‘HOW TRUMP WON’ blares the headline on this week’s Time magazine cover in 80-point Duplicate Ionic. ‘Now he just needs the votes,’ whispers the small subheadline underneath. Oh, just that little detail? Trump actually needs people to vote for him?
I’ve been encountering a lot of this lately: Coverage implying that Trump’s lead atop the Republican polls (which he’s held since mid-July) is a watershed event, perhaps even tantamount to his having won an election.
These headlines, needless to say, are presumptuous. National polls, even with barely more than three weeks to go until the Iowa caucuses, aren’t highly predictive of the eventual outcome of nomination races. I know, I know: You’ve heard this spiel from us (and others) before. In fact, if you read this website regularly, you’ve been hearing it for nearly as long as Trump has been atop the polls.
To our credit — and perhaps to our ultimate detriment — we’ve at least been fairly consistent about this. If you’re not convinced about the empirical validity of national primary polls in the first place, having more of them doesn’t necessarily persuade you, any more than someone with a shellfish allergy is poised to take advantage of Red Lobster’s Endless Shrimp promotion.
But if Trump’s polls don’t quite mean what the headlines imply, what do they mean exactly? And even if you take them with a dose of skepticism, could he win the nomination anyway? (Spoiler alert: Yes, although I continue to think Trump’s chances are lower than where betting markets put them.)
Before we all decamp for Des Moines next week, let’s take one more theoretical look at the Republican race. In fact, we’ll take three of them. As far as I can tell, there are three main theories about the rise of Trump; respectively, they pin the credit (or blame) for Trump on Republican voters, Republican elites and the news media. Each theory interprets his polls differently and comes to different conclusions about how much staying power he might have in the new year.
Theory 1: Trump’s support reflects a Republican populist revolt: In a nutshell: Trump is extremely popular among Republican voters, who are attracted to his combination of populism, nativism and anti-elite resentment.
What it makes of the polls: It loves the polls, almost as much as The Donald himself does! It takes them enough at face value to assert that Trump is the leading candidate for the Republican nomination. Furthermore, it treats Trump’s persistence in the polls as evidence that his popularity is authentic and not just a passing fancy.
Where you’re hearing the theory: You’re hearing it everywhere — this view underlies most of the mainstream coverage of Trump’s campaign. (It’s almost never stated as a theory; instead, it’s usually taken to be self-evident that Trump’s polls are evidence of his profound popularity among Republicans.) This view is also popular among left-leaning media outlets, which usually take an unsympathetic view toward Republican voters and often attribute Trump’s popularity to economic and racial resentment among working-class white Republicans.
Strengths of the theory: For one thing, it’s the simplest explanation of Trump’s polling; there’s often a lot to be said for parsimonious explanations as compared with more complicated ones. For another, it can invoke data about how a number of Trump’s positions — for instance, his proposed ban on Muslims entering the United States — are fairly popular among Republican voters. It can cite evidence about how Republicans want an ‘outsider’ nominee, and can cite numerous instances from gubernatorial, Senate and House primaries over the past six years when insurgent candidates knocked off establishment alternatives. And it can point toward some evidence of Trump’s popularity apart from polls, such as crowd sizes and the unprecedented ratings for televised Republican debates.
Weaknesses of the theory: This theory is a lot like that Red Lobster menu, seeming to present an endless array of options, but most of them are just the same limited palette of cheap ingredients reconstituted in different ways. In particular, the theory puts a tremendous amount of stock in the national polls despite their historical lack of reliability. What about those huge crowds and ratings that Trump is drawing? Clearly there’s a lot of interest and curiosity about Trump, but this theory tends to take for granted that this curiosity is associated with wanting to vote for Trump; that isn’t necessarily clear.
What the theory implies for 2016: It takes Trump’s persistence in the polls thus far as evidence that he’ll have a lot of staying power. ‘Trump’s voters have, for most part, been with him for months,’ as The New York Times’ Maggie Haberman puts it. ‘Caucusing is one thing but why wouldn’t they vote in primaries?’
