Back in July, NBC News ran a headline proclaiming “GOP 2016 Win Will Need More Than 40 Percent of Latino Vote, Says Study”. Other news outlets, such as the Los Angeles Times and Fox News Latino, ran similar headlines. But are such predictions correct? And, if so, can such a level of support be reached by the same party who is seeing Donald Trump in the lead of the polls?
The research that those headlines is based off of comes from political scientists David Damore and Matt A. Barreto, published through the political analysis firm Latino Decisions. Their analysis of voting trends and demographic data leads them to the conclusion that, in order to win the Presidency in 2016, the Republican candidate will need to win somewhere between 42 and 47 percent of the Latino vote. This leads them to the conclusion that “…the conventional wisdom often asserted by pundits and in media reports that winning 40 percent of the Latino vote will be sufficient for the Republicans to carry the presidency in 2016 is farcical.” This is terrible news for Republicans: Mitt Romney only got 27 percent of the Hispanic vote in 2012.
However, the above analysis is missing something. Damore and Barreto’s math holds true for the popular vote — the raw number of supporters for a candidate. However, as any Al Gore fan will tell you, the President isn’t selected through a popular vote, but rather through the Electoral College. What matters then is not the raw number of Latino voters supporting the Republicans in 2016, but how Latino voters will influence the outcomes of state decisions in the election. This difference is critical, as state-centric electoral structures play the unintentional role of reducing the electoral influence of Latino voters below what it would be in a popular election. Nate Cohn at the New York Times explains:
Hispanics are disproportionately concentrated in large states, like California, Florida and Texas. Incredibly, Hispanics represent an above average share of the population in only nine of the 50 states. There are very few Hispanic voters in most small states, like Wyoming or the Dakotas, and small states are overrepresented in the political process, thanks to the structure of the Senate. Effectively, the Hispanic share of the eligible Senate electorate is just 7.5 percent….
Hispanics are earning more clout in presidential elections…. Yet [they] are still punching beneath their weight… ln battleground states, Hispanics represent more than 5 percent of eligible voters in only Florida, Nevada and Colorado. As a result, President Obama would have easily won re-election even if he had not won the Hispanic vote; he would have won the Electoral College even if he had lost all three of those states.
This geographical imbalance can have real significance. It could even mean that the Latino vote might not be quite that important to the Republicans after all. So, exactly how important is it? Fortunately, we have the tools we need to check. Using a demographic election tool from RealClearPolitics, past voter turnout rates by race from the U.S. Census Bureau, and past voting preferences by race from the University of Connecticut’s Roper Center, we can create our own estimates. In order to determine its importance, let’s assume the best case scenario for Republicans and then see how much of the Latino vote they need.
We’ll assume that white voter turnout returns to its highest level in the last five elections (67.2 percent in 2004), while black voter turnout falls to its lowest level (53 percent in 1996). Then let’s assume that white support for Republicans reaches its five-election high (59 percent in 2012) and black support for Republicans does the same (12.5 percent in 1996, excluding those who supported Perot). Let’s then assume that 52 percent of the Asian vote goes for the Republicans (1996, again excluding Perot supporters) and that they return to their highest voter turnout rate in the last five elections (47.6 percent in 2008). In this scenario, with the Latino voter turnout at its five-election average of 46.84 percent, the Republicans would still need 38.2 percent of the Latino vote to win the Electoral College.
There are, of course, a number of major problems with this rough math. First, it is possible that any of the above numbers will break their past highs or lows, meaning that an even more extreme scenario could be possible. Indeed, with racial issues like immigration and police brutality standing out as major political issues during this election cycle, the racial voting demographics in 2016 may be substantially different than they were in any of the five previous elections, reducing the predictive power of past numbers. Second, this analysis also ignores the impact that a third party or third parties may have on the 2016 election, which is something that has certainly been disruptive before. Finally, relying exclusively on this type of analysis leads us to a crude sort of “demographic determinism”, where we begin thinking that race and other forms of demography become the sole determinants of election outcomes, ignoring the long list of other extremely important factors. Doing this would be a serious mistake.
However, the findings of this back-of-the-napkin math are still enormously significant: within the parameters of electoral participation we saw in 1996, 2000, 2004, 2008, and 2012, even the best case scenario would require the GOP to acquire 38.2 percent of the Latino vote. If we rely on data from the Pew Research Center, they have never gotten this level of support in the last 9 elections. If we rely on data from the Roper Center, the Republicans have only gotten it once (2004) in the last 10 elections, but may have gotten had John B. Anderson not run in 1980. Thus, regardless of whether or not Damore and Barreto’s estimates will prove to be extremely useful, their point stands loud and clear: Latino voters pose a serious threat to the Republicans. The GOP will need a level of support that will be difficult for them to achieve, even under normal circumstances.
The emphasis in that last sentence is on “under normal circumstances”, because the current field of GOP primary candidates is far from normal. At the moment, RealClearPolitics estimates that the leader of the field is currently Donald Trump, with an average of 30.4 percent of the vote. Donald Trump, who has made nativist and anti-immigrant sentiments the staple of his campaign, has been driving other Republican candidates to the right on the issue of immigration.
