In my piece on presidential debates in the September/October issue of the Washington Monthly, I argued that while debates can move the polls, they rarely decide the winner of the election. That is, they are typically not “game-changers” in the sense that they vault underdogs to victory. Only in 1960 and 2000 did the debates give the ultimate winner a potentially decisive nudge. More common, the candidate who is leading after the second party convention has gone on to win, according to Robert Erikson’s and Christopher Wlezien’s study of the 1952-2008 election.
For a while, 2012 looked as if it might be different. Indeed, a few pundits seemed to think that the first debate had shown political science was wrong. But at the end, 2012 largely confirms prior research. Romney gained somewhere between 2-4 points after the first debate, which is in line with previous debate bumps. The 4-point swing is most visible in polling averages, but there is some evidence that it was smaller, maybe in the 1-2 point range. (Part of the swing may have been driven not by the people changing their minds, but by the changing composition of “likely voters,” as Republicans became more enthusiastic and likely to declare themselves likely to vote.)
But no matter the precise size of the swing, the first debate was not the end of the story. In my piece, I noted how the impact of a debate can be overtaken not only by subsequent debates but by subsequent events in the campaign. In 2012, the second and third debates — in which Obama was judged to have “won” by pluralities or majorities of voters — did not clearly shift the polls toward Obama. But they did illustrate how difficult it is for one candidate to run the table and win every debate. More common is that the candidates fight to a draw. The late movement toward Obama, which is visible in the Pollster national average, also suggests how later events can matter. Perhaps in this case it was Hurricane Sandy.
Ultimately, the 2012 debates fit the historical pattern nicely. Much as in previous races, they had a noticeable impact but, in the end, Erikson and Wlezien’s conclusion holds: “the best prediction from the debates is the initial verdict before the debates.”