Cross posted on racesandredistricting.blogspot.com which has more election and demographic analysis.
Follow us on facebook at: www.facebook.com/...
After the Nevada caucus, Bernie and Hillary supporters are up in arms debating parts of it. There is the debate over whether people were yelling "English only" at Dolores Huerta. There is the debate over who has the most momentum but most importantly there is the debate over who won the Hispanic vote. The Sanders camp has long contended that they can win over voters of color and that Sanders is not simply a "candidate for white voters."
The debate starts here: roughly 70% of Nevadans live in Clark County, the most urban and Democratic part of the state (Obama won 58% there in 2008 and 56% in 2012) and Clark County has the bulk of Nevada's Hispanics (30% of Clark County is Hispanic, 45% of Clark County is non Hispanic White). The other 30% of Nevada includes Washoe County with Reno and the rural counties are nicknamed the "Cow Counties." The U.S. Census quick facts website shows that the Hispanic population percentage is lower in these counties than Clark County. Sanders narrowly carried Washoe County and the "Cow Counties." Clinton however won Clark County by 10 points which was the entire winning margin for Nevada by 5-6 points. This would suggest that Clinton won the Hispanic vote but some Sanders supporters are saying that the entrance polls show Sanders leading with Hispanics which means he won Hispanics. Heavily Hispanic precincts voted for Clinton but Sanders supporters say Sanders won Hispanics in precincts with lower Hispanic populations.
This article's goal is to not to prove that Hillary Clinton won Nevada's Hispanic voters but to show why exit/entrance poll crosstabs should not always be trusted, especially when judging the Hispanic vote.
We will examine this by looking at a few areas:
1. Clark County Voter Math
2. Why Exit Polls are not always reliable
3. Precincts with Medium Hispanic Populations
1. Clark County Voter Math
This point has been repeated thoroughly but in 22 heavily Hispanic majority precincts, Hillary Clinton won the majority of the delegates. The Hillary campaign used these numbers to estimate that they won 61% of the Hispanic votes. She also overwhelmingly won the casino caucus sites which have high Hispanic populations. She did not win by narrow margins either, one precinct I had a friend awarded 28 delegates to Hillary and 12 to Sanders. This map below is a demographic map of Las Vegas: (the red indicates white and green indicates Hispanic. As shown the large bulk of the Hispanic population is in East Las Vegas).
The map below shows how heavily Hispanic precincts voted.
African Americans were probably 14% of Clark County votes due to being 12% in the exit polls and the fact that outside of Clark County the African American population was low in Nevada. If the African Americans voted 76%-22% for Clinton (as shown by the exit polls) this means that Hillary Clinton won other voters by about three points. With heavily Hispanic East Las Vegas precincts voting for Clinton, it is safe to assume that the pockets of Sanders support came from areas with a higher White population.
2. Why exit polls are not always reliable
Exit polls have made mistakes with Hispanic votes previously. In 2004, they said Bush won 49% of Hispanic vote in Texas in 2004 and 56% in Florida in 2004. Results from heavily Cuban congressional districts and heavily Hispanic Osceola County seemed to confirm a result close to that, at least within a few points. Hispanics are a very diverse group of voters and usually do not vote in blocs and certain groups are known for being more Democratic than others (for example, Puerto Ricans have a reputation for being the most Democratic group of Hispanics). When looking at Texas however, the heavily Hispanic counties in the southern tip of the state (Rio Grande Valley) suggest that Texas Hispanics voted more than 50% for Kerry.
(map courtesy of the New York Times of the Rio Grande Valley with the Democratic counties in 2004).
Take Webb County as an example in the Rio Grande Valley. The population of Webb County is 94% Hispanic which means it can show us how Texas Hispanics voted without other demographics changing the results. Webb County voted 57% for Kerry. Other Rio Grande counties which were heavily Hispanic also backed Kerry, mostly by similar margins. One argument was that Bush performed better among Hispanics not in the Rio Grande Valley. The heavily Hispanic congressional districts outside of the Rio Grande Valley in San Antonio of TX-20 and TX-29 voted 55% and 56% respectively for Kerry according to the Almanac of American Politics 2006. Those numbers are lower than normal but still show that Hispanics voted more Democratic than 50%.
3. Using precincts with medium Hispanic populations
I have heard the argument that Sanders performed better with Hispanics in precincts with 20%-40% Hispanics because he won some in Clark County. He also won a few in Washoe County as well which does not have as large or as dense Hispanic population. This brings me to California's 21st Congressional District located in Bakersfield and some rural agricultural communities in the southern Central Valley. The district voted 52%-45% for Obama in 2008 even though it was 70% Hispanic. This created the misconception that Hispanics in the Central Valley were more conservative. In 2013 and 2014, Democrats lost SD-16, a State Senate seat there and Republican touted this by saying that Hispanics could vote Republican. At a closer examination, the data showed that Hispanics mainly voted Democratic. I examined precincts such as Lost Hills, CA which were 96% Hispanic and showed that Hispanics voted heavily Democratic. At the same time, many precincts that were 90%+ White voted 80% for McCain.
Another example is Tulare County. It is 63% Hispanic but voted 41% for President Obama. It was not however because Hispanics voted Republican but because Hispanic turnout was low in Tulare County. A few 90%+ Hispanic precincts were examined and while they backed Obama heavily (precincts in the southwest corner and in Orisi had 70%-80% margins for Obama), they had much lower turnout than surrounding whiter and conservative precincts. This shows that Hispanics in the Central Valley were not necessarily more conservative than Hispanics in Los Angeles. It shows that Hispanic turnout was low and the turnout from White voters was higher and most likely outnumbered Hispanics despite the higher Hispanic population. Reasons for the lower Hispanic turnout include that Hispanics are generally younger which means many are under 18 and many are not citizens and cannot vote. This is why it is important to use precincts to analyze Hispanics that are nearly 100% Hispanic because Hispanic turnout rates are generally lower than that of other demographic groups.
Map of Tulare County's population (red is primarily white, green is primarily Hispanic, a district with both is a mix)
Map of Tulare County's votes in 2008: (courtesy of Dave's Redistricting App)
Some analyses have pointed to precincts with sizable but not majority Hispanic populations to show areas where Bernie performed well. However it is unclear whether that Bernie support came from white voters or if it came from Hispanics. By looking at Tulare County though as an example, it appears that the White voters are overrepresented.
This explains my problem with justifying the Sanders win by looking at districts with medium sized Hispanic populations. As shown by Tulare County, Hispanics can be underrepresented in precincts with lower Hispanic populations. That is why I want to only look at East Las Vegas precincts because the precincts have such a high Hispanic population so other demographic groups cannot dilute their votes.
Overall, is Hillary or Bernie more popular with Hispanics? That question will only be fully answered when we see Texas vote on March 1st and see their exit polls. We can also examine the exit polls and see how closely they match Webb County and other heavily Hispanic counties in the Rio Grande Valley. So far, a YouGov poll shows Hillary with a 23 point lead among Texan Hispanics but as stated earlier, crosstabs are not always completely accurate but more likely than not they are.
My educated guess though is that Hillary probably won Hispanics (although we do not know the margin) because entrance/exit poll crosstabs are not always accurate but actual precinct results from very heavily Hispanic precincts are accurate. I will not however say with 100% certainty Hillary won Hispanics until Texas votes.