Obama’s bounce has arrived and so has a debate about its meaning. Some contend that Obama’s bounce will fade over the next few days and that Romney remains very much in the race. Others argue that analysts should hold out for two weeks before making proclamations. Bounces tend to overstate a candidate’s strength, so there are risks in extrapolating too much from Obama’s initial improvement. The critical question is whether it’s possible to learn any lessons from the bounces themselves, even before we learn whether they endure. In my view, the answer is “yes.”
In contests involving an incumbent president, the candidate with a higher share of the vote following his convention in Gallup polling has won every election since 1964. We can go further: no modern candidate has won the presidency without taking a lead after his own convention. And there is a strong relationship between the incumbent’s share of the vote and their eventual finish in November.* To take a recent example, Bush peaked at 50.4 percent in the RealClearPolitics average after the RNC, and eventually won 50.7 percent of the vote. And it’s worth noting the largest errors involved races with incumbents who took the presidency due to assassination or impeachment (Ford, Johnson), and a year in which the polls systematically advantaged the incumbent (1996).
If Obama leads after two weeks, we’ll be able to add an additional clue to the list: the candidate leading two or three weeks following the final convention has gone on to win every modern presidential election. But even right now, the other three empirical findings suggest Romney’s chances are in jeopardy. Although the race is close by historical standards, it has been remarkably stable and remarkably clear: Romney has never led in the RealClearPolitics average and, no, a candidate has not won the presidency without holding a lead in the polls by early September. If Obama approaches fifty percent of the vote, as Gallup and Rasmussen suggest, that would clearly repudiate the view that a majority of voters are unwilling to reelect the president, which is the entire basis for Romney’s case for a come-from-behind, Reagan-esque sweep of undecided voters.
In that sense, 2008 is very different than 1980. Yes, Carter took a lead after his convention. But Carter never strayed far from 40 percent, which should have been a sign that he wasn’t even close to commanding the support necessary to win reelection. In contrast, Reagan surged to fifty percent of the vote following his convention, mirroring his eventual finish. The mass of post-RNC Reagan voters became "undecided" following Carter’s convention, but their initial preference seems telling in retrospect. The undecided voters who ultimately broke toward Reagan previously indicated a willingness to support Reagan, never demonstrated a willingness to support Carter, and uniformly disapproved of Carter’s performance. And two weeks after the DNC, Reagan pulled ahead and never relinquished his lead.
This election’s unusual stability makes Obama’s potential movement even more significant. For two years, Obama’s approval ratings hewed within just a few points of 47 percent and the entire summer elapsed without any discernible movement toward either candidate. Romney wasn’t able to secure any post-RNC bounce, and Obama’s gains represent the first decided movement toward either candidate since Romney won the Republican nomination. Put differently, we now know there are voters open to voting for the president beyond the initial 47 percent he held in summer polls, but Romney has not demonstrated similar upside.
The skeptics are right about one important matter: these assessments are tentative and hinge on confirmation by additional pollsters later this week. But if a broad set of polls show Obama near or above fifty percent with a clear lead among likely voters, we won’t need to wait a few weeks before drawing initial conclusions. Obama’s bounce would be significant in its own right, even if it fades. If it lasts, it might be time for a conversation about whether Romney can actually win in November.
*Someone with a better set of historic data should confirm or dispute the relationship between an incumbent’s share in post-convention polling and their finish. With the exception of 2004 (RCP average), the numbers are based on Gallup polls.