WASHINGTON, D.C. -- As the 2012 presidential conventions get underway in Tampa, Fla., a Gallup analysis of 15 elections from 1952 to 2008 shows that in all but three instances -- 1988, 1992, and 2004 -- the candidate leading in the Gallup poll conducted just prior to the first convention (the "pre-convention poll") has won the November election.
Pre-convention polls are not good predictors of a candidate's final vote share, but they are useful in terms of simply forecasting which candidate will win the election. Overall, 80% of the pre-convention leaders went on to become president, although that success percentage figure includes the disputed 2000 election, in which George W. Bush was elected without winning the popular vote.
Barack Obama and Mitt Romney have been tied or very close in recent Gallup Daily tracking averages. Gallup's final pre-convention standing of the two candidates will be based on the Aug. 20-26 average and posted Monday afternoon on Gallup.com. If either candidate is ahead in that average, the historical data outlined below suggest that candidate is more likely to be the eventual winner, although close races such as this one suggest more potential for exceptions to that pattern.
Big Bounces, Lack Thereof Have Bucked the Trend
Each of the three elections in which the leader in the pre-convention polls went on to lose had differing scenarios.
In 1992, George H.W. Bush led in both a two-candidate race vs. Bill Clinton and a three-candidate race vs. Clinton and independent Ross Perot prior to the Democratic convention that year. Perot dropped out of the race during the Democratic convention. Clinton enjoyed a tremendous bump in support after the convention and never trailed again, even after Perot rejoined the race in October.
The other two elections in which a leading candidate eventually lost provide clearer examples of changes in the overall political dynamic caused by the convention/debate segment of the campaign. Entering the conventions in 1988, then-Vice President George H.W. Bush trailed Democrat Michael Dukakis by six points prior to the Democratic convention, and fell further behind after it. Bush's lagging in the polls might at least partially be ascribed to "party fatigue," stemming from the same political party's holding control of the White House for two terms (all four incumbent vice presidents running for president since Richard Nixon in 1960 trailed their opponent at this point of the campaign). However, a month later, Bush received a big bounce from his convention that pushed him into the lead, and he never trailed from that point on.
The 2004 election was generally competitive between George W. Bush and John Kerry throughout. Bush was slightly behind Kerry just before the first convention -- the Democrats', from which Kerry got no bounce -- but Bush went on to win the election by three percentage points that November.
Eventual Vote Tallies Tend to Differ From Pre-Convention Polls
Regardless of the eventual winner, the margins in the final popular-vote tally can differ significantly from what is suggested by the pre-convention polling. This can result from many factors -- the convention bounce, the debates, the effect of the campaigns, undecided voters finally choosing a side, and which registered voters actually vote in the election.
John Kennedy was leading Nixon by four points prior to the conventions in 1960, yet went on to win the vote by less than one point. Jimmy Carter was leading Gerald Ford by 17 points prior to the conventions in 1976, yet won by only a two-point margin. Ronald Reagan was ahead by three points in 1980, but he won by 10 points. Bill Clinton's pre-convention margin of 22 points over Bob Dole shrank to nine points in the final popular vote.
Margins fluctuate, but the strength of support in pre-convention polls can identify solid favorites. All candidates polling above 50% prior to the first convention, always good territory for office-seekers, went on to victory -- although Jimmy Carter nearly became an exception in 1976, with his two-point margin against incumbent Gerald Ford.
Close 2012 Race is Akin to 1960, 1968, 1980, 2004, and 2008
This year, incumbent President Obama has not been able to sustain job approval ratings above 50% and in similar fashion so far has not edged over 50% in Gallup Daily tracking polls of presidential vote preferences (although Romney has not either). Obama and Romney have for the most part been closely matched over the last four months. Thus, the current standing of the two candidates in the polls most closely resembles 1960, 1968, 1980, 2004, and 2008 -- all elections in which pre-convention polling showed one candidate leading the other by four points or fewer. Still, in four of those five, with 2004 the exception, the candidate with even a slim lead prior to the first convention went on to victory.
Close Races Going Into Conventions Most Likely to See Shake-Ups
Gallup's historical trends demonstrate that both candidates tend to get a "bounce" out of their respective conventions. In order to control for the possibility that support for the candidate nominated in the second convention would be overstated in polls directly after that convention, Gallup examined surveys conducted at least 14 days after the last convention (but before the first debate, if applicable).
Historically speaking, candidates well behind in the polls before the convention process begins are unlikely to improve their position in any meaningful way, even if they achieve a small bounce. And candidates who are well ahead also maintain their positioning. Presidential nominating conventions have influenced voter preferences and even created new favorites only when the final pre-convention-period poll showed signs of electoral gridlock in the form of no clear, indisputable winner. In that sense, 2012 looks to be an especially ripe year for the conventions to have a significant effect on the overall standing of the two candidates.
This is because there is a higher likelihood of a "shake-up" in who leads in the polls if the two candidates are running fairly close before the conventions begin. Furthermore, these changes can be consequential. In 1988, 1992, 2000, and 2004, the lead changed after the two conventions, and three of those candidates who emerged as leaders went on to be elected president. The fourth such candidate, then-Vice President Al Gore, did go on to win the popular vote, but not the election.
In another two instances, 1960 and 1980, the candidates running behind were able to pull even after the conventions, but both were eventually defeated.
Also, 1968 provides an example of how unusual circumstances produce an unusual process. Before the conventions, the Republican Nixon held a modest lead of two points over Democrat Hubert Humphrey. After both conventions, Nixon's advantage had swelled to 15 points, most likely as a result of the highly visible breakdown of law and order at the Democratic convention in Chicago. Still, by Election Day the race had become neck and neck again, and Nixon ended up winning by a slim one-point margin in the popular vote.
Bottom Line
As the 2012 presidential campaign enters the convention phase of the cycle, the large-scale, nationally televised rallies in Tampa and Charlotte will likely excite each party's fervent supporters and serve as an endless source of commentary among political analysts as to which candidate put on the better performance. Some of this energy may be reflected in the polls: Gallup has found that most candidates can expect a "bounce" in their support after their nominating convention, with the median uptick being five points.
Observers will be especially interested in the impact of the two conventions, given the closeness of the 2012 race, in which both candidates have rarely enjoyed a lead any larger than three points in Gallup Daily tracking. All else being equal, the leader of the Gallup poll prior to the convention has an 80% probability of winning the election, according to past data.
Of course, all else is not equal. When pre-convention polls show a tight race, as is the case this year, conventions have been more likely to create new leaders or galvanize support for a heretofore weak leader. Thus, both President Obama and Romney have the potential this year to gain an upper hand as a result of the convention process.
A future analysis will look over the relationship between polls conducted soon after both conventions and their worth in predicting the election outcome.
