Stop me when this sounds familiar. The times are uncertain, with a long-running war abroad and a sense of rising division at home. There’s violence at political events and in our streets and talk on the right of a “silent majority.” The Democratic Party is internally divided between an older, more establishment faction and a left-leaning insurgency fueled by young activists. For the insurgents, the Democratic Party’s process for nominating presidential candidates seems rigged and undemocratic.
That, in a nutshell, was 1968, when Vice President Hubert Humphrey’s support among Democratic Party insiders assured him the nomination while anti-war candidates Robert Kennedy and Eugene McCarthy rallied young activists as they competed in primaries. But given the extended 2016 primary contest between Hillary Clinton and Bernie Sanders — and given much of the media narrative that surrounds it — that account of a divided Democratic Party resonates even today. In 1968, the Democratic establishment won a temporary victory by nominating Humphrey, but the insurgents won the real prize: control over the party moving forward. They rewrote the rules for how the Democratic Party nominates presidential candidates, and four years later, their votes made the liberal, anti-war George McGovern the nominee. The New Deal coalition in presidential politics was in decline, as civil rights and anti-war activists took the helm of the party from organized labor and Southerners.
Does Sanders’s surprising strength in the Democratic primaries foreshadow another move leftward for the Democratic Party? Over at Vox, Matthew Yglesias thinks the answer is “yes.” As he argues, “What’s clear is that there’s robust demand among Democrats — especially the next generation of Democrats — to remake the party along more ideological, more social democratic lines, and party leaders are going to have to answer that demand or get steamrolled.” An important part of Yglesias’s argument is demographic. With young voters strongly in Sanders’s camp, the demand for Sanders-style politics is likely to grow.
Still, others contend that it is a mistake to conflate support for Sanders in one particular primary contest with enduring support for his brand of liberalism. Political scientists Christopher Achen and Larry Bartels instead see Sanders’s popularity as lying primarily in group loyalties and opposition to Clinton. In their words, “it is a mistake to assume that voters who support Mr. Sanders because he is not Mrs. Clinton necessarily favor his left-leaning policy views.”
Voting is not like ordering at a fast-food restaurant. Instead, you get only a few choices and can’t pick the precise combination of things that you want. Pundits may see Sanders’s policy positions as key to his appeal, but do Democratic voters see things the same way? As we’ll see below, some surely do. But there is another sizable group of Sanders backers whose support comes primarily from the fact that Sanders’s opponent is Hillary Clinton. If we assume that all Sanders voters lean left, we risk exaggerating the leftward march of the Democratic Party.
A durable liberal coalition or a flash in the pan?
Certainly, Sanders’s supporters lean left, on average. Using a survey that interviewed the same people in 2007, 2008, 2012 and early 2016, I found that Clinton backers rated themselves as more conservative, less supportive of gay marriage and less supportive of raising taxes on the rich in earlier surveys than Sanders supporters. But policy preferences are just one reason why people might support a candidate, so it’s hard to know how much of a factor policy is in Sanders’s coalition and therefore how lasting its effects on the Democratic Party will be. Imagine that Sanders was facing off against a more centrist Democrat not named Hillary Clinton. Just how much of his political support would he maintain?
One way to answer that question is to compare the geography of Sanders support to the support for other left-leaning candidates in prior Democratic primaries. If Sanders’s support really represents an enduring coalition, other candidates who have adopted similarly liberal stances should draw their support from similar places. In the chart below, I test whether that is the case. For instance, the top-left panel shows the correlation between Sanders’s performance in Pennsylvania in 2016 and Joe Sestak’s performance in the 2010 Democratic primary for Senate. In that race, Sestak faced Arlen Specter, an incumbent senator who had recently switched parties. Many leaders within the Democratic Party had hoped Sestak would stay out of the 2010 race, giving support for him an anti-establishment edge similar to what we saw with Sanders this winter and spring. As the figure illustrates, the relationship between Sestak support and Sanders support is sizable. One way to measure the strength of an association is to look at the expected change in one measure as another measure shifts. Hypothetically, if a county goes from casting all of its ballots for Specter to casting all of its ballots for Sestak, Sanders should expect a 34 percentage point jump in support. That is a strong association. In Pennsylvania, at least, some counties will reliably back a left-leaning candidate in Democratic primaries.
But it’s not just Pennsylvania. In 2014, Andrew Cuomo sought re-election as New York’s governor and faced a primary challenge from left-leaning academic Zephyr Teachout. As the top-right panel makes clear, counties that gave more support to Teachout also gave more support to Sanders, with roughly the same correlation as we saw in Pennsylvania: A swing from 0 to 100 percent of the vote for Teachout corresponds to a pro-Sanders swing of 29 percentage points. And this pattern isn’t confined to the Northeast. As the bottom-left panel shows, when Bill Halter challenged sitting Sen. Blanche Lincoln from the left in Arkansas’s 2010 Democratic primary, the counties that backed him were also the places that would give more support to Sanders six years later.