How Trump could still lose, even if the theory is right: It’s possible to split the difference, tending to see the polls as valid and meaningful measures of public support but not necessarily as perfectly predictive. Furthermore, even if the polls are taken at face value, Trump’s position in the polls is good but not great: he seems to be trailing Ted Cruz in Iowa, and is potentially vulnerable in New Hampshire. Finally, his questionable ground game could prevent him from capitalizing on a theoretically high level of support.
Theory 2: Trump’s support reflects a Republican Party power vacuum: In a nutshell: Party elites and insiders usually have a tremendous amount of influence on the identity of the nominee, with Republican voters eventually falling in line behind one of their preferred choices. The fact that Trump is leading now reflects a lack of consensus among those party elites. However, these elites will rally behind whichever establishment-approved choice performs best in Iowa and New Hampshire, elevating that candidate to the frontrunner’s position.
What it makes of the polls: It treats them as provisional, pointing out that candidate preferences often shift considerably at the last minute.
Where you’re likely to hear the theory: From political scientists; this is a rearticulation of ‘The Party Decides’ view of the nomination race. You’ve also read a lot about it at FiveThirtyEight: Our August article ‘Donald Trump’s Six Stages of Doom’ was an exploration of the various barriers Trump would face in converting support in early polls into a party nomination. Finally, in-depth reporting, like this Washington Post article about Republicans making contingency plans for a contested convention, often speaks to this theory.
Strengths of the theory: It’s the most empirically grounded of the three, explaining most party nomination races since 1980 reasonably well. In past cycles, a number of nontraditional or insurgent candidates have performed well in the polls in the year before the election, but (hello, Howard Dean) failed to turn that support into votes or (howdy, Pat Buchanan, Rick Santorum and Jesse Jackson) had a ceiling on their support, winning an important constituency within their party without approaching a majority.
Weaknesses of the theory: Like any theory built around historical election data, it’s limited by its sample size — and in this case, the sample size is not all that large, because the current nomination process is fairly new and because presidential elections take place only once every four years. Furthermore, the theory has a mixed track record of success since first being published in 2008: It was very helpful in predicting Mitt Romney’s victory in the Republican race four years ago, for instance, but less so in anticipating Barack Obama’s upset over Hillary Clinton. What’s more, there are reasons to think that Trump could be an outlier, defying the premises of the theory even if the theory is generally sound. Finally, the theory suffers from a bit of an underwear gnomes problem, implying that Trump and other insurgent candidates will be stopped by party elites but not necessarily identifying the mechanism by which that occurs.
What the theory implies for 2016: It remains highly skeptical of Trump’s ability to win the nomination. It also implies that it’s still pretty early. Under our ‘Six Stages of Doom’ interpretation, Trump has overcome only his first hurdle while the remaining five still await. Because he’d be such a disastrous nominee, according to the theory, party elites will be fighting Trump at every step, possibly including preferring a contested convention to conceding the race to him.
How Trump could still win, even if the theory is right: It’s possible that party elites could fail to identify a successful strategy to stop Trump even if they have a lot of desire to do so. It’s also possible that the Republican party establishment is in disarray and lacks the power it once did.
Theory 3: Trump’s support reflects a media bubble: In a nutshell: Trump’s standing in the polls substantially reflects the disproportionate amount of media coverage he’s receiving; it’s not that remarkable for a candidate to poll at 35 percent when he’s recently been getting 70 percent of the media coverage of the Republican race. That makes Trump’s position vulnerable if media coverage eventually evens out, or as the election approaches in each state as voters learn more about the candidates on their own and less through the lens of the national media.
What it makes of the polls: See above.
Where you’re likely to hear the theory: From media critics, who recognize how much difficulty the press has had in understanding the Trump phenomenon. From supporters of other Republican candidates who are exasperated with how much coverage Trump has received. And from FiveThirtyEight: The first substantial piece I wrote about Trump this cycle, ‘Donald Trump Is The World’s Greatest Troll,’ explored the relationship between Trump and the media and how his media coverage could become a self-perpetuating cycle.
