The post-convention polls are just a few days away and reporters are gearing up to see whether Obama or Romney will earn the biggest bounce. But while most will judge the bounces relative to historic expectations, there’s another metric to pay attention to: the share of the popular vote that each candidate attains at the peak of his post-convention heights.
Why pay attention at all to the post-convention bounce? Conventions can really change the race. (Here are some great takes on that essential point by Mark Blumenthal and John Sides.) Taken together, these pieces suggest that first, the average bounce is relatively modest; second, neither candidate is even assured of such a modest bounce in 2012; and third, conventions can have lasting consequences.
But there are limits to bouncecasting. Many focus on the size of the bounce under the assumption that a larger bounce is better. But there is not an especially strong relationship between the size of a bounce and performance in November. For example, Michael Dukakis surged following the DNC in route to a 7-point defeat, while Obama’s modest bounce didn’t prevent him from winning by 7 point a few months later. So while it wouldn't hurt Romney to get a 5 point bounce instead of a 3 point bounce, it wouldn't be wise to draw too many judgments either way.
What’s more important is whether a bounce proves durable. The problem, of course, is that it’s tough to judge whether a bounce will last. So are we doomed to uncertainty for the next few weeks? Or are there useful numbers to consider while the polling averages are adulterated by bumps and bounces? While assessing whether the candidates beat expectations is one one highly limited option, an alternative metric is also worth following: the absolute standing of the candidates at the height of their bounces, and particularly the incumbent.
Start by considering the RCP average from 2004. Notice that Kerry and Bush’s post-convention peaks (Bush 50.4, Kerry 48) closely matched the election’s eventual results (Bush 50.7, Kerry 48.3).
That might just seem like an interesting coincidence at first, but it actually makes a certain degree of sense. If you weren’t willing to support Bush after the extraordinarily well-executed 2004 RNC, when were you going to support Bush? My guess is never. Bush was a well-known incumbent who made his best case and spent a week depicting his opponent as a flip flopper. And on the flip side, if you were ready to oust Bush and vote for Kerry after he … reported for duty, weren’t you all but assured to vote for him in November?
In this model, conventions might represent rare clarifying moments when reluctant and ostensibly “undecided” voters are momentarily willing to offer their support to the candidate that they’ll ultimately vote for in November, even if temporarily. On the flip side, a few reluctant opponents characterize themselves as “undecided” after hearing a week of positive things about the candidate they oppose. Combined, these two forces form a bounce with relatively few voters actually changing their minds.
If the race involves a well-known incumbent, these quick clarifying moments could prove to be enduring and informative. In 2004, someone willing to vote for Bush on September 1 was probably very likely to do so in November. On the other hand, the “post-convention peak” theory probably isn’t useful in races involving lesser-known candidates or when the electorate is less polarized, since a greater number of voters might be truly undecided and new information could sway the race. In 2008, for instance, Obama peaked at 49.2 and McCain hit 48.3 percent. And even in 2004, Kerry’s bounce probably didn’t preclude him from picking up additional voters.
Looking further back, the candidate with the highest share of the vote after their convention tends to do pretty well, winning the popular vote in ten of the twelve presidential elections since 1964. The exceptions were 1988 and 2000, when the challengers zoomed ahead with 54 percent of the vote. But what’s especially striking is the relationship between the incumbent’s post-convention peak and their eventual standing in November. The correlation between an incumbent’s eventual share of the popular vote and their standing in Gallup polling is quite strong (r^2=.73), and even stronger if one excludes Ford, who wasn’t exactly a typical incumbent (r^2=.84). Given that voters are usually uninformed about challengers, it’s not surprising that many with big bounces, like Michael Dukakis, have collapsed once opposing campaigns introduce new information. In contrast, the incumbents who rise above 50 percent tend to win.
Like everyone else, I’ll be following the convention bounces to see whether the candidates meet expectations. But it’s important to recognize that size isn’t everything with bounces, and we won't necessarily know whether the conventions reshaped the race for a few weeks. In the interim, keep an eye on the candidates’ standing in the race, and particularly the incumbents. It has a decent track record and it’s well suited to this election, as I’ll explain in my next post.