Skip to main content
ABC News
The Republican Party May Be Failing

“The Party Decides,” the 2008 book by the political scientists Marty Cohen, David Karol, Hans Noel and John Zaller, has probably been both the most-cited and the most-maligned book of this election cycle. After re-reading the book, which underpinned a lot of our early analysis of the primaries here at FiveThirtyEight, I’ve come to another conclusion: It’s probably also the most misunderstood book of the 2016 campaign.

The caricature of the book seems to be this: “The Party Decides” posits a clash between “the establishment” and rank-and-file voters and claims that the establishment always prevails. But that’s not really what the book says. Instead, the book argues that the major American political parties are broad and diverse coalitions of politicians, activists and interest groups, many of whom would never think of themselves as belonging to the political establishment.

However, the book does presume that, in part because of their breadth and diversity, American political parties are strong institutions. Furthermore, it assumes that strong, highly functional parties are able to make presidential nominations that further the party’s best interest.

For a variety of reasons, the nomination of Donald Trump would probably not be in the best interest of the Republican Party. Such an outcome this year, which seems increasingly likely, would either imply that the book’s hypothesis was wrong all along — or that the current Republican Party is weak and dysfunctional and perhaps in the midst of a realignment.1

The party isn’t “the establishment”

You might associate “The Party Decides” with an empirical claim made in the book: Endorsements made by influential Republicans and Democrats are a good predictor of who will win each party’s nomination. At FiveThirtyEight, we’ve been keeping track of a subset of those endorsements, those made by current governors and members of Congress in each party.

Our focus on endorsements by governors and members of Congress is mostly a matter of convenience, however.2 In “The Party Decides,” the authors consider a much broader array of endorsers, including state legislators, labor unions, interest groups and even celebrities. This is important because, in contrast to earlier scholarship that thinks of parties as consisting solely of politicians and party organizations like the Republican National Committee, the authors of “The Party Decides” take a more inclusive view. Their parties include not just elected officials but also “religious organizations, civil rights groups … organizers, fundraisers, pollsters, and media specialists” and even “citizen activists who join the political fray as weekend warriors.”

That means the term “Republican establishment” (in addition to its other problems) is not a good approximation for the book’s view on the party. “Anti-establishment” members of Congress, such as the Freedom Caucus, are parts of “the party” as much as members who always vote with leadership. Lots of people within Washington, D.C., are considered to be part of the “party,” but so are people in Kentucky and Alaska. The editors of National Review magazine are probably3 part of the Republican Party as the book’s authors would define it, but so are bloggers at RedState and conservative talk-radio hosts in Iowa.

The authors of “The Party Decides” use phrases like “party elites” and “party insiders” to describe this collection of people. An alternative that I sometimes prefer is “influential Democrats” and “influential Republicans.” That’s really the bottom line: These people have some ability to influence the nomination,4 and they have some interest in doing so. That influence could take many forms, including holding a position of power, having access to a donor network, possessing scarce skills or knowledge, contributing time or money, or having the ability to persuade others through a media platform.

It might even be tempting to boil down “The Party Decides” to an idea like this: You ought to pay attention to what influential people who care about a party nomination are doing, since they can have a lot of say in the outcome. Indeed, that’s probably a better representation of “The Party Decides” than the idea that a monolithic establishment always wins.

But the book has something more than that in mind. Parties are not merely collections of influential people; those people are supposed to be working together to further the party’s interests. If they “can agree to work together for a candidate, as usually they can, they constitute a formidable political force,” the book says. But they cede much of that power when they remain splintered.

The mechanics of this are complicated, obviously. Some groups within a party care a great deal about winning office. Others are more interested in policy or ideological victories. Moreover, in a given election, a party can only nominate one candidate; if she wins office, she’ll have only so much political capital. Which issues get priority and which ones get short shrift?

But if the parties in “The Party Decides” are complicated, that’s because real American political parties are complicated, too. It’s not inherently obvious what anti-abortion activists, the National Rifle Association, the oil lobby and movement conservative intellectuals have in common — but all of them usually associate themselves with the Republican Party and they potentially stand to gain by working together under its banner.5 Historically, the result of this party-building process has been a punctuated equilibrium of parties that can be stable for decades at a time but which occasionally undergo rapid and dramatic realignments.

Strong parties nominate strong candidates

So the party always wins? Not quite. Blame the book’s title if you like (the long version is: “The Party Decides: Presidential Nominations Before and After Reform”). It seems to imply that the Republican and Democratic parties are all-powerful, with voters merely going along for the ride. That’s not quite what the authors say, however. “We do not claim that parties are juggernauts that always prevail,” they write in the first chapter.

