Do presidential debates really swing elections?

After giving Hillary Clinton an 89 per cent of chance of winning the US Presidential election on 14th August, the respected statistical website "FiveThirtyEight.com" started revising down its forecast. A few hours before the first presidential debate between Clinton and Trump last night, her chances were at a meager 54 per cent. Reporters therefore not only considered the presidential race to be extremely close, but speculated about what might tip the outcome Trump’s way. They settled on the easy answer: the debates, of which two more are scheduled for October. The conventional wisdom is that one candidate can guarantee victory at the polls with a well-timed (and carefully rehearsed) answer, or see his/her campaign going to the sewer if he/she stumbles. By Monday night, the expectation for a “game changer” event was through the roof.

And some pundits believe that there’s been one: the vast majority appreciated Hillary’s “calm, poised” performance and her detailed explanations. Her “command of facts” was praised. Media sympathetic to her went as far as declaring Trump’s performance a “meltdown.” A CNN poll of debate-watchers showed that 62 per cent of voters thought Clinton won, compared to 27 per cent for Trump. Now the expectation is for a “bounce” in the polls.

Unfortunately, things are not so simple.

In the presidential campaigns between 1960 and 2000, political scientist James Stimson found little evidence of “reversals of fortune” because of presidential debates. A larger study by political scientists Robert Erikson and Christopher Wlezien, including every available poll between 1952 and 2008, came to a similar conclusion: the candidate who is ahead before the debates remains ahead, the one who is behind remains behind. Erikson and Wlezien conclude that evidence of any “debate effect” is “fragile.”

Consider the first debate between John F Kennedy and Richard Nixon in the Autumn of 1960: Stimson, Erikson and Wlezien found that Kennedy’s margin of victory in the end was only slightly higher than his lead in the polls on the eve of the first event, suggesting that the debates merely consolidated a pre-existing trend in Kennedy’s favor. The one in 1960 was a close election (Kennedy won by a razor-thin margin) but we have no firm evidence that says the debates were crucial.

In 1980, the only debate between Jimmy Carter and Ronald Reagan occurred a week before the election, when Reagan led by roughly 2-points. The debate seemed to matter: the latter, having been an actor, could muster a performance. But more fundamental factors were in play against Carter’s re-election, and the debate simply nudged Reagan further toward victory.

The election of 2000 seems to provide us with a clear case of a small, but consequential, debate effect. Al Gore’s performance in the first debate was widely criticised and was considered by many pundits a blunder. After the debate, there was a swing of 2 or 3 points toward Bush, enough to give him a narrow lead in the polls. However, in the end Gore collected over half a million popular votes more than George Bush, and he lost the election only because of the baroque and antidemocratic mechanism of the electoral college, coupled to an unprecedented, and even more antidemocratic, intervention by the Supreme Court.

Why have presidential debates been limited in their political impact? One factor is that the candidates are often extensively prepared. This was evident in the bland start on Monday night: in the first 30 minutes it was plain that each candidate had read a thick stack of briefing papers and rehearsed carefully his or her lines of attack. Clinton, but also Trump, stuck to the message crafted by their handlers: later in the event the latter, of course, let himself loose.

Clinton couldn’t be easily rattled but each candidates’ arguments were immediately countered by the other. Clinton appeared more comfortable than Trump, but the difference in their respective performances appeared to me rather small. The Democrats can rejoice, but half of voters are likely to side with the aggressive, unconventional, sometimes rambling Republican candidate. And this ties into the second reason why the debates don't tend to change minds.

In a politically polarised country there are two audiences, not one. In the United States of 2016, real independent voters would probably qualify for protection under the Endangered Species Act. All the others root for “their” candidate, which means that they hear only things that confirm the opinions they already have. The split-screen format used by many networks adds to this effect: Republicans watched only the nervous Trump and Democrats were transfixed by smiling Clinton.

Sure, both had their good and bad moments: Clinton drilling her opponent about his still-secret tax returns scored points, but Trump hitting her on trade agreements like NAFTA, signed by Bill Clinton, and Obama-supported TPP, was cheered by a substantial part of the audience. This doesn’t mean that these rhetorical blows convinced undecided voters: if they had an effect at all, it was to fortify the positions already chosen by partisans.

Now, the question is: in a tight race, which camp will be more successful in mobilising its supporters and bringing them to the polls, particularly in the so-called swing States? The answer will come on November 8th.