Latino Vote Simulations in Five States
Benchmarking where and when the Latino vote could be pivotal in the 2020 election
Equis Research partnered with People for the American Way on a series of vote simulations to answer a few related questions:
- What levels of Latino support and turnout do Democrats need to win in key battleground states?
- What is the possible impact of shifts in Latino performance?
- How do we gauge whether Biden is doing “well” or “poorly” with the Latino vote, beyond comparing his current polling to the imperfect marker of 2016 numbers?
- And lastly: if the high support Biden has recently enjoyed from (non-Latino) white voters were to come back down to earth, what kind of Latino numbers would he need to make up the difference — so that he can work to get them now, before it’s too late?
Ultimately, this was the kind of analysis we wanted to produce, now possible with our vote simulations:
Our analysis also helped to illuminate an important difference between Latino bulwarks like Arizona, Florida and Nevada, where Latino performance is uniquely critical to a Democratic win in most scenarios, and places like North Carolina and Pennsylvania, where Latino performance can provide a critical boost in a tight contest. The Latino vote matters in all these crucial battlegrounds, but when and how it matters depends on the unique demographic and political context in each state.
What this is / what this isn’t
Our aim is to identify when and where the Latino vote has the potential to be pivotal to the outcome of the election, to help campaigns and funders make wise resource decisions and to contextualize polling results for other observers.
Our goal is not to make a prediction about the ultimate outcome of the election, or to put a number on the likelihood that each candidate will win (a la 538 or The Economist.)
We approach this slightly differently in each state (with more detail below), but the main ingredients for all our vote simulations are:
(1) information about the size and racial/ethnic composition of each state, and
(2) the credible range of Biden support and overall vote share for key racial/ethnic groups in each state. The latter are based on both current polling and recent election results going back to 2012.
In order to generate useful projections, we often have to simplify reality by holding certain things fixed — the level of support and overall vote shares of non-Hispanic/non-White populations, for example. We are transparent about where and how we do this and about its purpose, which is not to minimize the importance of these groups to the outcome of the election, but rather to allow us to pinpoint the importance of the Hispanic vote in combination with other groups.
In fact, it’s critical to note that increases in performance among (non-Latino) Black or AAPI voters could also serve to make up for vote deficits identified here.
Data on racial composition of the electorate
Estimates on the racial/ethnic composition of the electorate come from three main sources: exit polling, census data, and analysis of voter file data.
These estimates can differ widely, even for the same election. (E.g. exit polls have Latinos as 6% of all votes cast in PA in 2016; voter file analysis puts it anywhere between 2% and 3.5%.)
The range of Latino vote share scenarios used here is intended to gauge the impact both of (a) actual shifts in turnout and (b) the different assumptions made by pollsters and analysts.
How to read the tables
The tables in this analysis report Biden’s projected share of the two-party vote in different scenarios. Therefore, >50% is a win in the state (green) and <50% is a loss (red).
To identify the conditions under which the Hispanic vote can be pivotal to the outcome of the election, we calculate the two-party vote based on a credible range of Hispanic vote shares and support levels.
As Hispanic vote share increases, non-Hispanic white vote share decreases commensurately. Each scenario holds the level of white support constant.
For purposes of the exercise, vote share and support levels for non-Hispanic Black voters and other groups in the electorate are held constant throughout at the historical average for elections since 2012 (based on estimates by TargetSmart) — except when otherwise noted.
Calculating the Two-Way Vote
For an apples-to-apples comparison, Democratic support is presented as a “two-way” number, to remove undecided or third party votes.
Biden needs the Hispanic vote to win in Arizona even in the most optimistic white vote scenarios. Our current polling puts Biden in a fairly strong position, as long as white support remains high. If white support drops back toward Clinton’s 2016 performance, it becomes critical for Biden to push up both Hispanic support and turnout.
At 47% white support for Biden — representing an average of recent public polling, and a four-point increase over Clinton’s number in 2016 — Biden and Mark Kelly (in the US Senate race) would both likely win the state if the election were today.
At 2018 levels of white support, just one point lower than his current average, Biden would currently lose the state (and Mark Kelly would be in a tight race). But they would both win with increases in turnout and/or support from Latino voters.
With an additional drop in white support, Biden only wins if he has significantly pushed both Latino support and turnout.
With a return to the levels of white support we saw for a Democrat in the state in 2019, pre-COVID — still a percentage point higher than Clinton’s 2016 showing — only the highest levels of Latino performance would save Biden.
Below 44% white support, we project Biden loses under any scenario.
Our calculations show Biden winning Pennsylvania if he can keep Black voter turnout and white support above Clinton’s 2016 levels, along with a credible level of Hispanic support.
But if white support drops back toward 2016, Biden will need high Black turnout share combined with high levels of both Latino support and turnout in order to get to a win.
Given the unique importance of the Black vote in Pennsylvania, we constructed two sets of scenarios:
We label the first set of scenarios “optimistic,” with the Black vote share set at 12.5% (the average of estimates from in-state partners).