The authors are also aware of the limited data available on party nominations. “The greatest point of vulnerability, in our view, lies in the thinness of the data that underlie the analysis,” they write. “Our main analyses involve sixty-one candidates, but these candidates ran in only ten nomination contests — and ten is not a large number for making inferences about a process as complicated as presidential nominations.”6 Being aware of these limitations is not the same thing as working around them, of course. But generally speaking, I think the book does a pretty good job under the conditions. (For geeky readers, I have a longer discussion in the footnotes.7)

One reason I say this is because the claims made by “The Party Decides” are modest. The authors aren’t saying that parties can wave a magic wand and nominate whomever they like. Instead, they posit that American political parties are robust and diverse institutions. And they claim that these parties make fairly rational choices in whom they nominate for president. The closest the book comes to a thesis statement is this:

Parties are a systematic force in presidential nominations and a major reason that all nominees since the 1970s have been credible and at least reasonably electable representatives of their partisan traditions.

There are a couple of things to unpack here. First, that qualification “since the 1970s.” That refers, in part, to the nomination process that’s been in place since the McGovern-Fraser reforms, which greatly increased voter participation in the system. However, it conveniently also excludes the Democratic nominations of 1972 and 1976, which were contested under the new system but resulted in the choice of factional candidates, George McGovern and Jimmy Carter.

The book’s view is that party elites had yet to learn the nuances of the new rules, whereas McGovern and Carter had clever strategies to exploit them. (In McGovern’s case, focusing on delegate accumulation instead of the popular vote; in Carter’s, understanding that a strong performance in Iowa could produce media-fueled momentum that would give him a leg up in subsequent contests.) Perhaps, but these years also suggest that the power wielded by party elites is fragile and that unconventional candidates can win if they (like Trump) pursue unconventional strategies.

Nonetheless, truly disastrous nominations like McGovern’s have been rare. Instead, parties have usually nominated candidates who, as the book puts it, are:

  1. “Credible and at least reasonably electable”;
  2. “Representatives of their partisan traditions.”

You might describe these two dimensions (as we sometimes have) as “electability” and “ideological fit.” The goal for a party is to find a candidate who scores highly along both axes. George W. Bush in 2000, for example, was acceptable to all major factions of the GOP, but he also began the race as a “compassionate conservative” with a highly favorable image among general election voters. It’s no surprise that Bush won his nomination easily.

At other times, the party must contemplate a trade-off between these goals. Sometimes, it will choose a candidate who breaks with party orthodoxy in important ways, but who has a lot of crossover appeal to general election voters. Bill Clinton in 1992 and John McCain in 2008 are good examples. Or, it may go for broke with an ideologically “pure” candidate whose electability is unproven. Sometimes, the gamble pays off, as it did for Republicans with Ronald Reagan in 1980, but there’s also the risk of winding up with the next Barry Goldwater. Note that Bernie Sanders and Ted Cruz, if chosen, would arguably8 fit into the category of ideologically pure but electorally dubious nominees.

Listen to the latest episode of the FiveThirtyEight politics podcast.


It has been extremely rare, however, for a candidate to be nominated while scoring poorly along both dimensions. McGovern is probably the best example, insofar as he was too radical even for many Democrats in 1972 and a disaster of a general election nominee.

Donald Trump might be another of those cases. It’s not clear what policy positions Trump really holds, but to the extent he has articulated them, they’re all over the map and not that well aligned with those traditionally held by Republican officeholders. However, unlike previous “mavericks” such as Bill Clinton or McCain, Trump is not very popular with general election voters. On the contrary, he’s extremely unpopular with independents and would begin the general election race with worse favorability ratings than any candidate to receive a major-party nomination before.

To some extent — at least until we see how the first few states vote — this is a reason to be skeptical of Trump. It’s possible that even if party elites don’t have much say in the process, Republican voters will figure out on their own that Trump is a risky nominee.

Put another way, the case for being doubtful of Trump’s nomination prospects never had all that much to do with “The Party Decides.” It was not as though Trump fit the profile of a typical Republican nominee but just lacked endorsements from party elites. Instead, Trump is an outlier in nearly every respect and in ways that suggest he could be extremely damaging to the Republican Party as its nominee. To have doubted Trump is to have given the Republican Party credit, perhaps too much credit, for being able to avoid a potential disaster.

