The battle for the Democratic nomination is almost over, but the battle to define the meaning of the Bernie Sanders campaign has just begun.

Does the Vermont socialist’s improbable success — fueled by historic levels of support from younger voters — herald the rise of a new left-wing bloc in American politics? Or is the Sanders phenomenon closer to a passing fad, little more than a protest vote against Hillary Clinton?

Writing in the New York Times, political scientists Christopher Achen and Larry Bartels recently joined the debate.

Based on an analysis of exit polls and survey data — and drawing from their new book, Democracy for Realists — they argue that the difference between Sanders and Clinton voters lies not in ideology or policy views, but “social identities, symbolic commitments, and partisan loyalties.”

Sanders voters, in short, are not “the vanguard of a new, social democratic–trending Democratic Party,” but chiefly a group of “disaffected white men” and young people drawn to “campaign labels, not policy preferences.”

Of course, this is music to many ears. Nothing would please corporate Democrats and establishment pundits more than if the largest left-wing insurgency in American primary history could be exposed, once and for all, as an accidental combination of confused children and angry white dudes.

An ecstatic Paul Krugman, crowing that Achen and Bartels had laid bare the true essence of the movement, was even inspired to formulate a set of Myers Briggs–style personality types for Sanders voters, in which every shade of fool, narcissist, lunatic, and failure was duly represented.

A closer look at the survey evidence, however, casts some doubt on Achen and Bartels’s conclusions.

The first prong of their argument is a familiar one: according to exit polls, self-described “liberals” have not come out overwhelmingly for Sanders. For Achen and Bartels, this counts as evidence that Sanders voters do not truly lean to the left.

Yet this election season has not been kind to the predictive power of the liberal-moderate-conservative typology beloved by pollsters.

Last summer, Nate Silver developed a model of the Democratic primary based on the (supposedly stable) number of liberals in each state’s electorate. His formula spat out a grand total of two victories for Sanders.

Of course, as it happened, the share of “liberal” voters skyrocketed everywhere — in Indiana, for instance, it leapt from 39 percent in 2008 to 68 percent — and Sanders has already won over twenty contests.

Does this mean Indiana’s Democratic electorate surged to the left over the last eight years? Probably not. A better explanation is that voters’ ideological identities — at least as expressed through simplistic survey labels — are not fixed, but heavily dependent on context.

The context in this year’s Democratic primary is a race between one of America’s most famous liberals, Hillary Clinton, and a rival significantly to her left.

For the most part, that leftward dynamic has encouraged Democrats to adopt the “liberal” label in record numbers: among the limited choices offered to voters, it is theoretically the most left-wing.

But in other ways, the Clinton-Sanders clash has muddied the ideological waters in ways the traditional categories do not reflect.

Bernie Sanders, an independent and an avowed socialist throughout his adult life, is not one to adopt the “liberal” designation — and the same goes for many of his supporters.

After the West Virginia primary, pundits gleefully noted that Sanders ran far ahead of Clinton among Democrats who favored policies “less liberal” than Barack Obama’s.

But what do Obama-style “liberal policies” actually mean to West Virginia voters — free college tuition and a $15 minimum wage? Or bank bailouts, soda taxes, and global free-trade agreements?

Sanders, after all, won “less liberal” voters not only in conservative West Virginia, but in Wisconsin and Massachusetts, too.

Over the last thirty years, as the Democratic Party has increasingly aligned itself with the professional class, it’s not hard to imagine that for many struggling workers, “liberal” has come to mean little more than a synonym for “Harvard Law School graduate.”

If Sanders voters proudly back a program that involves taking on the “billionaire class,” enacting universal health care and child care, and launching an ambitious jobs program, we need not be concerned about whether they call themselves “liberals”: they are already social democrats.

Here, however, Achen and Bartels interject a second and more troubling finding. Sanders supporters, they report, are actually less enthusiastic than Clinton voters about left-leaning economic policies: a higher minimum wage, for instance, and increased spending on health care.

