Socially Liberal, Fiscally Conservative Voters Preferred Trump In 2016

The conventional wisdom is that a socially liberal, fiscally conservative independent presidential candidate — like, say, Howard Schultz, the former Starbucks CEO, who is considering just such a run — poses a threat to Democrats in 2020 by potentially dividing the anti-Trump vote. But there’s another possibility: that Schultz, or a candidate like him, could divide the pro-Trump vote instead.

I’m not going to purport to provide a comprehensive analysis of whether a Schultz-like candidate is more likely to help or hurt President Trump’s re-election chances, which is a question that can be approached from many angles. The answer will depend on what sort of candidate the Democrats nominate and what Trump’s political standing looks like late next year.

What we can do, however, is look back to 2016, when voters who are socially liberal but fiscally conservative also had to choose a presidential candidate.

Who did these voters, who are somewhere around 15 percent of the electorate, go for in 2016? The answer is complicated because it depends on which social and economic issues you look at: racial attitudes were more determinative of their presidential vote than views on gay marriage, for example. But according to the Cooperative Congressional Election Study, a comprehensive survey of more than 60,000 voters organized by Harvard University and conducted by YouGov, these voters were slightly more likely to vote for Trump than Hillary Clinton.

My approach in this story is fairly straightforward: I chose five questions on social issues and five questions on economic issues from the CCES, with the idea of pairing them up two at a time to see how voters who held both positions voted. For instance, how did voters who wanted to grant legal status to large numbers of undocumented immigrants (a socially liberal policy), but who also wanted to repeal the Affordable Care Act (a fiscally conservative policy), vote in 2016? And which of these issues had more influence on their vote?

Rather than predict exactly what Schultz’s platform would be — so far, he’s been short on detail, and it’s not clear that his candidacy is going anywhere anyway — let’s instead undertake a more general analysis of how socially-liberal-but-fiscally-conservative voters behaved in 2016. (If you do want more detail on Schultz’s stance on each issue, please click on the footnote after each item.) First, here are the five fiscally conservative positions:

• Prioritizing the budget deficit: These are voters in the CCES who said the budget deficit is of “very high importance” to them.¹
• Favoring cuts to entitlement programs: Given a three-way choice, these voters said they’d prefer to cut “domestic spending (such as Medicare and Social Security)” rather than cut defense spending or raise taxes to balance the budget.²
• Opposing a minimum wage hike: These voters said they were against raising the minimum wage to $12.³
• Favoring Obamacare repeal: These voters favored repealing the Affordable Care Act.⁴
• Opposing environmental regulation: These voters said they opposed strengthening enforcement of the Clean Air Act and Clean Water Act “if it costs U.S. jobs.”⁵

And these are the socially liberal stances I tracked:

• Favoring gay marriage: These voters said they favored “allowing gays and lesbians to marry legally.”⁶
• Favoring amnesty for immigrants who entered the country illegally: These voters favored granting “legal status to all illegal immigrants who have held jobs and paid taxes for at least 3 years, and not been convicted of any felony crimes.”⁷
• Favoring abortion rights: These voters favored “always [allowing] a woman to obtain an abortion as a matter of choice.”⁸
• Favoring gun control: These voters supported a ban on assault rifles.⁹
• Thinking that white people benefit from their race: These voters strongly or somewhat agreed that “white people in the U.S. have certain advantages because of the color of their skin.”¹⁰

Next, here’s how voters¹¹ who took each of these issue positions voted in 2016, according to the CCES, along with the share of the electorate that holds each position. There aren’t going to be any huge surprises here: Voters who took socially liberal positions tended to vote for Clinton, and those who held fiscally conservative ones mostly went for Trump.

Like I said, no huge surprises — although there are a couple of things worth your attention. One is that more people than you might think voted for Trump despite holding at least some socially liberal positions: Clinton won voters who favor abortion rights by “only” a 67-27 margin, for instance, and won those who back amnesty for undocumented immigrants only 67-26. Conversely, slightly fewer voters with fiscally conservative views were willing to defect from Trump: He won voters opposed to the $12 minimum wage 77-15, for instance. The big exception to this pattern was the question on race, which was more determinative of voting than the other social issues; voters who thought white people have advantages because of their race went for Clinton by an overwhelming 76-17 margin.