Strengths of the theory: Of the three theories, it seems to do the best job of explaining the movement in the polls to date. So far, the number of voters who list a candidate as their first choice has had a had a relatively low correlation with each candidate’s favorability ratings, but a strong correlation with the volume of media coverage the candidate is receiving at a given time. Any penalty Trump might suffer from controversial remarks like those about John McCain or Vladimir Putin are offset by an increase in media coverage, for example. The theory is also resonant with evidence suggesting that better-informed voters are less likely to support Trump and perhaps explains why Trump is underperforming in Iowa and New Hampshire (where voters are deeper into their information-gathering process) relative to other states. Finally, the theory is helpful in explaining the boom-and-bust cycles of past nomination races, as media coverage can perpetuate feedback loops in a candidate’s polling.
Weaknesses of the theory: It has a lot of trouble in differentiating cause and effect. Is Trump’s standing in the polls the result of intensive media coverage, or is the relationship the other way around? Relatedly, it can sometimes be guilty of treating media coverage as an exogenous variable (something which ‘happens’ to a candidate) when in fact it may reflect a candidate’s skill in manipulating the media and drawing attention to himself — a skill that a showman like Trump has in spades. Also, while the theory can explain some of Trump’s rise in the polling, it can’t explain all of it: Even if you ignore those first-choice numbers, his national favorability ratings have also tended to increase over the course of the cycle.
What the theory implies for 2016: Media coverage will probably even out as the field is winnowed; furthermore, voters will become less reliant on the national media as they prepare to vote in their respective states. Both those things could hurt Trump. Under this theory, the momentum that candidates receive from early states is liable to be especially important; Trump could fade out quickly after losses in Iowa or New Hampshire, or blow up into an even bigger story with wins there.
How Trump could still win, even if the theory is right: Even if Trump’s support is something of a media bubble, it’s plausible it won’t burst until many states have voted. Also, further ‘surprise’ events in the news cycle, like terrorist attacks in the United States or Europe, could work to Trump’s benefit.
So where does all of this leave us? In FiveThirtyEight’s coverage of Trump, we’ve tended to focus mostly on the second and third theories. But that’s largely because the coverage you’re reading elsewhere is so dominated by the first one. That doesn’t mean I’d dismiss out of hand the idea that Trump is authentically popular with the Republican base.
Instead, I try to take a mental average of the three theories. I think Trump’s more likely than not to underachieve his national polls once people get around to voting in the early states, but I don’t expect his numbers to fade to zero. Perhaps he has Buchanan’s constituency of 15 to 20 percent of the Republican electorate, plus another 15 to 20 percent of bandwagon voters that will join him or leave him at any given time. Those momentum-driven voters — plus how quickly the rest of the field consolidates — will determine how many wins he notches in the early going.
No matter how well Trump performs in the early states, Republican elites will do as much as they can to constrain him. But he could win the nomination if they fail to develop a good alternative to him, or if he rides a tidal wave of media-driven momentum after early victories — momentum is a more short-lived, but also potentially more powerful force than the gravity the party exerts.
Whichever theory of Trump you prefer, it’s votes and delegates in 2016 — not what the polls said in 2015 — that will be the real test of it.
 My heuristic throughout the campaign has been to assume that Trump’s chances are about half of what betting markets say they are. Betting markets currently put Trump’s chances at about 25 percent, so I think they’re about half that — 12 or 13 percent.
 Ever wonder what it’s like to live in Iowa just before the caucuses? The FiveThirtyEight politics team will be fanning out across the state next week to answer that question. Stay tuned.
 The ‘proof’ that Republicans want an ‘outsider’ nominee is largely based on the polls, for instance, but those polls may be a reflection of Trump’s campaign rather than a cause of his success.
Trump could lose New Hampshire either to a surging Cruz or if one of the several establishment candidates — Marco Rubio, Chris Christie, Jeb Bush, John Kasich — can consolidate the support of more moderate/establishment Republican voters.
 Personally, I think a Trump nomination would be extremely damaging to the ‘Party Decides’ theory, and I’m wary of the notion of defining him away as an outlier, but we’ll save that for another day.
 The second and third stages concern whether Trump translates high levels of initial support in national polls into votes in Iowa and New Hampshire — and there’s plenty of doubt about that, given the reporting about his ground game and that his numbers those states already lag his national ones.
 And if his controversial positions also happen to be popular among Republican voters — as is his proposed ban on Muslim travellers — he won’t just hold steady but will gain in the polls.
 Although, perhaps importantly, his favorability ratings in Iowa and New Hampshire are mediocre.