That said, not all contentious Democratic primaries are forerunners to the Clinton-Sanders race. Consider the bottom-right panel of the chart, which compares Sanders’s performance in Connecticut towns in 2016 and Ned Lamont’s performance while taking on Sen. Joe Lieberman 10 years earlier. Here, the correlation is quite modest: A swing from 0 percent for Lamont to 100 percent for Lamont is associated with a pro-Sanders uptick of just 5 percentage points. Lamont is a wealthy businessman who ran in part on his opposition to the Iraq War, so it isn’t surprising that his geographic coalition looks quite different than that of Bernie Sanders. But the example is instructive. What it means to be the liberal candidate can change with the times.
Still, with the exception of Connecticut, Sanders’s support seems to be drawn from the same places that have backed other liberal challengers in the recent past. There is, in short, a durable faction that consistently opts for left-leaning candidates when they are on the ballot.
The two types of Sanders supporters
Still, to say that there is a set of places — and presumably, of voters — who back liberal candidates with some consistency is not to say that that group is anything approaching a majority of the Democratic Party. We also need to take Achen and Bartels’s point seriously by considering just how much of Sanders’s support comes from people who are not so much pro-Sanders as anti-Clinton.
In the early 2016 wave of the Institute for the Study of Citizenship and Politics panel that I conducted with my University of Pennsylvania colleague Diana Mutz and others, we asked respondents who they support. We then asked a follow-up question about why. In their answers, Sanders supporters were more than three times as likely as Clinton supporters to name a policy position, with 19 percent of his backers naming policies such as wealth redistribution and health care. That is certainly in keeping with the idea that Sanders’s left-leaning positions are a reason some people were feeling the Bern.
However, most Sanders backers gave different reasons for their support, reasons that do not suggest that it is grounded in an enduring liberalism. An identical 19 percent of Sanders backers explained their support with reference to another person, overwhelmingly mentioning Hillary Clinton. These Sanders supporters made comments such as “he’s not Hillary” and “[I] don’t like other 2 candidates.” These differences in rationales are consequential. Among the Sanders voters who referenced a policy position in explaining their support for him, the average ideology on a 7-point scale was 2.85, somewhere between “liberal” (a 2) and “somewhat liberal” (a 3). But among the Sanders voters who instead named another candidate in explaining their vote, the mean ideology was 3.49, putting them almost as close to “moderate” as to “somewhat liberal.” (By way of comparison, the average Clinton supporter scored a 3.23, and a 7 represents someone who is “extremely conservative.”)
According to our January 2016 survey, a sizable share of Sanders voters — at least 1 in 5 — were voting against Clinton, and those are more moderate voters who shouldn’t be tallied among those wanting the Democratic Party to move left.
West Virginia provides a case study in this kind of oppositional voting. In 2008, when the choice in the Democratic presidential primary was between Clinton and then-Sen. Barack Obama, the state overwhelmingly backed Clinton. But in 2016, with Clinton as the face of the Democratic establishment, the state backed Sanders by more than 15 percentage points. It is an obvious mistake to view Clinton’s 2008 win in West Virginia as any enduring support for her particular set of policy positions: When she was no longer running against Obama, her support crumbled. These results suggest that there are many West Virginia voters whose backing of Clinton in 2008 and of Sanders in 2016 was a statement of opposition to the national Democratic Party’s direction and its front-runner.
Another way to see the extent to which Sanders became a vehicle for anti-Clinton voting is to look at West Virginia’s 2012 Democratic presidential primary. Yes, you might not remember the 2012 West Virginia Democratic primary. Obama was an incumbent president without a serious primary challenger, so you probably didn’t stay up late to watch the returns come in. But in West Virginia, Obama did face a challenge, from Keith Judd, a felon serving time in a federal prison in Texas. In fact, Judd managed to win just over 40 percent of the vote. And as the figure below shows, the counties that leaned toward Judd in 2012 also tended to lean toward Sanders in 2016. As we go from a hypothetical all-Obama county to an all-Judd county, we should expect Sanders’s vote share to tick up by 36 percentage points, a shift comparable to those we saw above.
Sanders surely picked up a set of left-leaning voters who consistently back liberal candidates when given the chance — and who will likely do so in the future. But paradoxically, Sanders’s backing among those left-leaning voters may well have given him the visibility, name recognition, polling numbers and fundraising to become the anti-Clinton Democrat. That might explain the unusual geography of his political support, as he managed to win liberal bastions such as Washtenaw County in Michigan (home of Ann Arbor) while also winning more conservative, rural counties such as Ottawa County (west of Grand Rapids). The Sanders coalition proved unexpectedly strong in the 2016 Democratic primaries. But just as Cuomo soundly beat Teachout, the liberal bloc alone was not enough to propel Sanders to victory — and it wasn’t Sanders’s sole source of support, either. The lesson of Sanders’s performance may be less about the growing appetite for left-leaning policies than about the unlikely bedfellows that underpin many presidential campaigns.
Owen O’Hare, Ashwin Ramesh, Gabrielle Rothschild and Elena Zhou provided research assistance.
CORRECTION (July 11, 1:40 p.m.): An earlier version of a chart in this article misstated the year of Ned Lamont’s primary against Joe Lieberman. It was 2006, not 2010.