But what happens when we start pairing these positions together? Between the five socially liberal positions and the five fiscally conservative ones, there are 25 possible combinations of positions, taken two at a time. And in the rather long table below, I’ve shown how voters who held each of those combinations voted in 2016:

For most of the combinations, Trump won more of these voters than Clinton, but before we really get into the conclusion about which party’s voters Schultz or someone like him might poach, there are a couple hints in this data about the size of a potential market for a Schultz-like campaign. One is that, depending on which combination of issues you choose, an average of about 16 percent of the electorate holds both fiscally conservative and socially liberal positions. That’s not a terribly large constituency. It’s smaller than what you’d get if there was no correlation between social and fiscal conservatism and voters were equally distributed between four quadrants.¹² The CCES data produces is a slightly larger estimate for the size of this population than other researchers have found, but it still isn’t anywhere near a winning coalition unto itself.

We do see, however, that it is a natural group for independent or third-party candidates to pursue. On average between the various issue combinations, 8 percent of socially-liberal-but-fiscally-conservative voters went for candidates other than Clinton and Trump in 2016, a bit larger than the overall third-party vote in 2016, which was around 6 percent.

But the headline is that, when choosing between the major-party candidates, these voters were more likely to go for Trump than Clinton. Among the 25 combinations of socially liberal and fiscally conservative views, Trump won the most votes 19 times, Clinton did so five times, and there was one draw. And on average between the 25 combinations, Trump won 52 percent of the vote to Clinton’s 40 percent. That’s not a huge margin: a 12-point edge among 16 percent of the electorate. But it adds up to enough voters that, if all of them had gone for a third party instead, Clinton would have won Michigan, Pennsylvania, Wisconsin and Florida, and therefore the Electoral College.

The big exception to the pro-Trump trend was the question on race, which tended to outweigh every other issue. For instance, among all voters who wanted to cut entitlement programs to balance the budget, Trump won 70-24. But among voters who want to cut entitlements and who think white people have an advantage because of their skin color, Clinton won 55-37.

An alternative approach is to look at which set of issues — fiscal views or social views — had influence on people’s votes by performing a series of regression analyses. Without getting too technical, this is probably more robust than the method I used above because it’s less dependant on the overall popularity of a particular policy view and instead reveals more about how two views compare to one another. With that said, it yields fairly similar answers, so feel free to skip ahead to the (brief) conclusion at the end.

Specifically, I ran a series of probit regressions to estimate the probability that a voter chose Clinton or Trump¹³ for each of the 25 combinations of policy positions. For instance, one of the regression models estimated the probability that a voter selected Clinton or Trump based on whether or not they support gay marriage and whether or not they wanted to repeal the ACA. I also ran a series of regressions that again tested the policy positions but controlled for whether or not the voter was white and whether or not he or she had a college degree.¹⁴ The results are detailed in the table below.

Don’t worry much about the technical interpretation of the numbers (coefficients) in the table; it’s their relative value that counts. In fact, they produce some fairly clear conclusions about which issues tended to influence how people voted:

• The question on racial attitudes, as already mentioned, was an especially strong predictor of voting behavior. Its influence diminishes slightly if you control for the voter’s race, but it still beats all the fiscal issues, except for …
• Obamacare. Views on repealing the ACA had the largest influence of all, beating even the racial question. Health care was a huge issue in 2016, as it was again in 2018 and as it will probably be in 2020.
• Prioritizing the deficit, which is one of Schultz’s central points of emphasis so far, was one of the least influential factors.
• Perhaps more surprisingly, views on immigration were not that important either, as the question on amnesty was the least influential of the five social issues.

On average, this method produces a more equivocal result than the one I reported earlier. Which social issues and which fiscal issues you choose matters a lot, with race (social) and health care (fiscal) being especially important. Still, the fiscal issues were slightly more influential, on average. Among the 25 pairings testing a fiscal issue against a social issue, the fiscal issue “won” (was more influential) 15 times and the social issue won 10 times in the basic version of the model. In the version that controls for race and educational status, the fiscal issue won 14 times against 11 times for the social issue.

But again, my goal here is not necessarily to convince you that a Schultz candidacy would certainly hurt Trump. Rather, it’s to discourage you from giving too much credit to the conventional wisdom, which asserts without much evidence that the opposite is true. Voters who are fiscally conservative but socially liberal — about 15 percent of the electorate — are often swing voters, rather than definitively being in one party’s coalition. In 2016, these voters swung slightly toward Trump, at least according to the CCES, which was perhaps enough to give him his Electoral College majority given that his margins in several decisive states were extremely narrow. Maybe they swung toward Democrats in last year’s midterms — the 2018 version of the CCES isn’t available yet — and maybe they wouldn’t be inclined to vote for Trump in 2020. But the effect of a Schultz-like candidacy is less predictable than what the pundits are telling you.

From ABC News:

Polling ‘shows there’s not a huge market for Howard Schultz’: 538’s Nate Silver