skip to main content

The “Bounce” Game

Both presidential campaigns can expect bounce in polls as a result of their conventions

GALLUP NEWS SERVICE

PRINCETON, NJ -- We are now approaching the convention period of this year's presidential campaign. The Democrats will hold their convention in Boston July 26-29, while the Republicans will follow with their convention in New York Aug. 30-Sept. 2.

The overt business of the conventions no longer has much importance. The two major parties' conventions have become little more than four-day pep rallies for each party's candidates and platforms, both of which in reality are determined well before the conventions get underway. There is minimal real news that bubbles up from the convention floor.

But the conventions provide a critically important public relations function. Even with reduced prime-time coverage by the traditional broadcast networks, the average American is exposed to a continuous diet of four days of focused attention on a party's presidential and vice-presidential nominees during a convention.

And this focus affects voters. An analysis of past polling in election years documents that a party's ticket typically enjoys a "bounce" in the horse-race polls after its convention.

The fact that these convention bounces occur has become part of the political conventional wisdom. Journalists and pundits discuss the bounces at great length. They have become one of the most well-documented phenomena in modern political polling.

Thus, the focus of polling conducted in and around the conventions is not so much on the fact that a bounce occurs as it is a matter of interpreting the bounce against expectations. A higher-than-expected bounce can be seen as abnormal and a plus for the party in question. A lower-than-expected bounce (or no bounce at all) can be interpreted as a sign of trouble for a party's candidates.

Indeed, the presidential campaigns now take pains to blunt any psychological impact from the bounce of the opposing party's candidate, setting expectations in order to inoculate journalists and, hence, the public. The apparent assumption: A public that has been led to expect a bounce will be less likely to accord it special significance.

This past week, for example, George W. Bush's presidential campaign made an "expected Kerry bounce" announcement, distributed to the press and conspicuously posted on the Bush-Cheney campaign Web site. Bush strategist Matthew Dowd wrote that he anticipates a Democratic bounce that will leave President Bush and Vice President Dick Cheney at least 15 points behind their Democratic challengers, Sens. John Kerry and John Edwards, by the end of the month, following the Democratic National Convention. As Dowd said, "Assuming that Kerry enjoys the average challenger's bounce … we should expect the state of the race to swing wildly to his favor by early August."

Of course, in a presidential election year, there are two conventions each summer, and thus two convention bounces. The Bush campaign memo did not discuss expectations for the state of the race by early September, after the Republican convention has been completed.

If Bush's Republican convention bounce is not the equal of Kerry's Democratic convention bounce, then Bush will begin the critical last two months of the campaign in a come-from-behind mode. If the Republicans can generate a large bounce of their own as a result of the New York convention, then the race will be tied or even show a Bush lead after Labor Day. In the long run, it is the ultimate net impact of both convention bounces that will really matter. But in the short run, it is apparent that the change in horse-race standings after each convention will be scrutinized in minute detail, with observers from all sides ready to use the size of the bounce as evidence that a convention was a success or a failure.

The Historical Record

This discussion of the expected size of a convention bounce is based on historical comparisons. Gallup has now compiled a database from previous elections sufficient to allow us to make historically grounded estimates of the average changes in presidential horse-race standings as the candidates move through various aspects of their campaigns -- including the conventions.

This summer, everything else being equal, history indeed suggests (as the Bush campaign has pointed out) that the Democratic ticket of Kerry and Edwards should move ahead in the polls after the Democratic National Convention in Boston July 26-29.

History also suggests, as noted, that the Republican ticket of Bush and Cheney would in turn recover positioning after the Republican convention to be held in New York Aug. 30-Sept. 2.

Thus, presuming that the Kerry-Edwards ticket has moved ahead in August, one of the most telling polls of the entire campaign will be the one conducted in early September that will measure how much of the lead the Bush-Cheney ticket has recovered as a result of the GOP convention.

There is always the possibility, of course, that real-world events will change the "usual" pattern of convention bounces this year. It is not outside the realm of possibility that the two campaigns will dream up ways of disrupting the flow of media coverage of the opposing party's convention, and thus disrupt the bounce.

However, barring these types of events, most everyone's attention after the Boston convention will focus on the status of the horse-race poll vis-à-vis historical expectations. And then, a month later, attention will focus on the "recovery bounce" enjoyed by the Bush-Cheney ticket.

More on the Bounce

By definition, bounces have a starting and an ending point. The presidential campaign is dynamic, and there is never any one point before Election Day itself when one can assume stability in the relative standings of the two candidates. Still, pollsters need to define a way of calculating a bounce.

The calculation of the ending point is fairly straightforward: the first nationwide poll taken after the convention ends.

The calculation of the starting point is a little murkier given that there are weeks leading up to a convention that include convention preliminaries such as the selection of a vice-presidential candidate. Nevertheless, for our purposes, we can say that the pre-convention starting point is the last full nationwide poll taken before the convention begins.

The bounce, thus, is the difference in the standings of the two candidates, comparing the last nationwide poll conducted before the convention with the first nationwide poll after the completion of the convention.

Gallup's polling history allows us to examine the bounces that have occurred for both major-party candidates in each election since 1964.

What do we find?