We label the second set of scenarios “regression to 2016,” with the Black vote share set at 10% based on the 2016 exit polls in the state.
Comparing the “optimistic” scenarios to the “regression to 2016” scenarios allows us to see how the importance of the Hispanic vote shifts if Biden fails to improve upon Clinton’s performance with Black voters in Pennsylvania.
Above 44% white support, we project Biden would win the state. If recent polling, which has him as high as 50%, were to hold up, he’d be in a very good position.
Even with a drop in white support from today’s polling, down to 44%, Biden wins — if Black turnout share remains high.
…But if Black turnout share looks like 2016, then a Biden win depends on optimistic Latino turnout at current levels of support. A decrease in the same could spell trouble.
Even if Biden’s white support drops to a point just above previous Clinton (and Obama) levels, and Black turnout share regresses to its 2016 figure, Biden can still win at realistic, though optimistic, levels of Latino performance.
If Biden’s white support regresses to 2012/2016 levels, then high Black turnout share will need to combine with decent Latino performance for him to win the state.
Although the Latino electorate in North Carolina is relatively small, it has the potential to be pivotal under certain conditions. In our simulations, those conditions are a function of two key factors: the level of white support and the Black vote share (assuming that Biden earns a high level of support from Black voters).
The following chart summarizes the results of our simulation across a range of white support and Black vote share.
- Pink squares indicate scenarios in which our calculations have Biden losing the state regardless of how well he does with Latinos.
- Blue squares indicate scenarios in which we show Biden winning the state even if he does poorly with Latinos.
- Green squares indicate scenarios in which the Latino vote is pivotal to determining the outcome of the election.
The zone marked by a green box shows scenarios that fall within the range of Biden’s white support in North Carolina according to polls taken between March and September (38% to 41%). Within this range, there are a number of credible scenarios in which a Biden victory will be contingent on the Latino vote.
As Biden slips toward the lower end of his current range of white support, the Latino vote becomes an important companion to the Black vote in getting past 50%. At current projections, Biden is directly in the zone of Latino pivotality.
The most recent poll of North Carolina has Biden with 38% of the white vote. At that level, the Latino vote will be pivotal to the outcome if Black vote share lands somewhere between Clinton’s 2016 performance (20%) and Obama’s 2008 high watermark (23%).
Even in this optimistic scenario for Black turnout and white support, Biden would only win at the higher end of the Latino turnout range, and would lose if his Latino support numbers were to slip.
In the US Senate race, Cal Cunningham (D) would lose at his current rate of Latino support (65% two-way), except at the highest level of turnout.
The Latino vote is critical to a Biden win in Nevada in nearly all scenarios we examined.
At his current level of Latino support, Biden is well positioned if he can hold on to the high level of white support he enjoyed at the end of 2019 and the beginning of 2020.
If white support drops down to what we saw for Obama in 2012, for example, he couldn’t afford any slippage in Latino backing.
At Clinton’s 2016 white levels, Biden still has a clear path to victory, but it will require him to improve on his current support among Latino voters and maximize turnout.
Above 43% white support, we expect Biden wins in all our likeliest scenarios.
Even if white support were to drop to 2012 levels (43%), Biden would be in position to win the state if the election were today. But any slippage in Latino support, paired with a lower turnout scenario, would flip the result.
If white support were to match Clinton’s 2016 result, Biden would need to increase his margins with Latino voters and rely on higher Latino turnout share to repeat her (narrow) win in the state.
If Biden can do as well with White voters as he was polling during the summer, our simulations show him winning the state regardless of his performance with the Latino vote.
However, Florida will find a way to be Florida. If Biden ends up in the range of Nelson/Gillum, getting above 60% with Latinos and pushing up voter turnout are both essential for victory.
In order to get to 60% overall Latino support, Biden will need to maximize his support among non-Cuban voters, even if he earns a high level of support from Cubans.
Above 41% white support, Biden wins in all our likeliest scenarios (he was polling as high as 44% in the summer).
At 41%, which is a slight improvement over average white support in Florida, Biden would win the state today. But a repeat of the Nelson/Gillum under-performance with Hispanics in 2018 would result in a narrow loss.
At current Hispanic support, and with the level of white backing earned by Obama when he won Florida in 2012 (40%), Biden narrowly would narrowly lose the state today. But he wins with even modest increases in Latino — or Black — performance.
With every point drop in white support, the math gets more difficult in Florida. Here (39%), Biden would need to at least match Clinton’s 2016 performance with Hispanics, or bank on more generous projections of Hispanic turnout share, to win.
At a historic average of white support (38%), Biden loses the state, except in cases of max Hispanic support and turnout share.
Below 38% white support, we project Biden loses in all scenarios. For comparison: Clinton was at 33% in 2016, per exit polling.
As evident above, Biden likely needs to cross the 60% threshold of support with the Hispanic vote in order to win Florida.
Getting to the upper bounds of that range depends on calibrating his vote share among the dynamic mix of sub-groups who make up the Latino vote in the state: getting very high support with the non-Cuban vote (>70%), and maximizing the Cuban vote (low-to-mid 40s).