Rejoinder: How Special was 2016?

Five responses from leading scholars of American politics have given us a great deal to think about. Several themes emerge from the responses. The first is the challenge of the addressing how relevant the 2016 election will be for understanding the future of American politics. Several of the discussants also challenge our thinking about the role of white working class pundits, and about how political scientists should think about demographics and politics more generally. In the study of comparative politics, the literature on case selection demands that scholars answer the question, “What kind of case is this?” before proceeding; see for example Gerring and Seawright (2008). Looking forward, is the 2016 typical with some unusual features, or will it in retrospect seem unusual? The answer to this question depends on the research question and the variables of interest. As a result, elections scholars may need to think more deeply about the kinds of questions we pursue and the theoretical assumptions we make. However, we must also wait to find out the impact of 2016 on subsequent contests. As we attempt to classify the 2016 election, we are stuck doing some guesswork. Noel urges scholars to ask how an outlier can sharpen our theories. Masket and Victor both pose the question of whether last year’s contest will turn out to have been anomalous or a new normal. Finally, Shapiro asks whether the election was really so unusual after all. These different classifications suggest not just different interpretations, but that the implications of 2016 depend on what the researcher seeks to explain.


The Primary and General Elections
Taken together, several of the responses point to a sharp distinction between the unusual nature of the nomination contest and the predictable aspects of the general election. Beginning with the nomination phase, Noel suggests that one thing we can learn is that party elites will not throw their support to a rogue frontrunner, even in the face of polling strength and media attention. Masket observes that while electability may have traditionally been a central consideration in nomination contests, Trump's election suggests that the emphasis on traditional qualities may have been misplaced. Taken together, these observations suggest several possibilities.
Masket's observation suggests that we might expect a wider range of inexperienced or ideologically extreme candidates in CONTACT Andrew Gelman gelman@stat.columbia.edu Department of Statistics and Department of Political Science, Columbia University,  Social Work Building, New York, NY . future presidential nomination contests. If Noel is correct that 2016 was an anomaly, and party elites regain control of the process in the future, then we might expect to see deeper and more intense clashes within parties. Such clashes in the past have led to institutional changes like the McGovern-Fraser reforms. In other words, it is possible that parties will revert to the kind of elite control that The Party Decides describes. Noel also argues that Trump's candidacy is better characterized as an elite effort than as a grassroots phenomenon. Nevertheless, if Trump's candidacy has opened up the possibilities for new types of candidates, there will be potential for significant clash between supporters of those candidates and what we used to call the party establishment.
If party hands have slipped permanently off the reins of nomination control, then Masket's electability hypothesis could instead portend chaos. Parties with weak mechanisms of formal control and little agreement about what kind of candidate might be acceptable are vulnerable to even worse coordination problems than those we saw in 2016 on the Republican side. Furthermore, it is worth noting that although the Democrats' process last year conformed to conventional expectations, there are no guarantees that they will escape outsider capture in the future.
While the nomination was highly unusual, many features of the general election ran according to schedule. Partisan polarization performed as expected, driving much of the result. However, several discussants point to polling challenges as a consistent obstacle to better predicting and understanding elections. Shapiro and Hunt both suggest technical issues in polling, adjusting for sampling issues and problematic assumptions about state-level trends. Victor's point about networks and voting speaks to gaps in our substantive knowledge about voter psychology and behavior that concern us, even if we can accurate predict the vote. Similarly, Victor questions whether the election really demonstrates that the ground game does not matter. She presents this as a methodological problem, insofar as we cannot observe a counterfactual in which the candidate made different campaign choices. But it is also a substantive challenge to our understanding of the classic question about whether, and how, campaigns matter-especially for an unconventional campaign such as Trump's, whose "asymmetrical warfare" strategy stands apart from the usual argument that, for the general election campaign for president, campaign effects largely cancel each other out. (Even the fabled Obama 2008 campaign wizardry was ultimately only credited in making a difference in one or two swing states such as Indiana.) In other words, a challenge that remains for social scientists is to figure out which of our problems are with our tools, and which are with our substantive knowledge and conceptual understanding of who votes and why.
Comparison of the primary and general elections also indicates another issue not addressed in our article, which is the interaction of these two phases of the campaign. The usual recommendation, attributed to Richard Nixon, is to run to the right (or, for Democrats, to the left) during the primary and then run to the center in the general election. 2016 was different, however, in that Trump's shaky standing among many Republicans implied that his immediate task in the general election was to retain core Republican support, thus, he had to stay to the right during the general election.
This raises the question of why Clinton was not able to take advantage of Republican disarray by firmly occupying the center. Suppose that, as it appeared likely at various points during the campaign, that the Trump candidacy had collapsed, either staggering along with the support of the Republican establishment or replaced with a Pence campaign reminiscent of Dole's hopeless 1996 effort. Perhaps, then we would be arguing about whether Hillary Clinton's centrist conservatism represents a break from political polarization or just a brief respite from the struggle between the two parties?