Post-Convention Increases in Support, 1964-2000

Election

Candidate (incumbent in boldface)

Bounce

Convention Order

 

 

 

 

2000

George W. Bush

4 points

1st

2000

Al Gore

8 points

2nd

 

 

 

 

1996

Bill Clinton

5 points

2nd

1996

Bob Dole

3 points

1st

 

 

 

 

1992

Bill Clinton

16 points

1st

1992

George Bush

5 points

2nd

 

 

 

 

1988

George Bush

6 points

2nd

1988

Michael Dukakis

7 points

1st

 

 

 

 

1984

Ronald Reagan

4 points

2nd

1984

Walter Mondale

9 points

1st

 

 

 

 

1980

Ronald Reagan

8 points

1st

1980

Jimmy Carter

10 points

2nd

 

 

 

 

1976

Jimmy Carter

9 points

1st

1976

Gerald Ford

5 points

2nd

 

 

 

 

1972

Richard Nixon

7 points

2nd

1972

George McGovern

0 points

1st

 

 

 

 

1968

Richard Nixon

5 points

1st

1968

Hubert Humphrey

2 points

2nd

 

 

 

 

1964

Lyndon Johnson

3 points

2nd

1964

Barry Goldwater

5 points

1st

As can be seen, the convention bounce varies from a record-high 16-point bounce for Bill Clinton following the 1992 Democratic convention, to no increase at all in Gallup's estimation of the bounce for Democratic candidate George McGovern after the 1972 convention. (Clinton's record 1992 bounce was in part due to the fact that independent candidate Ross Perot dropped out just as the Democratic convention was ending.)

The average bounce across all conventions in these 10 elections is 6.1 points.

The candidate with the bigger bounce, of course, doesn't necessarily go on to win:

  • In 2000, Democratic candidate Al Gore got a larger bounce than Republican George W. Bush did, but Bush won the election (although not the popular vote).
  • In 1988, Democrat Michael Dukakis got a one-point-higher bounce than George H.W. Bush did, but Bush won.
  • In 1984, Democrat Walter Mondale got a larger bounce than incumbent Ronald Reagan did, but Reagan won re-election handily.
  • In 1980, Democrat Jimmy Carter got a slightly larger bounce than challenger Reagan did, but Reagan won.
  • In 1964, Republican Barry Goldwater got a larger bounce than incumbent Lyndon B. Johnson did, but Johnson was way ahead in the polls and stayed that way, ultimately defeating Goldwater by an overwhelming margin.

Factors That Affect the Bounce

Not all conventions are created equal, of course. Idiosyncratic circumstances can make a big difference in the degree to which candidates enjoy the fruits of their party's labors at their conventions.

There are, however, several recurring dimensions that theoretically could affect the size of a convention bounce:

1. The order in which the conventions occur

There appears to be a very slight advantage in terms of a ticket's bounce when that party's convention comes first. But that average advantage across the 10 elections since 1964 appears to be based for the most part on the extraordinary 16-point bounce Bill Clinton received in 1992. Excluding that election, the bounces for the first and the second conventions appear to be no different.

Calendar order

Average bounce

Average bounce excluding 1992

First conventions

6.6 points

5.6 points

Second conventions

5.5 points

5.6 points

2. Whether the party holding the convention is currently in the White House

Challengers appear to have a slightly higher probability of a bounce than do incumbents. Again, a good deal of this average effect is due to the 1992 Clinton bounce. But even when 1992 is removed from the calculations, the challenger candidate gets a slightly larger 5.9 bounce compared to the average bounce for incumbents of 5.6.

Candidate status

Average bounce

Average bounce excluding 1992

Incumbents' conventions

5.6 points

5.6

Challengers' conventions

7.1 points

5.9

3. The partisan effect

With or without Clinton in 1992, the Democrats seem to generate slightly larger bounces with their conventions than do Republicans.

Political party

Average bounce

Average bounce excluding 1992

Republican conventions

5.2 points

5.2

Democratic conventions

6.9 points

5.9

So What Will Happen This Summer?

The Democratic convention this year has three potential, albeit it very small, pluses: (1) it is the first convention, (2) it includes the party challenging an incumbent in the White House, and (3) Democratic conventions generate larger bounces than Republican conventions do.

Based on the past 10 election cycles, the data suggest that the Kerry-Edwards ticket can expect to gain between 5.8 and 6.9 points in the comparison of the pre-convention poll to the post-convention poll.

The Bush-Cheney ticket can expect to gain between 5.4 and 5.5 points after its convention.

Obviously, if the race is roughly equal before the Democratic convention gets underway on July 26, then the best prediction based on history is that the race will be roughly equal or that Kerry will enjoy a slight lead when the dust settles after the GOP convention in early September. If either Kerry or Bush is significantly ahead in early September, then it will signify a deviation from historical averages.

There are confounding factors, of course.

Calculations of gains to be expected from each party do not necessarily mean that one can double them to get an estimate of a gap bounce. In other words, a gain of six points by one party's candidate does not necessarily mean that the other candidate loses that amount.

For example, George W. Bush gained four points after the 2000 Republican convention; Al Gore's percentage did not change at all. Bush gained four points because voters became less likely to declare themselves as undecided or to vote for third-party candidates. As a result, the Bush-Gore gap changed by four points, not eight, after Bush's gain. Thus, differences in the precise way in which a ticket gets a bounce after one of the conventions could affect the post-convention standings.

All in all, it is certainly possible that one or both of the conventions this summer will defy expectations. The expected bounces are averages only. The degree to which this year's conventions are abnormal will be one of the most critically important aspects of this entire campaign.


Gallup https://news.gallup.com/poll/12310/Bounce-Game.aspx
Gallup World Headquarters, 901 F Street, Washington, D.C., 20001, U.S.A
+1 202.715.3030