Take, for example, the polling work placing President Barack Obama in opposition to a "generic" Republican candidate, versus his showing in opinion surveys against specific GOP presidential hopefuls. Two recent national polls--from Rasmussen and NBC--now show Obama losing by six and four percentage points, respectively, when he's pitted against a "generic Republican" candidate. (Results in Rasmussen traditionally lean Republican, but that's not the case for NBC; the Rasmussen survey has a two-point plus-or-minus margin of error, while NBC's error-margin in plus-or-minus 3.1 percent.)
Yet the Democratic incumbent is doing much better against most of the actual Republican candidates. According to a pair of other recent polls from Quiinipac and ABC, Obama is (respectively) down two and four percentage points against former Massachusetts Gov. Romney, who is now seeking to solidify his position as frontrunner in the GOP field. (Those polls have respective margins of error of plus-or-minus 3.2 and 4 percentage points.) But in other head-to-head surveys, the president's polling numbers show him ahead of all other major Republican challengers. He is up by three and one percentage points against Texas Gov. Rick Perry in those same two polls. And he is up by even larger margins against Herman Cain, Newt Gingrich, Ron Paul, Michelle Bachman, Jon Huntsman and Rick Santorum.
The obvious takeaway at this early stage of the primary cycle is that the "generic Republican" is a more effective candidate than any specific Republican. In part, that's because there's a bit of a built-in bias in polling for more abstract forms of political allegiance. In these surveys, respondents are registering their reactions to Obama as a known political quantity, with a full array of perceived electoral strengths and weaknesses. Yet for the no-name Republican, "Respondents get to project the person they think is the most electable actual Republican or even an imaginary Republican that is not in the race," as Wharton economist Justin Wolfers has observed.
Polls also produce a related "anti-incumbent bias" for much the same reason. Particularly over presidential election cycles, the incumbent figure typically tends to poll lower than unknown or lesser-known opponents do. Incumbents tend to gain ground late in the cycle and on Election Day--because at that point, the knowledge gap that voters have experienced between the opposition and the incumbent in a given race finally starts to narrow.
The prediction markets continue to have Obama at slightly less than 50 percent likely to win re-election; how does this finding jibe with what head-to-head opinion surveys say about the race? Going by the findings of the prediction markets, I can say with some confidence that Obama has roughly a 50 percent likelihood of facing off against Romney in November 2012--and a near-equal chance that the GOP nominee will be someone else. And going by recent poll results, these respective scenarios would mean that Obama would be 50 percent likely to be slightly down in the polls against his future opponent (i.e., Romney) or just about as likely to be slightly ahead of a non-Romney nominee. So factoring in the anti-incumbent bias of most polling, we can conclude further that the prediction markets, by leaning slightly toward a Romney-Obama matchup at the moment are actually more pessimistic about the likelihood of Obama's re-election than the current polling is.
David Rothschild is an economist at Yahoo! Research. He has a PhD in applied economics from the Wharton School of Business at the University of Pennsylvania. His dissertation is in creating aggregated forecasts from individual-level information. Follow him on Twitter @DavMicRot and email him at PredictionBlogger@Yahoo.com.