The unpredictable and surprising result of the 2016 GOP presidential nomination contest has called into question what many had identified as the paradigmatic political science understanding of the modern presidential nomination process, The Party Decides, by Marty Cohen, David Karol, Hans Noel, and John Zaller (Reference Cohen, Karol, Noel and Zaller2008). Nate Silver (Reference Silver2016), for example, called it the “most cited,” “most maligned,” and “most misunderstood” book of the election cycle (see also, Friedersdorf Reference Friedersdorf2016; Kolowich Reference Kolowich2016; Sheffield Reference Sheffield2016). Donald Trump’s nomination does not disprove the theory, but it does illustrate its limits. Indeed, Cohen, Noel, and others have offered several explanations for why the 2016 Republican nomination did not follow the storyline, noting that conditions may be changing in ways that make party insiders less likely to coalesce behind a candidate in recent election cycles. Footnote 1 My thesis is that insider endorsements can influence the outcome of a presidential nomination race, but they do so but under more limited circumstances, often less extensively, and to less effect than what is argued in The Party Decides. The basis for this claim lies in recognizing that party elites have varied in their propensity to endorse presidential candidates. Footnote 2
The ability to put candidates on the ballot makes political parties the arbiters of representative democracy (e.g., Schattschneider 1960). As arbiters, the parties have the capacity to choose candidates that serve partisan interests, potentially at the expense of the broader public (Bawn, Cohen, Karol, Masket, Noel, and Zaller Reference Bawn, Cohen, Karol, Masket, Noel and Zaller2012). But party insiders are conditional arbiters of presidential nominations. Their ability to influence the outcome is conditional on the extent of their participation in trying to influence the race, the timing of their involvement, and the degree of convergence that is attained. The Party Decides (TPD) focused on the timing and convergence of endorsements, but did not consider participation rates. Looking at participation indicates that party insider convergence on a frontrunner is often less extensive than suggested by looking only at the concentration of endorsements. This raises the questions of when and why party insiders participate and converge or not. This study follows Steger (Reference Steger2015) in offering a correction to The Party Decides by arguing that political party insiders unify sooner and more extensively when the party coalition is stable and there is a candidate in the race who has demonstrable national support. Party elites remain on the sidelines when the party is divided (Steger Reference Steger2015) or when there is uncertainty about the popular appeal of candidates (Anderson Reference Anderson2013; Ryan Reference Ryan2011; Steger Reference Steger2015; Whitby Reference Whitby2014). Evidence that candidates’ popular appeal influences decisions to make an endorsement further suggests that party insiders have less impact than it appears. As arbiters, party insiders facilitate more than direct the nomination of a preferred candidate.
When insiders divide or remain uncommitted, voters become relatively empowered because they have a larger number of viable candidates to select from and they can exercise a more independent voice.
The failure of elites to converge on a candidate during the invisible primary is not a new phenomenon. There have been two patterns of presidential nomination campaigns since 1980 (Steger Reference Steger2013). One is the TPD party-centric pattern, in which party insiders and groups coordinate and signal their support to the media, donors, activists, groups, and rank-and-file party identifiers which candidate is viable, electable, and preferable on political and policy grounds. The 2016 Democratic presidential nomination nicely illustrates this pattern. By the end of the invisible primary, Hillary Clinton had the endorsements of the vast majority of elite party officials. She gained a substantial advantage in campaign organization and fundraising and she received most of the news media coverage of the Democratic campaign. She maintained majority support in national opinion polls despite withering attacks from Republicans on the Right and from Bernie Sanders and progressives on the Left. Footnote 3 While Senator Sanders mounted a credible campaign and attracted substantial support in the caucuses and primaries, he came up short. Party insiders appear to have helped Clinton weather the storm by coalescing behind her, talking up her campaign in the media, and sending helpful signals to attentive publics. Footnote 4
The 2016 Republican nomination follows the other, candidate- and campaign-centric pattern in which party insiders fail to engage and unify behind a candidate (Steger Reference Steger2013; Reference Steger2015). Only a third of Republican governors, senators, and representatives endorsed any candidate, and no candidate received more than 27% of the endorsements made before the Iowa Caucus. Footnote 5 When insiders divide or remain uncommitted, voters become relatively empowered because they have a larger number of viable candidates to select from and they can exercise a more independent voice. Candidate and campaign-centric factors like candidate appeal, campaign spending, media coverage, and the sequential selection process become relatively more important for determining the nominee (e.g., Aldrich Reference Aldrich1980; Bartels Reference Bartels1988; Haynes et al. Reference Haynes, Gurian, Crespin and Zorn2004; Norrander Reference Norrander1993; Reference Norrander2006).