According to the two political scientists, even the young Democrats so central to Sanders’s rise are less likely to support these redistributive measures.

This finding contradicts many other studies. Pew’s survey of primary voters earlier this year placed Sanders supporters further to the left than Clinton supporters on some economic questions, including whether corporations make too much in profits, and found that they held comparable positions on other issues, including health care and Social Security.

When Harvard’s Institute of Politics conducted an extensive poll of young people this spring, they found overwhelming evidence that voters under thirty are much more progressive than their elders.

So what accounts for the discrepancy? Achen and Bartels’s numbers come from a January 2016 survey conducted for the American National Elections Studies. It contains some of the most detailed information we have on the political views of different primary candidates’ supporters.

But as the political scientists Christopher Hare and Robert Lupton have pointed out, there’s one major problem with the ANES survey: it asked respondents to choose a Democratic candidate “regardless of whether you will vote in the Democratic primary this year.”

As a result of the open-ended wording, a large proportion of Republicans and Republican-leaning independents appear to weigh in: about 40 percent of them selected a favorite Democratic candidate.

Many of these Republicans and Republican leaners (very similar groups of people) chose Bernie Sanders. Some of them, no doubt, were sincere. After all, Sanders has consistently eclipsed Hillary Clinton in general election polls.

But it’s doubtful that more than a handful of these Republican respondents — like the 50 percent of Democrats who picked a favorite GOP candidate — expressed anything like a meaningful political commitment.

If we want to truly understand where Sanders voters stand within the broader Democratic electorate, it makes little sense to use a survey sample that is fully one quarter Republican.

Interestingly, when we remove these GOP respondents from the pool, the sharpest differences between Sanders and Clinton supporters occur not on economic policy but on questions involving gender and race.

And for all the online chatter about sexist “Bernie Bros,” the ANES data offer little evidence that Sanders voters embrace him out of a desire to buttress their male identity.

Sanders backers, for instance, were more likely to strongly endorse requiring employers to pay men and women equally for the same work. They were also much more assertive in their support for mandatory paid parental leave:

Nor do the ANES data furnish much evidence that Sanders voters have been motivated by white racial resentment. Among Democrats and non-Republican-leaning independents, in fact, white Clinton supporters were more inclined than white Sanders supporters to say that blacks are “lazy” or “violent,” and that black people should work their way up “without special favors.”

Based on the ANES results, Achen and Bartels describe Sanders backers as less hardy in their support for “concrete” progressive economic policies than Clinton backers. But omitting Republicans from the sample neutralizes that judgment, as Hare and Lupton demonstrate.

Achen and Bartels may be right to suggest that there are no major ideological gaps between Clinton and Sanders voters. It is Clinton, after all, who has won the lion’s share of support from black voters (especially older black voters), who are generally more left-wing than their white counterparts on economic issues.

The critical fact in politics, however, is not how voters organically position themselves on a static spectrum of ideology or policy: it is where they line up in the context of a dynamic political contest.

And this is where Achen and Bartels’s argument really falls short.

After attaching Sanders supporters to the “folk theory” of democracy — the civics 101 idea that “citizens can control their government from the voting booth” — the pair goes about demolishing its precepts. By the end of the essay, the reader is left wondering whether it isn’t just the promise of Sanders’s political revolution that’s empty but popular democracy as such.

The mistake, however, is to believe that the folk theory represents a realistic (or even desirable) ideal against which we should judge American democracy.

Voters’ political preferences and identities aren’t molded willy-nilly by political elites. But neither are they immutable doctrines that spring solely from the minds of voters. They’re the result of a complex mix of influences and experiences — personal, political, material, cultural, and so on. The folk theory washes all this away, reducing democracy to a simple transmission belt where voters funnel preferences to elites, who smoothly convert them into policy.

A grade school conception of popular government, it doesn’t recognize the interplay between institutions and the rank-and-file, between leaders and ordinary citizens.

