There is no rest for the weary. After nearly a year of the media’s mystical ascription to ephemeral and unreliable polling data, the 2016 presidential contest finally got some actual results this week. Yet the hard vote and delegate totals racked up in Iowa on Monday—with the Republicans bunched at the top with Ted Cruz at 27.6%, Donald Trump at 24.3%, and Marco Rubio at 23.1%, and Hillary Clinton narrowly edging Bernie Sanders 49.9% to 49.6% on the Democratic side—have already been re-filtered through the same murky prism as before. Instead of considering what the voters have actually told us, the media returned to speculating about what influence their own perceptions and coverage of the race will have moving forward. Their verdict: it’s more important for a candidate to exceed expectations than to deliver results. This conclusion brings the media back to the territory it knows and loves best: the close-minded world of inside sources, campaign turmoil, and donor angst. While this sorry state of affairs has been rightly pilloried by many as a perversion of the actual substance of democracy in order to further worship at the altar of process, the underlying point often goes unchallenged. As it turns out, the assumption that expectations are actually the key measuring stick for judging the early states is generally false; the media’s great expectations are not only regrettable but also forgettable.
There are three main problems with the expectations game. First, it completely ignores the fact that voters are often blissfully unaware or unconcerned with media expectations. This point really needs no explanation. Ask the average New Hampshire voter what Trump’s predicted vote share in the final Des Moines Register poll was, or the amount that Rubio over-performed his polling average, and you’ll likely receive a look as blank as fresh New England snow. Even informed voters are not always politics obsessives; they are usually aware of the high-level political coverage but not the trivial daily horse race. For that reason, it’s far more likely that voters will know how candidates performed than how surprised they should be at the result.
Second, expectations analysis overlooks the uncertainty inherent in primary polling, even when pollsters have performed well. All polls have a margin for error, which means that when a race is close, a poll can get it exactly right and yet be unequivocally wrong. As Nate Silver has shown, primary polling is particularly difficult to conduct due to a number of factors (including strategic voting, late decisions, and lower turnout). This means that the average primary poll misses the mark by approximately eight percentage points. Viewed in this light, it’s not entirely clear that Trump underperformed his actual level of support in Iowa, even if he did underperform his five-point polling lead heading into the caucuses. If that is so, then lower vote totals in later states may simply reflect his true potential and not the effect of negative momentum from missed expectations.
But third, and most importantly, the consensus that a candidate who exceeds expectations will get the biggest boost from an early primary state turns out not to be very effective in predicting future performance. This is particularly true once you control for the obvious fact that a candidate who performs well in a primary—regardless of whether that performance was predicted—will benefit that candidate elsewhere. The expectations myth survives in part because it proves true in one limited yet significant respect: unsurprisingly, the one area where media expectations have a meaningful effect is on media coverage itself. Even if the media’s expectations setting is ineffective, there is no doubt that campaign coverage is the oxygen every candidate needs to survive. For this reason, shocking performances, even if delivering overall results which are modest in comparison to those of other candidates, do have the potential to lift the obscure and ignored to prominence. Afterthought Jimmy Carter’s triumph over every other candidate in the 1976 Iowa Caucuses is one example. Senator Gary Hart’s “surprising, though distant, second place behind” former Vice President Walter Mondale, leading to his commanding victory in New Hampshire, is perhaps the best. Due to the media’s undeniable power to inform and misinform, a candidate starved for attention has to do what is necessary to be noticed. And if the media insists upon comparing performance to predictions, the slighted and ignored must play along. That, regrettably, often means devising a campaign strategy to accommodate media narratives. For this reason, Chris Christie and John Kasich really do need to exceed expectations and perform exceptionally well in New Hampshire to survive the inevitable narrowing of the media’s attention.
Setting this exception aside, however, there is surprisingly little support for the argument that unexpectedly good results alone—again, controlling for the fact that overall success naturally begets future success—is the crucial predictor of momentum. In other words, it matters if a candidate gets first place, but it doesn’t really matter how they got there. The following examples, by no means exhaustive, show this clearly. In the first election after Jimmy Carter put the Iowa Caucuses on the political calendar, George H.W. Bush scored a surprise upset over frontrunner Ronald Reagan in the 1980 caucuses (although interestingly, a computer glitch prevented about 6% of Reagan-favorable precincts from reporting their results, which, as a CBS News investigation discovered long after it mattered, meant that Reagan, not Bush, had narrowly won). Bush’s reward for upending expectations was a crushing defeat to Reagan by nearly 30 points. Eight years later, the shoe was on the other foot. Bush, by then the vice president, lost to Senator Bob Dole by a stunning 18%, with Bush sliding all the way to a “humiliating” third place behind televangelist Pat Robertson. A month later, however, it was Bush’s turn to score the comeback win in New Hampshire, while Robertson, Iowa’s biggest over-performer, was a mere afterthought. One of the most frequently cited examples of expectations-beating begetting momentum also fails to support the theory. In 2004, Senators John Kerry and John Edwards were in single digits in Iowa in December, far behind Rep. Richard Gephardt and former Vermont Governor Howard Dean, until a Des Moines Register poll picked up late momentum for the pair. Kerry and Edwards vastly outperformed even that survey, pulling in a stunning 38% and 32%, respectively. Kerry parlayed that victory into a large win in the New Hampshire primary. Edwards, on the other hand, came in a disappointing fourth in that state, with a mere 11% share. It was the winner, not the candidate who over-performed his polls by the greatest amount, who benefited the most. Actual success, not unexpected support, turns out to be the key to a surge in future states.