It matters which pattern emerges in a nomination race because the two scenarios differ in who wields power over the nomination. Party insiders and ancillary groups appear to play a powerful role in determining the nominee to the extent that they participate and coalesce behind a front-runner during the invisible primary. When they have done so, caucus and primary voters have gone along with the party’s choice since 1980 in every instance. Party stakeholders’ influence dissipates when they take a wait-and-see approach or when they divide their support among candidates. When this happens, more candidates remain viable options as the scope of conflict expands in the caucuses and primaries. Schattschneider’s (1960) scope of conflict thesis holds that the balance of power might be altered as new participants join the struggle, which is the reason many long-shot candidates decide to enter the race. New participants have the greatest potential to determine the outcome when no candidate has an outsized advantage in elite support, campaign finance, media coverage, organization, or resonance with party identifiers. Clinton had an outsized advantage in 2016 and Sanders could not overcome that. None of the traditional Republican candidates did and Trump benefitted from that.
ELITE ENDORSEMENTS DURING THE INVISIBLE PRIMARY OF 2016
Cohen et al. (2008, 232) argued that party elites and groups unified behind a candidate before the caucuses and primaries in nine of the 10 open presidential nominations from 1980 to 2004, with 2004 being the exception (see also Steger Reference Steger2000, Reference Steger2008). These studies measured candidate’s endorsements as a percentage of all the endorsements received by the candidates. This measure, however, can overstate the extent of party support for the front-runner because it ignores the rate of participation by potential endorsers. But first it is necessary to explain why participation is important and to note differences in the data used in various studies.
As I explain more fully elsewhere (Steger Reference Steger2015), endorsements matter mainly in the aggregate. The impact of an endorsement by a given politician may or may not move party activists and identifiers, depending on whether the target audience is aware of an endorsement and their opinion of the endorser. This varies across endorsers and cue recipients. For example, an endorsement by Senator Elizabeth Warren probably carries weight with progressive Democratic activists but it might not carry as much weight with African American Democrats compared to an endorsement by Rep. John Lewis (GA -5th), and vice versa. When large numbers of politicians from across demographic, geographic, and ideological spectra of the party endorse a candidate, there is a greater chance that people will know which candidate is being signaled as preferable and that the various constituencies will receive a positive cue. If only a small fraction of elites participate in this game, their endorsements send a weak signal to other insiders, the media, donors, groups, activists, and the identifiers of a party who vote in caucuses and primaries. This is notable when similarly situated politicians refrain from making an endorsement, as was often the case for Southern Democrats in the 1980s. More generally, non-participation signals uncertainty and trepidation about the appeal of candidates in the race (Ryan Reference Ryan2011; Whitby Reference Whitby2014).
The 2016 nominations illustrate these points nicely. Virtually the entire array of Democratic insiders, including many progressives, endorsed Hillary Clinton during the invisible primary. That potent signal was communicated repeatedly to voters through the media. Clinton also benefitted indirectly as various officials talked up Clinton in the media, allaying doubts and affirming her status as a candidate who could beat the Republicans. Republican insiders, in contrast, mostly sat on the sidelines while their (latently) preferred candidates floundered and Trump gained momentum. Early in the invisible primary, the “establishment” candidate was Jeb Bush, but party insiders gave him very tepid support—almost no elected officials outside of Florida endorsed Bush. Party insiders just did not add their voice to this campaign until it was too late, and then the efforts were half-hearted and incomplete (e.g., Azaria Reference Azaria2016).
Republican insiders, in contrast, mostly sat on the sidelines while their (latently) preferred candidates floundered and Trump gained momentum.