Sanders is fond of saying that his campaign “isn’t about me.” That isn’t quite true: while leaders don’t singlehandedly create social conditions, they can mobilize people to help steer the course of history.

But acknowledging that “the people” aren’t the only ones with their hands on the wheel doesn’t spell the death of democracy — or necessitate a retreat to a place where, as Achen and Bartels end up in their book, democracy is only preferable because competitive elections minimize corruption and allow for an easy transfer of power.

What’s especially curious is the pair’s almost obsessive focus on the foibles of voters, rather than the myriad institutions and forces that frustrate popular democracy in the United States.

In Democracy for Realists, Achen and Bartels devote at most a few sentences to the political power of the well-heeled (this despite the work of Bartels himself on the political implications of inequality). There’s little mention of the ways in which the American system, with its innumerable checks and balances and veto points, blurs the lines of decision-making and makes holding politicians accountable a tall task. There’s no appreciation of the fact that powerful market actors set the contours of economic policy-making.

Popular sovereignty, the two seem to say, faces its gravest threat from inept citizens — not from Sheldon Adelson’s wallet or Madisonian institutions or the cold judgment of the bond market.

Sanders has advanced a different understanding. In his packed stadium rallies, democracy is not the atomized affair of the folk theory, in which the ideal citizen impartially weighs the positions of the various candidates and then somberly enters the polling station to cast his ballot. It is instead an activity of passion and agitation and, yes, even struggle against a “billionaire class.”

The last point has been crucial, in two respects. First, even with the reduced pitch of an electoral campaign, Sanders has reminded observers that collective action and organized people remain the driving force behind social progress. The rough and tumble quality of strikes and protests and mass rallies don’t debase democracy — they are among its highest expressions.

Second, through constant excoriation, Sanders has formulated what Achen and Bartels would call an identity around which his supporters cohere: they are not the billionaire class. They stand against it.

Of course, when coupled with the social-democratic remedies Sanders pushes, this is just old-fashioned class politics — the idiom of any viable left project.

But that’s the point.

As Jerome Karabel has recently argued, the most important question that the 2016 Democratic primary asked is whether a candidate running on a bold left-wing platform could win mass support. The answer to that question has been an emphatic “yes.”

43 percent of voters — and 70 percent of young voters — opted for the social-democratic candidate, even though he was facing an opponent with greater institutional backing than any non-incumbent in history. (If more independents were allowed to vote, the result would have been even closer.)

Achen and Bartels are surely right that social identities and symbolic allegiances animate voter behavior more than pure ideology. But although a large portion of Sanders’s electoral backing has come from white men, the ANES data do not suggest they are particularly attached to their racial identities.

Only 22 percent of white Sanders supporters indicated that “being white” was “extremely” or “very important” to them (compared to 43 percent of white Clinton supporters).

By contrast, 74 percent of Sanders supporters (compared to 56 percent of Clinton supporters) reported that “the difference in incomes between rich people and poor people” has grown “much larger” in the last twenty years. Sanders supporters placed income inequality among their most important political issues twice as often as Clinton supporters.

Achen and Bartels file these results under the heading of economic pessimism and dismiss their significance because Sanders supporters did not respond with sufficient gusto to a bland ANES question about “government services.”

But if abstract policy preferences aren’t so important after all, perhaps we should take another look at those inequality numbers. What if they actually show the growth of a deeper allegiance — a compound of social identity and symbolic attachment that we might even dare call “class consciousness”?

From the New Deal to the New Democrats, the symbolic allegiances that have most damaged American social-democratic politics have been whiteness and maleness. For better or worse, they are with us still.

But by bringing so many white men into the social-democratic tent — not through sexist innuendo or racist dog whistles, but by appealing to a profound sense of class grievance — the Sanders campaign has pointed a way forward.

The promise of class politics, after all, is not only that it can threaten the interests of the few, but that it can unite the struggles of the many. After the final primary elections this month, the Sanders campaign may come to an end. But class politics isn’t going anywhere.