White Working Class Voices
After 2008 and 2012, a variety of commentators have engaged in questions about how identity politics should factor into Republican strategy, whether the party should double down on its appeal to older white voters or instead broaden its demographic base. After Trump's victory in 2016 the questions have gone the other way, as to how and whether an identity politics appealing to minorities was getting in the way of any Democratic efforts to win back white, non-college educated voters. In our original essay, we commented on these questions in several ways. We argued that "a white working class pundit is something to be, " citing the combination of authenticity and insight offered by commentators like Michael Moore on the left and J.D. Vance on the right. Noel cautions that a working class pundit might be a thing to be, but not to listen to. We respond that there are different ways to listen and take perspectives seriously. The challenges of making predictions have been exposed in the past few years, and on some level, it is no great feat to suggest that one of two major candidates in a national race might win it. However, our original contention is that commentators like these offered insights not often seen in national political coverage. Punditry is punditry, and anyone who dabbles in it will be confined by the limits of the genre. But pundits have the potential to reach wide audiences, and the national discourse will be richer for the inclusion of a similarly broad range of voices. If anything, our initial claim did not go far enough in calling for the inclusion of those perspectives, not to take their conclusions as the final word, but as a starting point.
Our original piece also commented on the strategic importance of white voters to winning the presidency because of the structure of the Electoral College, based on the calculations of Gelman and Kremp. This is not the same as the much thornier questions about the normative implications of emphasizing the concerns of this particular group. In the time that has elapsed since our initial submission, the debate about these questions has advanced in a particular way. Political theorist Mark Lilla (2016) has written extensively, and to great fanfare, about the need for Democrats to transcend "identity politics" and instead "concentrate on widening its base by appealing to Americans as Americans. " Numerous analysts, including political scientists like Lee Drutman (2017), have argued instead that Democrats may find it impossible to deemphasize race issues at this point, and that the party might instead focus on electoral reform (read: gerrymandering) and turnout. In other words, arguments that the Democrats should focus on lost white voters could be untenable on multiple levels. Shapiro suggests that such an approach might be problematic, both strategically and normatively. Masket picks up this line of thinking, asking, "How many profiles were written of urban African American voters after Barack Obama's election? How many of Latinos during the early stages of Donald Trump's candidacy? What of women who supported Hillary Clinton? If there is a bias toward covering a certain segment of the population, why?" Masket's question of media coverage-suggesting a bias toward presenting the opinions of small-town conservative whites, ironically as a reaction to the media's self-perceived urban liberal perspective-has significance for party politics scholars, who may view a permanent parting between the Democrats and the white working class as the final end of the New Deal coalition and the consolidation of a new realignment. For scholars of campaigns and voting behavior, questions about mobilization remain. Why were Trump's appeals so effective in winning white votes within the Republican party and then retaining the votes of skeptical white Republicans during the general election? Hypotheses about health, economics, and, of course, racial resentment abound. Ashley Jardina (2017) points out that white identity, too, is a significant force in American political life. On the other side of the coin, why was Clinton unable to draw more turnout and win more votes among minority groups, as commentators expected based on the Trump campaign and also Clinton's resounding success with minorities during her primary election victories? This is both a question about social conditions and about the election process.
Caveats about highlighting stories about white voters over others are warranted. Masket's concern about this problem, combined with Shapiro's observations about the parallels between 2016 and 1948-another year in which racial concerns weighed heavily-remind us that these campaigns do not occur in a historical vacuum. Scholars and journalists cannot turn their attention to white working class or disaffected voters without considering how white interests and preferences have historically been treated as more important than other perspectives.

History and Qualitative Methods
Finally, the responses to our piece highlight the need for a multi-method approach to studying electoral politics. Shapiro establishes important parallels with previous elections. Hunt notes, "But qualitative research in which researchers listen to voters explain their views and feelings about the candidates and the country provides a much deeper and richer understanding than a single number. " The study of elections has become dominated by quantitative methods, which is logical given the available data. However, while these methods are amenable to some types of questions, they provide limited insight into how voters make and characterize their choices, and how they absorb and perceive political information.
Many of the puzzles of the 2016 election remain. As scholars and analysts attempt to understand what happened and offer advice for how the nation should move forward, more questions and debates emerge. The five thoughtful responses to our analysis illustrate the importance of paying close attention to our theories and data sources, as well as to the kinds of questions we ask.