But now that Trump has gone from “black swan” to prospective nominee, it’s worth asking another question: If Republican voters are on the verge choosing Trump, why aren’t party elites doing much to stop them?

So why isn’t the GOP stopping Trump?

Some of the reasons could be circumstantial. One way in which a factional candidate might win the nomination is by claiming the plurality when the vote among mainstream candidates is divided. That might be some of what’s happening this year. An unprecedented number of traditionally well-qualified Republicans entered the race. Although some of them have since dropped out, there’s still a pile-up in New Hampshire – a state where Republicans might otherwise have a chance of stopping Trump — with Marco Rubio, Chris Christie, Jeb Bush and John Kasich each polling between 8 percent and 12 percent of the vote.

Another reason could be if a candidate rides a wave of media-driven momentum to victory. Usually, we’d think of a momentum candidate as someone like Carter, who parlayed an unexpected “win”9 in the Iowa caucuses in 1976 to emerge from obscurity and top a disorganized field. Can Trump, who was a nationally renowned figure before he entered the race, really be placed in the same category as Carter? Maybe. The media coverage of Trump has been disproportionate and seems to be self-reinforcing, with polls and coverage begetting one another in a virtuous cycle.

But these aren’t the good excuses for party elites that they might seem. In fact, they are exactly the sorts of outcomes that party elites are supposed to intervene to prevent. One of the reasons to coordinate during the early, “invisible primary” phase of the nomination is to avoid a logjam of candidates later on. As for Trump’s media coverage, surely some of it reflects the fact that he’s a perpetual attention machine who generates good ratings. But in “The Party Decides,” party elites are supposed to serve as a counterweight to “the poor quality of media coverage, which covers leading candidates much more than others.” So far, they haven’t been able to change the narrative on Trump.

What power does the party really have?

If the “Party Decides” theory is at a loss to explain why GOP elites have failed to stop Trump, it may be because elites never had all that much power to begin with. Indeed, the book can be frustratingly opaque when describing how party elites motivate rank-and-file voters to go along with their choices. “The inner workings of the invisible primary are, as the name implies, hard to see,” it says at one point.

So I’ll try to fill in the some of the blanks by borrowing from the political scientist Joseph Nye’s distinction between “hard power” and “soft power” in international relations. Hard power consists of military and economic might. Soft power consists of non-coercive forms of influence, such as gaining global esteem by exporting popular culture.

In the context of presidential nominations, the analogy to “hard power” is rule-setting authority plus control over scarce resources. Modern political parties do have some of this. They control the rules by which delegates are chosen, for example, though attempts to rig the rules in favor of the elites’ preferred candidate can backfire. They would have quite a bit of power in the (unlikely) event of a contested convention. Party elites also have access to financial resources, though not a monopoly on them.10 The party may own quite a bit of data, an increasingly important resource.

For the most part, however, “The Party Decides” seems to think that party elites possess “soft power”: the power of persuasion. It assumes that party elites have largely the same goals as rank-and-file voters, but are more informed about which candidates to support, leaving the electorate “open to suggestion”:

All of which leads us to reason as follows: An electorate that is usually not very interested, not very well informed, and attracted to candidates in significant part because they are doing well is probably an electorate open to suggestion about whom to support. If, as we know to be the case, many primary and caucus voters are also strong partisans, what they want in a candidate may be exactly what party insiders want: someone who can unite the party and win in November.

This is a plausible story in some respects. In particular, it coincides with the finding that polls are not very predictive until quite late in the nomination race and even then can undergo dramatic shifts in the span of weeks or days. Voters usually like several of their party’s candidates; it may not take all that much to nudge them from one candidate to another. There are also reasons to be skeptical, however. For example — perhaps especially in the Republican Party — there has been an erosion of trust between party elites and rank-and-file voters.

For a long time, this seemed to be the question that the Republican nomination would turn upon. Trump seemingly had plenty of support from voters, but almost none from party elites, making him the “perfect test case.” How much power did the party really have? Would voters continue to support Trump once the party threw its whole playbook at him?

But just when it looked like we were about to get some answers, a funny thing happened on the way to Des Moines.

Party elites haven’t been doing much to stop Trump

It became clear a month or so ago, if it hadn’t been already, that Republicans didn’t have much of a strategy for stopping Trump. In fact, other than occasionally tsk-tsking at some of his more inflammatory remarks, they weren’t doing much of anything about him. They weren’t waging a concerted negative campaign; there have been remarkably few negative ads of any kind against Trump. But party elites also weren’t throwing their support to any of the other candidates, at least judging by those candidates’ still-lackluster pace of endorsements.

