Some insights from Equis Research data & analysis

This is the second in a series of posts where the Equis Research team is digging into its data to provide context on the Latinx vote in this year’s presidential election. Equis Research works with a team of polling firms — GBAO, Global Strategy Group, LatinoDecisions, Myers Research, and TargetSmart — to survey Latinx attitudes. We have interviewed 28,227 Latino voters in 11 states since July 2019.

In our first post, we looked for lessons from the 2020 primary. You can read it here.

In this post, we’ll look forward at Latino turnout — what it looks like to increase vote share; the unique importance of Latino non-voters; and how reaching new heights of voter participation in this community requires moving beyond a traditional mobilization framework.

What are likely turnout scenarios in 2020?

Elections are won on the margins. So it is helpful to start by projecting how many Latino voters are expected to show up in 2020, and beginning to consider how many marginal votes can be added beyond that figure.

In order to build an initial estimate of 2020 Latino vote share in our target states, we worked with TargetSmart and a combination of the TargetSmart presidential turnout and race models. First, TargetSmart created an initial turnout estimate for all voters in a given state, based on past trends and other (pre-COVID19) assumptions about this cycle, then rank-ordered the entire voter file on turnout score, from top to bottom.

Because race data in most states can be spotty, estimates of the Latinx population were largely based on modeling. There are lots of good uses for this modeling — but it’s definitionally imperfect. These imperfections are exacerbated when looking at Latinx voters, when there is such a diverse number of ways a voter could self-identify into this group. As such, most numbers — including ours! — likely undercount the number of Latinx voters, both those who have voted in the past and those who may vote for the first time in 2020.

Still, we choose to go with conservative estimates, knowing that regardless of the number, the goal of campaigns and organizations is to try and increase this vote share.

The chart below lists (1) our current projections for Latinx vote share in 10 battleground states, and (2) compares them to our own (conservative) estimate of 2016 rates.

And (3) for illustrative purposes, the chart also shows how many additional Latinx voters would need to turn out, beyond our current projections, to increase statewide vote share by half a point. (Real turnout goals in some states, like Arizona, will be greater, whereas in other states, like Wisconsin, they might be smaller.)

What states should be the focus of Latino program?

When talking about the Latinx vote, much of the attention is duly paid to the states with the largest share of the population — your Floridas, Nevadas, and Arizonas. The turnout chart above shows where the Latino vote plays the biggest role in statewide results.

But in this election cycle, in particular, campaigns and organizations would be wise to broaden their scope. When a state like Michigan is decided by a margin of 10,704 votes, as it was in 2016, you can point to a variety of factors that could have changed the outcome — and analysts have. But one that doesn’t get discussed enough: that slim margin is equivalent to 8% of the Latinos who were eligible to vote but sat it out (or 5.3% of all eligible Latinos). An additional point or two of Latino turnout, as you’d expect from a GOTV treatment, isn’t huge in an aggregate sense but contributes greatly in a narrow election when both sides are scrambling to find new pools of votes.

Of course, increases in turnout won’t happen without investment. To the casual observer, 6,000 extra votes doesn’t seem like a heavy lift. But when you factor estimates of the cost of GOTV tactics — let’s say, for these purposes, four votes per $1,000 spent — then turning out 6,000 votes could cost $1.5 million and 20,000 votes could cost $5 million.

These numbers, weighed with other considerations, suggest some unexpected opportunities. In Wisconsin (2016 margin: 22,748), for example, the density and youth of the Latino electorate, combined with the highest anti-Trump numbers in the states where we poll, could make investments in Latino engagement an efficient use of progressive resources.

Why Latino non-voters are so important

Now that we know what expansion of turnout could look like, we turn to the question of where those votes come from.

At our current projections, 57% of eligible Latinx voters in battleground states will sit out the 2020 election.

‍(Again, this is based on past vote behavior, not current attitudes.)

Beyond the obvious fact that this is a LOT of voters, there is another reason these non-voters matter so much in the context of 2020. As pointed out in various analyses, there is much reason to believe that likely non-voters, as a whole, are more favorable of Trump than likely voters.

A rare exception is Latinos.

