Do they bounce, and if so, how high?
It’s the question of the next two weeks, as the presidential candidates stand beneath the bright lights and balloons of their nominating conventions. Candidate support can spike during these high-profile events, sometimes in an election-changing way.
It’s called the “convention bounce,” a phenomenon so common its measurement has been a staple of pre-election polling and commentary. But not all bounces are the same. Some have proved short-lived, the race quickly reverting to pre-convention form. Others have represented a profound coalescing of voter preferences that’s charted the course ahead.
Bounces this year could be confounded by the calendar. The parties are holding back-to-back conventions for the first time since 1956; they still could be sweeping up confetti from next week’s Democratic shindig when the Republicans convene in St. Paul four days later. Sandwiched in between is Labor Day weekend, when many Americans are apt to be more focused on end-of-summer cookouts than on presidential politics.
But if the compressed schedule potentially could suppress bounces, other factors could boost them. It’s an opportunity for Barack Obama to define the “change” at the heart of his candidacy and to answer questions about his experience and readiness. John McCain needs to answer questions about his age, vitality and readiness to make a break from the Bush years. Each wants to offer a more persuasive prescription for the economic woes at the heart of voters’ concerns in this election, as well as addressing the war in Iraq and the uncertain nature of international dynamics.
Today’s highly polarized politics could play a role; with party lines drawn especially sharply, there may be less room than usual for the candidates to move. At the same time, 28 percent of registered voters in the last ABC News/Washington Post poll said they hadn’t made up their minds for sure. That’s plenty enough to fuel a bounce, if the conventions warrant it.
Any way you cut it, presidential nominating conventions mark unique and highly fraught periods in the election cycle, when public attention focuses, candidates pass – or fail to clear – the basic bar of acceptability to a broader audience, and their support can undergo its biggest swings of the contest.
Convention bounces became apparent in 1968 (election polling was too infrequent for reliable conclusions before then), but the focus owes much to Bill Clinton and the Mother of All Bounces: He soared from a dead heat against incumbent George Bush before the 1992 Democratic convention to nearly a 30-point lead after it, and never trailed for the remainder of the race.
While no bounce has matched Clinton's, others have been impressive. Jimmy Carter rode a 16-point bounce to a 33-point lead after his 1976 convention, lending authority to his challenge and underscoring incumbent Gerald Ford's weakness. Ford in turn mustered just a 7-point bump following his own convention; though the race did tighten at the close, Carter’s better bounce foretold his ultimate victory.
The absence of any bounce can be a danger sign. Neither Hubert Humphrey nor George McGovern took significant bounces out of their nominating conventions in 1968 and 1972, both en route to their losses to Richard Nixon.
The fullest picture of the bounce can be drawn not by looking only at change in support for the new nominee, but at the change in the margin, to include any drop in support for the opposing candidate (offense sometimes being the best defense in politics). The 1968 Republican convention, notably, did more to reduce Humphrey’s support than to bolster Nixon’s.
Focusing on the margin, however, also accentuates small changes in poll results; for this and other reasons, such as timing and plain old vagaries in polling, bounce measurements can differ. In 2004, Gallup polls figured John Kerry's bounce from a starting point measured five days before his convention began, and assigned him a net loss of 8 points – its first negative bounce since McGovern's 32 years earlier. ABC and the Post started with a pre-convention measurement done four days later than Gallup's, and found an 8-point bounce in Kerry's favor, much nearer the norm.
Sometimes there’s been a pre-bounce, a lift for the candidate that begins even before his convention doors open; no such has been sighted this year. But it’s the event itself that seems most important: Candidates dominate political center stage for a week, working to lay out their vision and policies, burnish their credentials and – directly or through surrogates – undermine their opponent.
Other factors may contribute – outside events such as the Chicago riots of 1968 or the on-again, off-again Ross Perot candidacy of 1992. However, while conventional wisdom holds that Perot’s departure created Clinton’s surge, data from those days indicates that Clinton’s support actually jumped before Perot left the race – suggesting that Perot was more a casualty of Clinton's surge than its cause. Indeed if external events, strength of support, political polarization or economic discontent may be factors, atop the heap most likely stands the persuasiveness of the candidates themselves.
Exposure isn’t the sole cause; while airtime for network coverage of the conventions has declined sharply over the years, the bounces haven’t. The two national conventions received a total of 73 hours of broadcast network coverage in 1968, declining sharply in ensuing years to a low of six hours in 2004 (per "Vital Statistics on American Politics 2003-2004," CQ Press). Audience ratings likewise dropped. Yet there's no significant relationship between hours of network coverage and size of convention bounces. Indeed the largest bounce on record, Clinton’s, occurred in the modern era of less network news coverage – eight hours for his convention – while George McGovern’s bounceless 1972 convention was one of the most heavily covered, at 37 hours. (Cable and internet, of course, are handy outlets for those who can’t get enough convention coverage.)
Using the change in the margins, the average bounce has been 10 points among registered voters in Gallup polls from 1968 through 2004, and, for comparison, a similar 13 points in ABC/Post polls from 1992 to 2004. As noted, individual bounces vary, but the averages are consistent across a range of parameters: In Gallup data, 11 points for Democratic candidates (9 points leaving aside Clinton's 1992 bounce), 9 points for Republicans, 8 points for incumbents, 11 points for challengers, 10 points for better-known candidates (incumbent presidents and incumbent or former vice presidents), 10 points for lesser-known candidates, 12 points after each cycle’s first convention, and 9 points after the second convention.
While the averages by the candidate's political party are similar, more of the action has been among Democratic candidates – a standard deviation of 10 in their bounces (8 without Clinton's in 1992), compared with 4 in the Republicans'. The average Democratic bounce correlates significantly with the average bounce overall, while the average Republican bounce does not. Influences there may include the objective quality of individual candidates, a generally declining Democratic advantage in partisan self-identification (until 2004), and perhaps, more steadfast support among Republicans for their party's nominees.
The varying size and durability of convention bounces suggest that they’re not founded simply on the quantity of that week’s news coverage, but on more substantive evaluations of the content the parties and their candidates present. A focusing of the public’s attention may inspire the bounce, but a more deliberative judgment determines its size, staying power and ultimate impact.
Convention Bounces, 1968 to 2004* Dem Bounce Rep Bounce2004 Kerry +8 G.W. Bush +5 2000 Gore +16 G.W. Bush +9 1996 Clinton +5 Dole +15 1992 Clinton +30 G.H.W. Bush +16 1988 Dukakis +11 G.H.W. Bush +11 1984 Mondale +16 Reagan +8 1980 Carter +17 Reagan +13 1976 Carter +16 Ford +7 1972 McGovern -3 Nixon +8 1968 Humphrey +4 Nixon +14 *ABC/Post polls, 1992-2004; Gallup, 1968-88
(This piece is adapted from one I’ve written for the forthcoming Encyclopedia of Survey Research Methods from Sage Publications.)