There are significant differences in the data that give rise to the conclusions of this paper and those of Cohen et al. (Reference Cohen, Karol, Noel and Zaller2008). This study and others that have looked at participation use endorsements by elite elected officials and/or superdelegates (Anderson Reference Anderson2013; Ryan Reference Ryan2011; Vavreck and Sides 2012; and Whitby Reference Whitby2014). TPD data include these endorsements but also those of lower level elected officials, former elected officials, party officials, group leaders, and celebrities. Depending on the year, endorsements by governors, senators, and US representatives constitute about 55% of the data points used by TPD. The distinction matters. Party and elected officials’ incentives differ from those of groups and activists (e.g., Keech and Matthews Reference Keech and Mathews1976; Butler Reference Butler2004). Party and elected officials give primacy to selecting a candidate who can win—who will help or at least not harm candidates lower on the ticket. Groups and activists, in contrast, give relatively greater weight to policy as Cohen et al. argue. Party activists also appear to view candidate viability and electability through rose-colored lenses—viewing a candidate who is preferred for policy reasons, as more viable and electable than may be the case from an objective standpoint (e.g., Abramowitz Reference Abramowitz1989; Kenney Reference Kenney1993). Bernie Sanders’ supporters seem to reflect this tendency in the 2016 nominations. The Party Decides blurs this distinction, blending elected officials, group leaders, and activists together theoretically and empirically. The potential divergence between elected officials and others needs to be kept in mind since the evidence used here does not necessarily nullify their arguments about party activists and groups.
While the measures differ, TPD data show a moderately strong correlation, r = .75 between endorsements by elite elected officials and endorsements by others for the years of overlap. In most years, when party elites converge on a candidate, so too do others making endorsements in the campaign. The most common differences are that some factional, “second-tier,” and “outsider” candidates pick up more endorsements from non-elected officials or state level officials. These differences generally reduce the degree of convergence, which would work against the TPD thesis so it does not change substantially the result of this study. The exceptions to this may have been Walter Mondale in 1984 and George H. W. Bush in 1988 who received more endorsements from groups than they did from elite elected officials, many of whom refrained from making a public endorsement before the caucuses and primaries. Footnote 6 Limiting the analysis to elite elected officials has the very important advantage of providing a known population of potential endorsers, which makes it possible to calculate endorsements as a percentage of those who could make an endorsement. Thus the data provide a measure of party coalescence while accounting for participation rates of potential endorsers.
Figure 1 shows three endorsement measures for Clinton, the Democratic Party front-runner for the 2016 presidential nomination. Clinton ended the invisible primary with 97% of all the endorsements received by a candidate (the dotted line). By this measure, Clinton was the consensus pick of Democratic insiders. Further, 86.92% of elite elected Democratic officials made an endorsement during the invisible primary—indicating very widespread participation (the solid line). Clinton thus received the endorsements of 84.4% of all of the elite elected Democratic officials who could have made an endorsement during the invisible primary (the dashed line). This epitomizes the TPD argument. When there is widespread participation by party insiders and they converge on a nominee, there will be no substantial difference in the picture portrayed by the different measures of endorsements. Clinton’s nomination is certainly consistent with the argument that party insiders can influence a race by coordinating their support for a candidate, talking up the candidate in the media, and discouraging partisans and donors from supporting rivals for the nomination. She was clearly the party’s choice.
Figure 1 The Democratic Front-Runner’s Share of Endorsements, the Percentage of Elite Elected Officials (EO) Making an Endorsement, and the Front-Runner’s Share of All Possible Endorsements During the Invisible Primary for 2016
Invisible primary endorsements by elite Republican elected officials do not fit this pattern (see figure 2). The candidate shares of endorsements indicate that no candidate was the “clear” establishment choice, as elite Republican officials divided their support among the candidates. Jeb Bush had the most endorsements with 26.95%. Marco Rubio had 23.48% of the endorsements. Ted Cruz had 15.65% of the endorsements. John Kasich and Chris Chistie each received fewer than 7% of the endorsements. The remaining endorsements were spread among the other candidates—except for Trump, who received no endorsements before the Iowa caucus. Trump’s endorsements are represented by the X axis. Republican insiders could not even coordinate their opposition to Trump (e.g., Azaria Reference Azaria2016).
