Continued from page 1

• Sunshine and rain;

• Shark attacks. By which we mean: shark attacks.

Strange but true: In the summer of 1916, four people were killed during a weeklong series of shark attacks along New Jersey’s coastline. A few months later, President Wilson lost as much as 10 percent of his expected vote in the areas where the attacks occurred — never mind that he was a well-known former governor of the state and not an incumbent local lifeguard.

Now consider college football. In a 2010 study titled “Irrelevant Events Affect Voters’ Evaluations of Government Performance,” Mr. Malhotra and co-authors Andrew Healy and Cecilia Mo analyzed wins and losses for 62 NCAA Division I schools over a 44-year period, as well as contemporaneous election results for the counties containing those schools.

Using statistical modeling and a series of tests, the trio found that gridiron performance seems to alter behavior in the voting booth, with the positive relationship between winning football and incumbent vote share going beyond mere correlation.

In a related experiment, they also found that victories by local teams in the 2009 NCAA men’s basketball tournament led people to give President Obama a higher approval rating.

Moreover, the increase in incumbent vote share was larger for surprise college football wins that defied pregame gambling odds — roughly a 2.5 percent bump — and also in areas where the sport is especially popular, with incumbent vote share rising to 3.35 percent.

How significant can the latter number be? Increase Mr. Obama’s 2008 Franklin County vote totals by 3.35 percent — hardly inconceivable, given that Buckeye football is a civic religion in and around Columbus — and he would receive almost 20,000 additional votes.

“We were struck by how big the effect was,” said Ms. Mo, a political science professor at Vanderbilt University. “We asked ourselves, ‘Is this too big?’ But we ran so many tests to undermine our results and couldn’t.”

Other studies show similar effects. German researchers found that positive performance by the nation’s soccer team between 1993 and 2002 resulted in increased popularity for leading candidates from the party in power. Conversely, soccer losses led to less-favorable voter assessments.

Mr. Miller recently analyzed American mayoral elections from 1948 to 2009. He compared ballot-box results with professional football, basketball and baseball performances by teams in the same city.

To his surprise, Mr. Miller found that sports success could result in as much as a 6.1 percentage point boost in an incumbent’s vote share — bigger than the well-established negative effects associated with unemployment rates and larger than the margin of victory for New York City Mayor Michael R. Bloomberg in 2009. Mr. Miller estimates that 1 in 17 modern, major-city elections ultimately may have been decided by variations in sports records.

Speaking of Mr. Bloomberg: In 2009, the New York Yankees won the World Series. Coincidence?

“The effect is not trivial,” Mr. Miller said. “I estimate that going from none of a city’s teams in the playoffs to all of them in the playoffs boosts a mayor’s chance of re-election by around 13 percent. And the causal chain is pretty intuitive. Winning teams make citizens happier, and happier citizens are more likely to re-elect incumbents.” 

The irrational voter

Story Continues →