One such example of the rightward shift on immigration is the fact that previously taboo ideas like the removal of birthright citizenship have been forced into the mainstream, as candidates seeking to steal Trump’s fire copy his positions. Ted Cruz, Bobby Jindal, Rand Paul, and Rick Santorum have all joined Trump in arguing that the right of citizenship shouldn’t be automatically given to everyone born in America, with Ben Carson, Chris Christie, and Lindsey Graham effectively taking his side without explicitly taking a position. Scott Walker, in the style of any good politician, has danced around the question. Though such an idea may be popular among Republican candidates, a 2011 poll from the Pew Research Center found that only 23 percent of Hispanics supported the idea of “Chang[ing] [the] constitution to bar citizenship for illegal immigrants”. It’s common to see candidates appealing to the more “extreme” ideological members of their party during the primaries, but the idea only had 57 percent of support among Tea Party supporters — hardly an overwhelming majority.
Though Trump believes, in his words, that “the Hispanics love me”, he is actually viewed unfavorably by a whopping 82 percent of Hispanic people, and is viewed strongly unfavorably by 68 percent. That may have something to do with his now-infamous accusation that immigrants are criminals, drug smugglers, and rapists. Alternatively, it may be inviting people like Ann Coulter, who believes that “There’s a cultural acceptance of child rape in Latino culture” and wants to make watching drone strikes on undocumented immigrants a tourist attraction, introduce him at speeches. But even if Trump fails to capture the nomination (which almost everyone expects will be the case), forcing such a trend towards harsh measures on immigration may mean that he’s damned the final Republican candidate to lose anyways. Any candidate who seriously wants to win the nomination is having to take an anti-immigration stance right now, despite the fact that such a stance will almost doom them in seeking the level of Latino support that they need. Polling shows that Latino Americans are more liberal on immigration than either white or black Americans, while simultaneously being more likely to demand that politicians agree with their views on immigration in order to earn their vote:
The issue for the Republican Party here is brutally clear. Republicans need a far larger portion of the Latino vote than they typically get while simultaneously needing more than ever to take stances that Latino voters oppose. The leadership of the Republican Party is completely aware of this issue, probably far more so than most actual Republicans.The report that the party put out briefly after losing the 2012 election, suggesting reforms to the party to improve their election chances, includes the following:
If Hispanic Americans perceive that a GOP nominee or candidate does not want them in the United States (i.e., self-deportation), they will not pay attention to our next sentence. It does not matter what we say about education, jobs or the economy; if Hispanics think we do not want them here, they will close their ears to our policies…
We are not a policy committee, but among the steps Republicans take in the Hispanic community and beyond, we must embrace and champion comprehensive immigration reform. If we do not, our Party’s appeal will continue to shrink to its core constituencies only.
No wonder the Republican establishment dislikes Trump. Over the past 10 years, the number of Americans identifying as Republicans has dropped from 29 percent to 23 percent, and the party knows that failure to adapt to changing demographics will only continue this process of having it “shrink to its core constituencies only.” And yet, this type of change, so deeply neccesary for the party’s survival, is being made impossible by the very man who is currently in first place in the race to lead it.
[2018 update: Following Donald Trump’s presidential victory, the Pew Research Center published numbers on the racial turnout rates for the 2016 election, and the Roper Center published their racial voting numbers. White turnout was 65.3%, with 57% supporting the Republican; both just minor rises over 2012, and below the rates I assumed in my scenario. Black turnout fell significantly to 59.6%, but only 8% supported the Republican; the turnout was higher and the support lower than the numbers I used in my scenario. Asian turnout was 49.3% and Asian support for the Republican was 27% (dramatically lower than the high rate I used). Latino turnout was 47.6% (higher than the number I used), and Latino support for Republicans was 28%.
Given all of this, what went wrong? Simple: the tool I used. I plugged the actual numbers back into tool, and it suggested that the exact same racial distribution of votes that actually occured in 2016 would lead to a blow-out in favor of the Democrats.
Noticeably, the problem appears to be in the tool’s translation of demographic data into a popular vote estimation: it estimates a result of 53.8% to 46.2%, dramatically different from the actual result of 48.2 to 46.1%. Part of the problem here is that the tool doesn’t take third party votes into account: something I should have taken note of the first time around. The tool’s model has a third party vote percentage of 0%, as opposed to the actual percentage of 5.7%. This is a pretty major omission, especially since third party vote share was so large in 2016 (compare that 5.7% figure to 2012’s 1.7%.) In fact, since the model’s prediction of Donald Trump’s vote share is only a tenth of a point off of his actual share, it almost seems like the model managed to predict the size of the third party vote and give all of it to Clinton; not the case, of course, but an odd coincidence.
Of course, third party votes alone can’t nearly account for the discrepancy between the model and the reality. The majority of third party votes in 2016 were for Libertarian Party candidate Gary Johnson, and evidence suggests that libertarians tend to support Republicans over Democrats, suggesting that much of the third party vote should “take away” from Trump’s vote share rather than Clinton’s (to the extent that third party voters would have voted for either Trump or Clinton if they didn’t vote third party).
So where did the model go wrong? The exclusion of third party votes (and, thus, the assumption that every non-Republican vote would go to the Democrats), surely distorted it, but another problem may have come as a result of the model’s methodology, specifically in its data selection.
To be honest, I don’t know exactly what went wrong with the model. If anything, though, I think that there’s a greater underlying lesson I’ve drawn from all of this, which is that models are exactly that: models. In a number of ways, Trump’s rise and ultimate victory represented a significant anomaly in American political tradition, making it very hard to see coming for electioneers who utilize past trends to forecast future outcomes. Any number of things may have occured either with the model itself or with the uniqueness of the 2016 election that made predictive modeling in this fashion futile. Trump’s presidency is a warning of caution to everyone, myself included, about using statistics and mathematics to try and see the future.]
Originally published at theodysseyonline.com on September 21, 2015.