Figure 2 The Percentage of Elite Elected Republican Officials Making an Endorsement, and the Major Candidates’ Shares of Possible Endorsements During the Invisible Primary for 2016
Only 34.2% of elite Republican elected officials made any endorsement during the invisible primary (the dashed line in figure 2). This is expected when party elites are uncertain about which candidate will resonate with constituents or when party elites find the candidates unacceptable on policy grounds (Anderson Reference Anderson2013; Ryan Reference Ryan2011; Whitby Reference Whitby2014). Accounting for participation, Jeb Bush had the support of only 9.25% of the elite elected officials who could have made an endorsement. This is thin evidence for calling Bush the establishment candidate. As Bush faded in the polls during the fall of 2015, Rubio emerged as the establishment candidate but he had the support of only 8.06% of these potential endorsements. Cruz, who emerged during the primaries as the main alternative to Trump, had only 5.37% of possible endorsements at the end of the invisible primary. Kasich was endorsed by only 2.1% of these officials. In short, Republican Party insiders largely stayed on the side-lines and those who did participate divided their support among the candidates, thus depriving partisan voters of a clear signal about the candidates. This pattern is not as uncommon as it seems. A majority of elite elected officials refrained from making an endorsement in half of the open presidential nominations since 1984.
In open Democratic presidential nominations between 1984 and 2016, there have been only two races in which a majority of elite elected officials endorsed the candidate who became the party’s nominee (see figure 3). In the other races, only a minority of elite elected officials endorsed a candidate during the invisible primary—including the 1984 race featuring former Vice President Walter Mondale. While Mondale received other endorsements not included in this analysis, elite elected officials who were concerned with how the nominee would affect their own electoral chances, were less likely to endorse Mondale. Only Gore in 1999 and Clinton in 2015 had majority support in national opinion polls, and in both cases, a majority of elite elected officials endorsed them during the invisible primary.
Figure 3 Democratic Nominee’s Share of Endorsements as a Percentage of all Possible Endorsements by Elite Elected Officials, 1984–2016
In open Republican races, there have been more races with elite elected officials making an endorsement during the invisible primary—but only one case in which a majority of officials made an endorsement during the invisible primary (the 2000 nomination). In two cases—G. H. W. Bush and Bob Dole—a large minority of elected officials endorsed a candidate during the invisible primary (see figure 4). In both of these cases, the endorsed candidates averaged between 35% and 45% in national polls during the invisible primary year and prior years leading up to the election. In the remaining cases, fewer than 20% of elected officials endorsed a candidate during the invisible primary.
Figure 4 Republican Nominee’s Share of Endorsements as a Percentage of Possible Endorsements by Elite Elected Officials, 1988–2016
Party insiders are conditional arbiters of presidential nominations. They can help their preferred candidate get nominated—but mainly to the extent that they engage in the campaign and unite in support of that candidate. Whether and the extent to which party insiders coalesce during the invisible primary depends on the unity of the party coalition and whether a candidate enters the race with a substantial lead in early polls of party identifiers and leaners (Steger Reference Steger2015).
Regarding the party unity, the Republican Party has experienced serious internal divisions in recent years (Olsen and Scala Reference Olsen and Scala2015). Such divisions make it hard for a candidate to satisfy all of the various factions. Surveys by the PEW Research Center show that Republicans continue to be divided. Footnote 7 Even after Trump had locked up the nomination, Republican identifiers and leaners who had supported other candidates were much less likely to support him in the general election. The Democrats also had intra-party divisions, with Sanders’ supporters being less likely to support Clinton as the nominee, but these differences were not as great as they were for Republicans and Sanders’ supporters moved more to support Clinton after the primaries were over. Footnote 8 Divisions in the parties may reflect long-term patterns of realignment (e.g., Paulson Reference Paulson2007) or transient divisions as election-specific factors like candidate appeal and issues affect participation by party constituencies (Steger Reference Steger2015). As a transient factor, Trump’s personal and policy characteristics contributed to Republican support of, and dissatisfaction with, his candidacy. However, given that Republican Party divisions have grown since 2000 (Olsen and Scala Reference Olsen and Scala2015), the divisions in 2016 appear to be more than a transient deviation driven by election year idiosyncrasies. As the Party has divided internally between moderates, mainstream conservatives, very conservative religious and very conservative secular factions, it has become exceptionally hard to have a single candidate who would draw strong appeal to activists and identifiers across all four groups. Elected officials face greater political risk to their own careers in endorsing a candidate, knowing that doing so can potentially alienate activists and identifiers in their own party. The safer route for elected officials is just to stay quiet and let their party’s activists and identifiers settle on a nominee.
