The Impact of Debates? It's Debatable

With the Hillary Clinton-Donald Trump debates upon us, the quadrennial question comes begging: Do these showdowns matter?

The chances for impact seem ripe this year. The two most unpopular major-party candidates in the history of ABC News/Washington Post polls are facing off. Donald Trump's unorthodox campaign style and provocative positions have piqued public interest; the debates may be the single best opportunity for him to allay concerns about his qualifications, temperament and policy promises alike. Hillary Clinton, for her part, needs -- like Trump -- to ease questions about her trustworthiness, as well as to spark greater enthusiasm among her supporters.

Can either candidate move the needle? It will be tough: A review of data since 1960 suggests that past debates have almost never directly and measurably changed the candidates' relative standings. That's admittedly a high standard, though, and at least some debates may have had more subtle impacts.

Debates have been held in 11 presidential races since 1960. We find just one after which the lead in presidential preference changed hands by a significant margin: In 1980, when Ronald Reagan uttered his "are you better off?" line. He gained 7 points in a post-debate poll.

Still, there are other cases in which debates (or post-debate evaluations) may have influenced campaign dynamics. Debates are, after all, an essential window on the candidates' styles and grasp of the issues. They can seem to add or reverse momentum. And polls are only so precise.

Consider 1960: After Richard Nixon's pasty-faced performance, John F. Kennedy went from 46 percent support to 49 percent in Gallup data; Nixon, from 47 percent to 46 percent. Those changes were within the polls' margin of sampling error. Nonetheless, the numerical "lead" switched, and collective memory maintains that the debates spelled Nixon's narrow defeat.

Nor did Gallup polls show significant movement immediately after the 1976 debate in which Gerald Ford said Poland was free (and news reports pounced on the misstatement); Jimmy Carter gained a single point, Ford lost 3. Still, even if not measurably, the gaffe may have halted what had been a slide in Carter's lead.

Then 73-year-old Reagan's teasing jibe in 1984 about 56-year-old Walter Mondale's "youth and inexperience" didn't meaningfully change the numbers in ABC News/Washington Post polls, but perhaps helped put aside questions about Reagan's age and acuity. In 2000, Al Gore's wonk-heavy performance saw him go in at +2 and leave at -3, but again, that was within sampling error.

There's also the question of missed opportunities, as in Mike Dukakis' bloodless answer, in 1988, to what he'd do if his wife were raped and murdered. He didn't seize on the chance to show a non-technocratic side. And ABC/Post polls did not meaningfully move.

ABC News/Washington Post polls in 1996, 2004, 2008 and 2012 also showed no significant movement around the debates. In 1992 there was more movement, including a significant gain for Ross Perot, but never enough to change Bill Clinton's advantage.

In instant reactions, at least, polling has found that debates typically reinforce preconceived notions rather than change them. Most of each candidates' supporters say it's their candidate who won, and precious few say their minds were changed.

But post-debate evaluations can shift. In 1992, right after the first debate, 24 percent of viewers said Perot had won. By the next night, amid positive reviews of Perot's performance, that grew to 37 percent among people who'd watched, heard or read about it. And Perot's support did advance, from 6 percent before the debates to 17 percent after them.

That shift may have been real, but it also may have reflected the challenges of immediate post-debate reaction polls. Conducted in mere minutes, among individuals pre-recruited to participate, they're subject to considerable sampling and weighting limitations. Further, some ask who "won," others who "did the better job"; some accept "tie" as an answer and some don't.

Instant reactions also bypass the role of considered judgment. Hard as it may be to believe in the Twitter Age, sometimes people actually need a little while to think about things. The clearest response to the debates this year may take some days to emerge.

Rundown

Here's a rundown of the measurable debate effect (or lack thereof) in each election since 1960. There were no debates in 1964, 1968 or 1972.

Click here for data tables associated with each election.

1960: Gallup had Nixon +1, but that was 12 days before the first debate; it had Kennedy +3 after it. Twelve days is a long time and 4 points is a small number. Gallup didn't poll between the remaining three debates, but showed Kennedy +4 after the last one.

1976: News reports jumped on Ford's misstatement of Soviet domination of Eastern Europe in the second debate. Carter's lead had been diminishing, and that did stop after this debate. But the measured change after the second debate was tiny -- Carter gained 1 point, Ford lost 3.

1980: This one looks to have mattered: Gallup's pre-debate poll had Carter +8; post-debate, Reagan +3. (John Anderson did not participate.)

1984: Mondale crept up, but Reagan stayed ahead by double digits nonetheless.

1988: We don't have an ABC poll done right before the first debate; in a Gallup poll it was George H. W. Bush +8. Our poll after the second debate showed no meaningful movement.

1992: Perot moved up 5 points in ABC News polling after the first debate and 4 points after the third debate. Before the debates he had 6 percent support; after the last debate he was up to 17 percent. Clinton moved up by 5 points after the second debate, Bush down by 6, then Clinton down by 6 after the third. But Clinton led throughout.

1996: The race looked perhaps slightly tighter after the second debate, but Clinton maintained a double-digit lead throughout.

2000: Gore went in +2 and came out -3 -- a change of lead, but within sampling error. And a week after the last debate it was back to a dead heat, 47-47 percent.

2004: There were some wiggles in the debate period -- a 6-point lead for George W. Bush before the first debate was a dead heat before the third -- but when all was said and done the race after the debates looked a lot like the race before them.

2008: The main shift in vote preference came before the first debate, as Obama seized the reins of economic discontent after the failure of Lehman Brothers in mid-September. McCain's +2 in early September, just after the GOP convention, proved to be his best. The race shifted to Obama +9 in an ABC/Post poll completed Sept. 22, four days before the first debate. Obama never trailed again.

2012: Obama was +2 ahead of the first debate in ABC/Post polls, within the margin of sampling error, +3 after the second debate and -1 after the final debate, all within sampling error.