Donald Trump and Hillary Clinton engaged in the final scheduled debate last night, passing a milepost marking the home stretch of the campaign. The overall dynamics and candidate strategy on display last night closely resembled those of the first two debates. Clinton was once again well-prepped and bent on goading Trump into counterproductive responses on his main points of vulnerability. Trump was once again extemporaneous and free-associative, focusing on broad themes instead of policy details.
It's hard to avoid the conclusion that Trump has been badly hurt by the debates. On September 26, the day of the first debate, the FiveThirtyEight model estimated that Clinton was leading Trump by 46 percent to 45 percent in the national popular vote and by 278 to 260 in the electoral vote, with just a 55 percent projected chance of winning the election. Today, Clinton is estimated to hold a lead of 50 percent to 43 percent in the popular vote and 343 to 194 in the electoral vote, adding up to an 87 percent chance of victory on November 8.
This shift no doubt partially reflects other developments that have occurred over the period that the debates were held—especially the Billy Bush tape and subsequent accusations against Trump. And Clinton was already the favorite to win the race even before they occurred. But I think the bulk of the evidence points toward the debates having a significant independent effect on the relative standing of the candidates, especially because they generated negative news coverage of Trump that persisted for days after the events themselves. Trump's refusal last night to commit to respecting the outcome of the election is likewise poised to dominate this week's coverage—to his further disadvantage—which means that the debates may continue to hurt his chances further over the next several days.
If true, the power of the debates to shift public opinion is another way in which the 2016 election departs from the usual pattern. Though debates receive a lot of attention every year, and media figures always spend a lot of time explaining which candidate "won" each face-off, previous research had concluded that the effects of debate performances on the horse race tended to be quite temporary when they existed at all. If Clinton winds up winning the election by a margin comparable to her current lead, we may regard the debates in retrospect as significant events in the trajectory of the race.
Political scientists have taken our share of lumps this election from our critics, largely because most of us didn't expect the Republican Party to nominate Trump (a conclusion which, to be fair, we were hardly alone in reaching). One recurrent point of difference between political science and popular media is that many journalists and pundits tend to interpret electoral outcomes as mostly reflecting the different personalities and strategies of the candidates, while political scientists more commonly emphasize the role of fundamental factors like partisanship and economic performance in shaping the choices of voters. (This view is sometimes caricatured as a belief that "campaigns don't matter," which no political scientist I know has ever claimed.)
If the debates are revealed to be a major factor in determining the vote margin in the 2016 election, however, it's fair to point out that campaign effects turned out to be bigger than some of us assumed. When analysis built on investigation of previous elections fails to hold in a new case, there are three possible explanations:
1. The analysis was flawed even when applied to previous cases.
2. The analysis was sound in the past, but the current case doesn't fit because the world has changed—and future cases will resemble the current case more than past cases.
3. The analysis was sound in the past and will be again in the future, but the current case represents a temporary deviation from the long-term norm.
Our least charitable critics will probably argue that (1) is likely to be true—we just don't know what we're talking about and never did. But as even they must acknowledge, there's plenty of reason to believe that this particular election is just different from previous elections, with Trump's nomination either a cause or a symptom of this difference.
We can even come up with very plausible hypotheses about why the debates would matter more for a candidate like Trump than a candidate like Mitt Romney or John McCain: perhaps voters had less confidence in Trump's ability to do the job of president, rendering a substandard debate performance all the more damaging to his chances. Or maybe the press coverage of Trump has been much more negative than previous candidates over the same period. Alternatively, though it's tempting to rely on a Trump-centric explanation to account for everything that's distinctive about this election, maybe the debates mattered more because of something unique about Hillary Clinton. Perhaps her relatively strong debate performances helped her consolidate the support of younger voters and independents who never liked Trump but preferred Bernie Sanders to Clinton and were won over once the debates focused their attention on the choice before them this November.
The bigger challenge is to distinguish condition (2) from condition (3). In the heat of a campaign's final weeks, when it's very hard to step back and gain a broader perspective, we often assume that the current state of the world will pertain indefinitely into the future—for example, that Trump's particular brand of politics is here to stay in the Republican Party even if he loses the election. Sometimes that's right, but sometimes we're simply in the midst of a temporary departure from the usual order of things. Until we are able to gain the benefit of experience—or, as political scientists might put it, more data—we won't know for sure how much of what's extraordinary about the 2016 election is merely a product of the moment, and how much is a foreshadowing of the new normal.