The other critical factor affecting party insider coalescence is the availability of a candidate with demonstrable appeal in national polls several years before the election year. The Democrats had Clinton who consistently polled above 50% in surveys in 2013–14. The Republican Party lacked a clear front-runner in 2013–14. None of the Republican candidates polled above 20% during these years. With no clear leader at the onset of the race, a large number of ambitious Republican office holders and former office holders were willing to enter the race, making it harder for party insiders to discern which candidate would catch fire with party activists and identifiers (Steger Reference Steger2015). This is the first Republican nomination in the modern era in without candidate who had at least a large plurality of support in these early national polls. This has often occurred in Democratic nomination races, in which the leader in early national polls did not enter the race (e.g., Ted Kennedy in 1972 and 1976; Gary Hart in 1988; Mario Cuomo in 1992; and Hillary Clinton in 2004). In the races for which there is data, the majority of Democratic elites refrained from endorsing a candidate during the invisible primary in each of these races.
The power of party insiders to influence the nomination is weaker than it appears because party insider convergence depends in part on the popularity of the candidates who run.
The power of party insiders to influence the nomination is weaker than it appears because party insider convergence depends in part on the popularity of the candidates who run. Party elite participation and unity correspond to candidate support in national polls taken three and four years before the election. These early polls were not analyzed by Cohen et al. (Reference Cohen, Karol, Noel and Zaller2008). Elites are more willing to endorse when there is a clear front-runner in these earlier polls. When there is no candidate with a majority in these early polls, or the leading candidate does not run, then there is low participation and less unity among party elites. This does not mean that elected officials are more reactive than proactive. They want a nominee who will be electable and who will help or at least not hurt candidates lower on the ballot in the coming election cycle. Early polls provide an indicator of which candidate(s) have the kind of stature to do that. Elites simply use information that they have available. Polls are probably one piece of information used by elites who tend to take a wait-and-see approach in absence of evidence of a popular candidate.
Even if we discount early polls as exercises in name recognition, party insiders may still rally to a candidate in anticipation of that candidate resonating with party activists and identifiers. Anticipated reactions reverses the temporal sequencing in the model. If anticipated reactions drive endorsement decisions, then it is still the preferences of mass partisans that drive endorsements—even though the data will indicate that endorsements precede mass support in polls. Studies of individual endorsement decisions are consistent with this interpretation (Anderson Reference Anderson2013; Ryan Reference Ryan2011; Whitby Reference Whitby2014). If this is the case, then elites’ collective influence is essentially that of making the selection process more efficient by weeding out the chaff from the wheat before the caucuses and primaries. Efficiency matters from the standpoint of party elites because it reduces the expenditure of campaign resources that could be more beneficially directed toward the defeat of the nominee of the opposing party. In this sense, the arbitration role of party insiders may be more one of facilitation than direction.
The Party Decides thesis also may no longer have the potency that it did in the 1980s and 1990s because candidates have greater capacity to compete without elite facilitation of access to campaign resources. Presidential nomination campaigns require a lot of money and media coverage. Part of the TPD argument is that collusion among party elites and group leaders enables a front-runner to gain more of these campaign resources. Party elite and group leader support does in fact facilitate fundraising from donors tied into party networks (e.g., Dowdle, et al. Reference Dowdle, Limbocker, Yang and Sebold2013; Mitchell et al. Reference Mitchell, Sebold, Dowdle, Limbocker and Stewart2015). Candidates depended on this access more in the 1980s and 1990s than in the most recent nomination cycles. Raising big money in those years required access to fund-raising networks of politicians, bundlers, and direct mail experts because the campaign finance regulatory framework capped the contributions of individual donors. The unraveling of campaign finance regulation has enabled candidates to raise money quickly from a small number of “angel” donors. The Internet has enabled popular candidates to crowd-source the financing of a campaign without support from established party donor networks. Sanders, for example, ran a viable campaign because he was able to get the necessary resources from over two million individual donors. Trump represents a third kind of candidate who does not need the support of traditional donor networks. Trump has the personal resources to run a viable campaign, though he did not need to because of an unprecedented capacity to use print, broadcast, and social media to appeal to partisan voters. The continued fragmentation of the news media, talk radio, blogs, and social media have also diminished the value of party insiders “talking up” a candidate with journalists. Candidates have great capacity than ever to appeal to audiences directly and through word of mouth. The Sanders and Trump campaigns, particularly early on, illustrate the potency of alternative media and social media for generating buzz independent of what party insiders are conveying to party voters through traditional media.