Recently, the race took an even stranger turn. There were stories like this one, from Philip Rucker and Robert Costa at The Washington Post, suggesting that party elites were warming to Trump. Soon after, Bob Dole was suggesting that Trump wasn’t such a bad guy, while Iowa Sen. Chuck Grassley was appearing with Trump and urging voters to “make America great again.”

Importantly, these actions seem to have been taken mostly in opposition to Ted Cruz, instead of in support of Trump. Nonetheless, these reports caused me to renounce much of my remaining skepticism of Trump’s chances. Even tactical and tacit support for Trump is remarkable from the “Party Decides” perspective because the book suggests he’s just about the last person party elites would want to nominate.

I mentioned the main reasons for this before: Trump scores poorly on the two dimensions — electability and ideological fit — that party elites are supposed to care most about. Maybe you could make a devil’s advocate case on Trump’s behalf, but I’m not sure how convincing it would be.11

Moreover, Trump is not the sort of candidate to whom you’d expect the party to extend the benefit of the doubt. Under “The Party Decides,” parties are supposed to prefer candidates who are acceptable to as much of the coalition as possible to those who are polarizing. Trump generates considerable enthusiasm among some Republican groups but strong opposition among others. Party elites tend to prefer candidates who have worked their way up through the system and developed a network of relationships within the party. Trump, a relative newcomer to Republican Party politics, worked around the system instead.

What’s more, Trump has touched on any number of “third rail” issues, from banning Muslims from entering the United States to denouncing super PACs, that Republican candidates usually avoid. This is part of Trump’s appeal, of course: He says what other candidates won’t say, but which may nevertheless be popular with Republican voters. But Republican candidates usually avoid mentioning these topics for good reason12: They tend to expose the seams in the Republican coalition — splitting the base from the “donor class,” dividing some Republican constituencies from others, or damaging the party’s brand for the general election.

Perhaps you can argue that Cruz is just as bad as Trump from a “Party Decides” standpoint. Cruz is far enough to the right that he could cost the GOP points in the general election. He’s such a purist, in fact, that he also might be too far to the right for some groups of influential Republicans, such as those who support government subsidies for ethanol. Furthermore, Cruz is perceived to be difficult to work with. But there’s no reason party elites can’t oppose Cruz and oppose Trump just as vocally.

One explanation could be that party elites are misinformed or confused about Trump. “The Party Decides” tends to assume that party elites are highly sophisticated — able to see past the spin, the non-predictive early polls and the media talking points of the day. But perhaps this isn’t the case. Party elites often have relatively little communication with rank-and-file voters and may not understand the reasons for Trump’s popularity because they don’t encounter very many Trump supporters. At the same time, they exist within the political echo chamber and are inundated with constant media chatter about Trump’s polls and momentum. The party elites may even be engaged in a self-fulfilling prophecy of sorts: Because everyone thinks that Trump is impervious to attack, no one is bothering to attack him.

Have Republicans lost their team spirit?

Maybe the most incredible passage of the campaign cycle comes from a recent Jonathan Martin article in The New York Times. It suggests that some Republican professionals are supporting Trump because they think he’ll lose:

Of course, this willingness to accommodate Mr. Trump is driven in part by the fact that few among the Republican professional class believe he would win a general election. In their minds, it would be better to effectively rent the party to Mr. Trump for four months this fall, through the general election, than risk turning it over to Mr. Cruz for at least four years, as either the president or the next-in-line leader for the 2020 nomination.

I’m a bit skeptical, but Martin seems to be referring to Republican lobbyists and consultants, in which case the reporting makes a certain amount of sense. If Trump wins the nomination but loses the general election, we’ll have another extremely vigorous competition for the Republican nomination in 2020, which means lots of work for consultants.13 Lobbyists and consultants would stay busy with either the transactional Hillary Clinton or the wheeler-dealer Trump as president, but less so with an ideologue like Cruz.

Other types of party elites have their own incentives. Republican members of Congress apparently think they’ll do worse with Cruz on the ballot than with Trump. I haven’t seen much evidence to support this claim (or much evidence against it), but so long as the members believe it, you might expect their self-preservation instincts to kick in.

The other Republican campaigns, meanwhile, may have tactical reasons to avoid attacking Trump even as they pillory one another.

The theme is that many Republican elites have no professional incentive to oppose Trump even if they personally dislike his politics or think he’d be a poor nominee. That’s fair enough. But the whole point of forming a party is to work together to facilitate the party’s interests. In that sense, the GOP would qualify as a weak, fraying party if it can’t avoid nominating Trump, a candidate who might at once reject large parts of the party’s traditional platform and potentially cost it a highly winnable general election.

