Election organizers and political observers have long believed that the weather can change election outcomes, and scientific studies have supported this idea. A new paper proposes that, while this is true, the reasons are not what everyone thinks. Besides keeping people from the polls, rain and snow may also affect which candidates they prefer.
In the United States, Democrats fear bad weather. Turnout drops when it rains – particularly if there are queues – and this hits the blue team harder. Several studies have backed this up. In 2007, Dr Brad Gomez, then at the University of Georgia, found reductions in both votes cast and Democrat share in counties where it rained heavily on presidential election days.
However, Dr Woo Chang Kang of the Australian National University and Dartmouth College's Professor Yusaku Horiuchi noticed something odd in Gomez's data. The Republican share increased in heavily rained-on counties more than could be explained by the lower turnout. In fact, Republicans actually got more votes in absolute terms, not just percentages, when it rained.
In American Politics Research, Kang reports that from 1948 to 2000, rain leads to Republicans winning 1 percent more of the electorate, while Democrats get 2.1 percent less, and 1.1 percent more stay home.
The observation is so unexpected it's natural to assume Kang has made some mistake, and it's always possible a statistician will find an error. Certainly it's hard to believe Republican supporters actually prefer voting in the rain.
Kang has an alternative explanation. He points to extensive evidence that bad weather can change moods and make people more risk averse. Combining this with other research that shows people prefer right-wing candidates when in an anxious state of mind, he proposes the increased Republican vote comes from literal fair-weather Democrats. Perhaps some of those who are committed to doing their civic duty – determined to vote without having made up their mind who to vote for – shift depending on the day's weather.
It seems extraordinary this would apply to at least one person in 100, particularly in America where so many people never vote, but Kang applied robustness checks to the data and the results stood up.
Kang acknowledged to IFLScience that plenty of questions remain. For example, he's not aware of any studies considering temperature rather than precipitation. Today, cold weather might dampen concerns about global warming, increasing Republican support. However, this can hardly have been a factor in Eisenhower's victories, for example. Kang noted that “people feel more comfortable with incumbents,” but admitted he had not tested if this meant presidents are more likely to be re-elected when it rains.
Increased numbers of people voting prior to polling day may have blunted polling day effects in recent years, but Kang sees plenty more to explore.