‍This is borne out in our data, where vote history is available: the significant chunk of Latinos who skipped both the 2016 and 2018 elections tend to give Trump worse approval numbers than those Latinos who voted in both of those elections.

For Democrats, turning out more Latinos is an efficient way to create a more anti-Trump electorate. And this year, they’ll need it. Trump has shown a unique ability to bring out infrequent voters, largely non-college white voters. If Democrats are to counteract such a surge, they will need to dig deep into their own reserves of available votes — and the data shows Latinos are a good place to start.

Some part of the opportunity is in registration. But registration alone doesn’t do the job.

So, again, where will those votes come from?

The orange band in the chart below shows the size of the unregistered (but eligible) Latinx population across these 2020 states, based on a combination of census estimates and voter file data. That is a great deal of untapped potential.

But the green band should give us pause: it shows the large number of Latinos who have already gone through the trouble of registering to vote but aren’t expected to turn out this year.

This suggests the impact of registration efforts could be undercut when there isn’t follow-up engagement to turn out new registrants.

The ability of campaigns and organizations to register voters in a post-COVID world is also in doubt. Site-based registration, so critical for registering Latino voters, isn’t a viable option for the foreseeable future.

That is yet another reason campaigns shouldn’t sleep on the green band; the voters it represents are a reasonable focal point for universe expansion in this environment.

Case study: Arizona

Let’s look at how these trends apply to Arizona, as an example.

By our intentionally conservative estimates, some 7-in-10 eligible Latinos in Arizona are expected to skip the polls this year.

The opportunities for expansion are vast: while Latinos make up 13% of all likely voters, they represent a staggering 41% of all likely non-voters in Arizona.

To quantify the impact: an addition of 30,000 Latino voters beyond current projections, or nearly 2.5% of eligible Latino voters, at our estimated support levels would get Joe Biden 3 points closer to his win number in the state. The election could be won or lost on these margins. The stakes of moving Latinos off the sidelines are great.

Moving beyond a mobilization framework

One of the ways we look at the Latinx electorate is in segments based on a combination of vote choice and vote intent. The below chart reflects the combined data from our December wave of polling (minus Florida, which merits a separate write-up), as analyzed by GBAO.

You’ll notice the “Mobilization” cohort — voters who are very likely to vote for a Democrat for president but less certain to vote — is relatively small. Naturally no Democratic campaign can win without maximizing the turnout and support of this segment.

But hitting vote goals, or exceeding them, requires success with another, less straightforward segment of the Latino electorate. The “Ambivalent” cohort are voters who aren’t solidly locked into their presidential vote choice, or else prefer a third party candidate.

Most of the voters in this bucket are there because they express a preference for a Democrat against Trump…but aren’t prepared to say they’ll definitely vote for them. This is before Biden became the presumptive nominee, and could shift somewhat as we begin to poll with a named candidate. The soft support remains noteworthy, however, especially among the youngest cohorts and among women.

Most political science will tell you to disregard this expression of reluctance — if those voters vote, they’ll vote the way they say they’re only leaning. Here’s the thing: campaigns — and Democrats, in particular — should be concerned not so much that an ambivalent voter will defect to the other side but that they will defect to the couch.

Our analysis of the 2016 ANES data shows that Latinos who were ambivalent broke for Clinton over Trump, but nearly 1 in 3 ended up not voting. That’s nearly double the rate of non-participation as among ambivalent white voters, and slightly higher than among ambivalent black voters.

Ambivalent voters need persuasion, not just to influence their candidate choice, but to influence their decision on whether to vote at all.

The good news for campaigns: persuasion for purposes of turnout looks a lot like any other kind of persuasion: it starts at the launch of the campaign, with a focus on both issues and values that matter to the target voter, with messengers that matter, across the channels ( TV, digital) they frequent. This is what a campaign might normally do for a soccer mom. Perhaps for these purposes it’s useful to remember that Latinas are, in fact, the original soccer moms.

That’s all for now

In future installments, we can talk more about who these ambivalent voters are.

You can also expect to see more from us on the most important topic we haven’t yet covered: gender.

And for those asking, WHAT ABOUT FLORIDA?, stay tuned…

‍