For the candidates—namely Trump and Cruz—who have already received more than their share of coverage, there is no evidence that the expectations baseline will affect their future support. That is not to say that the results themselves in Iowa are contained to that state. Voters, especially in primaries, vote strategically by taking cues from polling and previous results as to whom they should support among the many candidates they view favorably. Relatedly, and even more importantly, early caucus and primary results tell subsequent states which candidates are viable and worth consideration. The vast majority of people, even likely voters in New Hampshire and South Carolina, will only start to truly focus on the campaign as election day draws near. In other words, Rubio’s 23.1% (just 4.6% from victory) will give his campaign a significant boost, regardless of where he was polling beforehand. It also, however, does not outshine Cruz’s victory.
Instead of the frivolous exercise of measuring success by arbitrary and often negligent pre-election judgments of a candidate’s standing, the proper way to interpret primary and caucus vote totals is to examine who voted and how. Indeed, the Iowa results are significant for what they tell us about the preferences of different slices of the electorate. Although these details do not themselves create momentum, they do provide predictive power for other states with similar, or different, ideological and demographic makeups. In the Democratic race, Clinton supporters have argued that given the recent Sanders surge, her paper thin victory is still a win and thus a critical inflection point in the nomination race. On the other hand, the Sanders camp suggests that the virtual tie works to their candidate’s advantage given that Clinton had a small but steady lead in the Iowa polling average going into the caucuses. It seems far more important, however, to recognize the staggering demographic splits between the Clinton and Sanders factions and what they might portend for the future. Age was the single most predictive factor in the race; Clinton won 58% of voters aged 45 to 64 and 69% of voters 65 and older; Sanders won a remarkable 84% of the under-30 crowd. Sanders was the favored candidate of lower-income Iowans, while Clinton had her highest margins among the richest Hawkeyes. In addition, Clinton won 58% of moderates and a narrow majority of “somewhat liberal” voters, while Sanders took 58% of “very liberal” Iowans. And as expected, Clinton held a strong advantage among Iowa’s small non-white population, winning 58%. It is these differences, just as they did in 2008, that will determine who wins the remaining Democratic contests. It is why Sanders will remain a strong favorite among the white, liberal, and less affluent Democratic primary voters in New Hampshire, and why Clinton continues to be the best bet in the heavily African-American South Carolina. More broadly, this is why Sanders is unlikely to amass the delegates necessary to win nationally.
These demographic results are also exactly why, despite the current rush to bury Trump, the differences between the Iowa and New Hampshire electorates will be far more important in a week than Trump’s first-place expectation, second-place finish in Iowa. As conservative writer Reihan Salam recently argued: “Trump appeals most to working-class Republicans in rural stretches of the Deep South, Appalachia, and the Northeast, and in particular to those of a more secular bent. New Hampshire, then, is likely to prove far more favorable ground for Trump than Iowa.” That is certainly borne out by the results in Iowa, where Trump had the most support among the least educated, least religious, and oldest voters. Trump also won a large plurality of moderates and was tied for first among independents. The move east will benefit him, as New Hampshire is not only more moderate and less religious, but is also an “open” primary allowing independents and Democrats to participate (as long as they do not vote in the Democratic primary). Rubio, the candidate who will supposedly be most helped by Iowa, did best in larger cities and the suburbs, suggesting he may have trouble in the large rural swaths of the Granite State.
The narrative today is that “Rubio is the real winner in Iowa” (like a herd of cattle, that quotation comes from dozens of headlines from mainstream media organizations), while Trump is “just another loser.” Yet at last count, the two candidates received roughly the same amount of support in the caucuses, with Trump actually coming out ahead. While the fallout from Iowa is unclear, it is simply preposterous to claim that the rapid rise of Rubio in one state at the very end will overshadow everything else. Results are results, and candidates like Jeb Bush, Kasich, and Christie have a long way to go explain how their paltry support and finishes of sixth, eight, and tenth respectively give them a viable path to the nomination simply because no one expected them to do well. The emphasis on expectations is as laughable for them as it is when one of the few voters to back Jim Gilmore argued: “He exceeded expectations. I mean, look at some of the other candidates. Look at the money they spent. They got 3,000 votes. He didn’t spend any money and got 12.” Demographics, delegates, and actual, on-the-ground success will determine the presidential nominees. Can we count on the media to cover it that way? Don’t expect much.