The first presidential general election debate is on Monday. And with the polls as tight as they are, some Democrats and liberals are feeling a gnawing sense of anxiety that these upcoming debates could throw the election to Donald Trump.
But that gets to a bigger question political scientists have been grappling with for many years: How much do the debates even matter, anyway?
The evidence isn’t entirely conclusive, but in my read of it, debates have the potential to make a small but real impact on the race. Polls have often shifted by a few percentage points during debate season, and in a close race, that could really matter. (By the way, Hillary Clinton’s lead over Donald Trump in the RealClearPolitics polling average has now shrunk to less than 3 percentage points.)
Now, the effect of general election debates has been overhyped by some. There’s little historical evidence that they’ve ever swung polls by more than a few percentage points. General election debates aren’t like primary debates — there are strong partisan loyalties, the vast majority of debate viewers have already made up their minds about who they’re voting for, and few are willing to change their minds because of what happened in one debate.
But, in a close race, with a very polarized electorate, a shift of just a few percentage points could matter a great deal. And even if debates don’t swing the presidential outcome, if they help or hurt a presidential candidate by a few percentage points, that could have a domino effect in down-ballot races — such as the battle for the Senate.
How the polls have moved during past debate seasons
Again, the potential of debates to completely shake up a presidential general election campaign has often been exaggerated. Political scientist John Sides had a good, thorough rundown of the relevant evidence in 2012, and he found that many supposed game-changing debate moments had a relatively minor impact, if any.
The data shows overall that big changes in the polls during debate season just haven’t happened in recent decades — but it also shows that small changes, of a few percentage points or so, are common. Here’s the key chart, from Robert Erikson and Christopher Wlezien’s book The 2012 Campaign and the Timeline of Presidential Elections:
Essentially, the x-axis here plots the Democratic nominee’s polling numbers a week before the first debate, and the y-axis plots his polling a week after the final debate. The further away a point is from the diagonal black line in the middle, the more the polls changed during the debate period.
And you see that they rarely move very much. The past nine presidential elections —from 1980 to 2012 — are all quite close to that diagonal line, meaning that polls had only shifted a few points on net by the time debate season ended.
Only in 1976 did the polls look dramatically different at the end of debate season, as challenger Jimmy Carter’s large lead over incumbent president Gerald Ford shrank to only a small edge. But that seems to have had little to do with the debates — Carter’s lead had been shrinking before the debates started, and it was in fact Ford who is said to have made a famous gaffe in the debates (by saying there was "no Soviet domination of Eastern Europe").
So, we haven’t seen major poll changes that are attributable to the debates. But that doesn’t mean we’ve seen no change. The chart shows that it’s actually quite common for polls to shift by two to three points during debate season.
Now, there’s one caveat: It’s probably not always the debates themselves that are moving polls here. Many other things happen between the start of debate season and the end. For instance, Barack Obama’s bump of 3 points or so during the 2008 debate season may have owed more to the unfolding economic crisis than his debate performances.
Still, some effects are pretty clearly from the debates themselves — as in 2012, when President Obama’s 4 point lead in the polls abruptly vanished after what was judged as a weak performance in his first debate against Mitt Romney. Obama regained a bit of ground after subsequent debates, but polls showed a very tight race from then on out.
And in two famously close elections — 1960 and 2000 — it’s quite possible that debates played a role in swinging the outcome. Both Richard Nixon and Al Gore’s debate performances were panned by pundits for stylistic reasons (Nixon sweated, Gore sighed). Afterward, each candidate seems to have lost a couple of points in the polls (though the relative lack of polling in 1960 makes it tough for us to say that with certainty, according to Sides).
Overall, it seems that debates likely haven’t changed the outcome of any recent election except, perhaps, for 2000, which was so close that any number of things could be said to have made the difference. Still, it’s certainly possible that a movement of a few points could tip a close election in the future.
The Trump factor
But of course, it’s far from clear how much of these historical trends will still hold true when one of the two major party candidates is Donald J. Trump.
In most of these past debates that have produced minor swings, the debaters have been relatively ordinary mainstream politicians, who have generally prepared well and had poll-tested messages. When two mostly evenly matched opponents like that are pitted against each other, it makes sense that the debates wouldn’t make a gigantic impact.
Yet Trump could be a very different animal. And that could hurt him — or, conceivably, even help him.
Perhaps the short-tempered, irritable developer will lose his cool. Perhaps voters will perceive him as remarkably ignorant next to Clinton, or he’ll be extremely unprepared. Perhaps the media will be quick to pounce on any misstep by Trump as a devastating gaffe.
But it’s easy to imagine the opposite happening, too. Perhaps expectations for Trump are so low that the media will deem even a mediocre performance a win for him. Perhaps moderators will roll over for Trump like Matt Lauer did in NBC’s candidate forum. And perhaps Trump will manage to come off as relatively "ordinary," and win some hesitant voters to his side.
The media spin war over the debate could matter as much as the debate itself
Furthermore, though the debates themselves will last for just 90 minutes, the spin war in the media over who won will last for days. And some political scientists have found that watching media coverage of the debate can do a great deal to shape perceptions about what actually happened.
Now, depending on how high your estimation of the media is, you might think that the media’s view of what happened will bear at least some resemblance to what actually happened. But the media often chooses to pull out and focus on just a few "moments" of a 90-minute long event. And groupthink can be a powerful force, particularly when journalists are all reading each other’s Twitter feeds.
Alternatively, different media outlets could come to different conclusions about what happened, and that can make an impact on viewers. Kim Fridkin and several co-authors at Arizona State University conducted an experiment around a 2004 debate between George W. Bush and John Kerry. They screened the debate alone for one group of people, they showed the debate plus 20 minutes of NBC commentary to another group, and they showed the debate plus commentary on CNN.com for third group.
Overall, participants who watched the debate alone, or watched it and read CNN.com commentary, were more likely to think Kerry won. But participants who watched the debate plus NBC’s analysis afterward were overwhelmingly likely to conclude Bush was the winner:
Now, this just relates to the question of who "won" the debate. Obviously, viewers can conclude that a presidential candidate won a debate without deciding to vote for him or her.
Still, the question of how the media judges the performances of Clinton and Trump could turn out to be very important indeed. If the media judges Trump by extremely low expectations, or if his outrageous conduct is normalized, that could really affect how some viewers understand what happened.
Overall, if Hillary Clinton were still leading Trump by 9 percentage points, then she and her supporters could feel confident that the debates would be highly unlikely to change that. A lead of less than 3 percentage points is a different story, though.