carolina primary suggested Donald Trump will head into Tuesday night’s event with a wide lead over the field, garnering the support of more than 4 in 10 of those who say they plan to caucus.