Finally, the 2016 presidential nominations may be notable for the strength of populist appeals that deviate from the mainstream party policy orthodoxy. Populism has a lot of different meanings and uses (e.g., Girdron and Bonikowski Reference Gidron and Bonikowski2013), but the common thread is that populism is anti-elite or anti-establishment sentiment. People who are angry or opposed to the party establishments can hardly be expected to follow the lead of party insiders. Both the Sanders and Trump campaigns drew support from populist elements in their respective parties, making it difficult for party insiders to prevent these candidates from disrupting the party coalitions with policy appeals that deviate from party orthodoxy.
The left-wing populism of Sanders has fused progressive economic policy ideas, cultural inclusivity, and anti-establishment rhetoric that targets Wall Street and “enablers” from the Democratic Party. The right-wing populism of Trump has fused economic nationalism, cultural exclusivity, and anti-establishment rhetoric accusing both political parties of abandoning white working class Americans. Both approaches pose serious challenges to the cohesiveness of the party coalitions, but Trump’s populism deviates more from Republican policy orthodoxy and has greater potential to disrupt the Republican Party coalition. It remains to be seen whether these challenges will result in more durable divisions of the parties or force party elites to alter their policy positions to accommodate disaffected partisans. For now, we can say that these movements limit the extent of party insider influence in nominations. The result of it will require more research as events unfold.
CONCLUSIONS
The Party Decides presents a strong case that party insiders are incentivized to coordinate their efforts behind a candidate who is signaled to be preferable or at least acceptable on policy grounds and who can win. Cohen et al. (Reference Cohen, Karol, Noel and Zaller2008) describe the invisible primary as a long national discussion among party officials and groups who signal each other and the media, donors, activists, and identifiers about which candidate should be supported in the caucuses and primaries. That argument is largely true, but party insiders are conditional arbiters of presidential nominations. Their influence on the race depends on how extensively they participate and coalesce behind a candidate during the invisible primary. The party’s elite elected officials, at least, vary in their engagement in the nomination race, which raises the questions when and why party elites vary in their participation.
While there may be additional factors that influence these decisions, party unity and the availability of a candidate with popular appeal among partisans appear to be factors that systematically influence these decisions. Party elites are more willing to make endorsements when the party coalition is unified and stable—which was the case for Republicans from the Reagan to George W. Bush years and less so since. Democrats were divided from the 1960s through the early 1990s (Mayer 1996) and their elite elected officials were much less likely to engage in the endorsement signaling game during those years. Democrats continue to have intra-party divisions, though the extent of the divisions has diminished as their conservative Southern wing evaporated. The remaining Democratic elected officials have been more willing to make endorsements. While parties can decide, their ability to unify also depends on the availability of a candidate with demonstrable popular appeal among partisan identifiers. Party elites have been much more likely to make an endorsement when they have as a candidate a politician who had strong support in national opinion polls three and four years before the election.
That elite elected officials are concerned with the popular appeal of a candidate should not be surprising. Their own electoral fates and those of their partisans in office may depend on it. Elected officials, more than group leaders and party activists, are highly cognizant of the fact that they must win office and control government to deliver benefits to their constituents. Winning is a necessary condition for getting policy. This does not mean that party insiders lack influence. Party elites still wield some influence even if they react to prior polls and/or act in anticipation of future appeal. Elites’ collective influence is essentially that of making the selection process more efficient by weeding out the chafe from the seed before the caucuses and primaries. Efficiency reduces the expenditure of campaign resources that can be more beneficially directed toward the defeat of the nominee of the opposing party. It also helps avoid visible intra-party fights that have potential to tear apart a party coalition. Party elites wanting to win can benefit from moving the campaign toward a candidate who can satisfy these needs. That is important, but it is a lesser degree of influence than what is argued in The Party Decides. The circumstances in which party insiders participate and unify are more conditional than what The Party Decides holds. Party insiders are conditional arbiters of party coalitions, potentially facilitating the selection of nominees.