That’s not to say the Republican Party would disappear after a Trump nomination — there would almost certainly still be something named the Republican Party — but it could conceivably be transformed into something more in Trump’s image, perhaps more in the direction of a European populist party. Trump’s nomination could even trigger a political realignment. Such things are rare, occurring perhaps once every several decades, but nominees like Trump are awfully rare, too.

Or maybe not. “The Party Decides” isn’t wrong, not quite yet. While most of the book’s focus is on the “invisible primary” phase of the campaign, it also presents statistical evidence that party elites continue to exert influence on the outcome well after Iowa and New Hampshire have voted.14 Furthermore, on the previous three occasions when party elites failed to reach a consensus before Iowa (these were the 1988 and 2004 Democratic nominations and the 2008 Republican race), the parties nevertheless wound up nominating fairly conventional candidates (Michael Dukakis, John Kerry and McCain). If Marco Rubio winds up the Republican nominee after all, the theory will come out looking pretty good. And if it’s Jeb Bush, somehow, the party’s powers will seem miraculous.

Check out the latest polls and forecasts for the 2016 presidential primaries.


  1. I encourage you to read “The Party Decides” yourself. Failing that, this interview with co-author Hans Noel at the Columbia Journalism Review is a good introduction to the book. So is this paper, “A Theory of Political Parties,” which was written by the book’s four authors and two other political scientists, Kathleen Bawn and Seth Masket.

  2. Other endorsements are potentially relevant — we’re just not keeping track of them. Maintaining our list of 585 potential endorsers is more time-consuming than you might think. Expanding the universe to tens of thousands of potential endorsers would be more difficult still.

  3. ”The Party Decides” is relatively silent on the question of whether explicitly liberal and conservative media outlets are part of the Democratic and Republican parties. I agree with Jonathan Bernstein that they probably belong within the book’s definition, however.

  4. Above and beyond merely voting; “The Party Decides” would not consider someone to be a party elite if she voted in a caucus or primary but engaged in no other partisan political activity.

  5. I’m picking on Republicans here, but Democrats have an equally broad array of concerns.

  6. This refers to the 10 competitive nominations between 1980 and 2004.

  7. Best practices in a small-sample environment usually involve two things:

    • You should use all the data available and keep the statistical analysis fairly simple to avoid overfitting and multiple hypothesis testing (“p-hacking”).
    • You should be making claims that have a high underlying degree of plausibility (claims which have strong Bayesian priors). Ten or 12 cases might be enough to tip the scales between “plausible” and “probable.” They probably aren’t enough to prove some extraordinary claim, however.

    For the most part, “The Party Decides” does a pretty good job with these, in my view. My main objection is in how the authors tend to treat 1972 and 1976 as transitional cases that aren’t all that relevant to the current party system.

  8. I say “arguably” because the candidate who’s the best ideological fit for his party isn’t necessarily the most liberal or most conservative one. Sanders is too liberal for many Democrats, while Cruz is too conservative for many in the GOP.

  9. Carter finished with more votes than any other candidate in Iowa, although “uncommitted” took the plurality.

  10. And that power may be diminishing in an era when a single rogue donor can contribute tens of millions of dollars to a super PAC.

  11. On electability, it’s true that Trump was able to transform his image among Republicans and that he’ll probably move toward the center if he wins the nomination, but he’ll start out at a disadvantage: His favorability ratings among the electorate as a whole have been remarkably steady and remarkably negative.

    On ideology, it’s true that you can turn Trump into a sort of moderate Republican if you squint, but only if you average out some extremely conservative positions with a few outright liberal ones. That’s not usually the sort of candidate whom party elites prefer; they’d rather have someone who’s within the party orthodoxy on every issue than someone like Trump or Rand Paul who frequently misses in either direction. I’ve also seen Trump’s ideological flexibility cited as a selling point in some of the stories about the party warming to him; he’s someone Republicans think they can cut a deal with. Under the “Party Decides” view, however, parties want candidates with predictable policy positions whom they can “trust to safeguard their most intense concerns.” That Trump is neither all that conservative nor all that predictable makes him doubly problematic.

  12. Or dog-whistle to voters rather than mention them explicitly.

  13. Trump also might invite a primary challenge in 2020 even if he wins the general election this year.

  14. This matches my own research for our primary forecast models.

Nate Silver founded and was the editor in chief of FiveThirtyEight.