Boston University Arts & Sciences

CAS News

Modeling Election Results

Douglas Kriner

Perusing the morning newspapers on November 7, I found no shortage of explanations for President Obama’s reelection victory. In 2011, the Obama campaign aired a significant number of ads during 2 Broke Girls—what else could explain the president’s success among young women voters? The Chicago campaign went negative early and often to paint Governor Romney as an out-of-touch, outsourcing executive who put profits over people, while the Romney campaign waited too long to respond. Governor Romney simply made too many unforced errors: $10,000 bets; disparaging the “47%”; dressage; and the like. And of course, there was Hurricane Sandy. Whereas Romney had won the momentum after Obama’s lackluster performance in the first debate, in the assessment of the Washington Post’s Scott Wilson and Philip Rucker, “for a candidate whose political career has been touched at times by luck, Hurricane Sandy arrived with a week left in the race and disrupted Romney’s effort.”

Definitively determining why an election unfolded the way in which it did is a Sisyphean task. There are simply too many potential causal factors, and not enough of the data needed to distinguish among them. However, before simply embracing your favorite narrative spun by the media, it is worth re-examining the fundamentals of the 2012 race. Every presidential election, political scientists construct a variety of forecast models that endeavor to predict the outcome of the race, usually before the party nominating conventions, using only a few simple variables. Most of these models include some measure of the state of the economy, the president’s approval rating (again, usually measured before the conventions), and some measure of incumbency. Like polls, different forecasters use different formulas and arrive at different answers. However, the average forecast from the 2012 PS Symposium was for President Obama to win 50.4% of the two-party vote; the median forecast had the president capturing 50.9%. These predictions are very close to the final result and far out-performed many of the polls conducted during the campaign itself.

How could most forecasts have predicted an Obama win, given the continued economic difficulties? First, the strength of the economy depends on how you measure it. In terms of unemployment, the labor market remains quite weak. On other metrics, however, the economy’s performance is stronger. The stock market (and Americans’ 401(k)s) have rebounded nicely from their post-crash lows. Consumer confidence in October 2012 rose to a five-year high. And looking into the future, many more Americans predict that the economy will improve over the next year than believe it will regress. This, coupled with the fact that a clear majority still blame President Bush rather than President Obama for the country’s economic problems has also bolstered President Obama’s approval ratings, which throughout 2012 hovered right around 50%. Together, these fundamentals suggested a narrow electoral victory for the President, even without natural disasters, clandestine videos from wealthy donor fundraisers, Olympic equestrian events, and even…advertisements on 2 Broke Girls.

Return to the Newsletter

Keep Us in the Loop

Let us know about news or upcoming events. Please send news items to megmay@bu.edu, or call Megan Maynard Winderbaum at 617-353-2540.



Boston University


Boston University
Arts & Sciences
725 Commonwealth Avenue
Boston, MA 02215
bu.edu/cas