Saturday, March 17, 2012

29 October 2010 – Rain, campaigns, and automobiles


Next Tuesday's midterm elections offer an opportunity to validate, through empirical observation, a thesis published a few years ago that attempted to build a bridge from the physical world to the social sciences.  In a 2007 edition of the Journal of Politics, Brad Gomez, Thomas Hanford and George Krause published a paper entitled "The Republicans Should Pray for Rain: Weather, Turnout, and Voting in U.S. Presidential Elections".  In this paper, the authors noted that the role of weather in determining the outcome of elections through its impact on voter turnout has been repeatedly referenced in the media, and lamented that very little actual empirical evidence was available to substantiate the claims being tossed about by various political and journalistic talking heads.

The authors decided to provide some. 

By using data from more than 20,000 US weather stations, they investigated correlations between rain and snow, voter turnout figures, and the end result of elections, and determined that snow reduces voter turnout by about half a percentage point per inch, while rain reduces voter turnout by nearly a full percentage point per inch (the difference is not surprising, given that an inch of rain is a much more discombobulating event than an inch of snow).  More interestingly, though, they also discovered that poor weather tends to favour Republican candidates.  This suggests that Republicans are less likely to be deterred from voting by inclement conditions.  The authors draw the conclusion that good weather across the US likely contributed to Kennedy's victory over Nixon in 1960, where the two were separated by only 118,000 popular votes; while bad weather in Florida in 2000 led to Gore losing the state, and therefore the presidency (some guy called "D. Chad" was also implicated).

Why is this interesting?  Well, right now it looks like bad weather - cold and wet - is on the way for the US eastern seaboard and the Pacific northwest next Tuesday evening.  According to polls, Republicans are already poised to sweep the House, to make significant gains in the Senate, and to take two-thirds of the gubernatorial races.  If the Gomez-Hanford-Krause thesis is correct, poor weather along the east would further tip the outcome of next Tuesday's elections towards to the right, while poor weather in the Pacific northwest could impact Democrat turnout and put districts assumed to be "blue" strongholds into play.  With the Senate already likely to end up in a near-split, the balance of power in Congress for the next two years could be decided by something as commonplace as rain.

This is not, interestingly, a purely US phenomenon.  David Childs, in Britain Since 1945: A Political History (2006), notes that bad weather during the 1951 elections resulted in a drop in turnout, helping to give the Conservatives a clear victory over Labour (p. 40).  There are counter-arguments, of course; a1994 study by Steve Knack using NCDC data and looking only at the '84, '86 and '88 elections found "no evidence of partisan differences in the turnout-deterring impact of inclement weather."  But this study looked only at 3 election cycles.  Overall, empirical evidence suggests that bad weather disproportionately depresses turnout for the Democrats.  Just offhand, I wonder whether weather impacts voter turnout unevenly in Canada?

What's disappointing in the G-H-K paper is that the authors decline to make any attempt to understand why conservatives benefit from bad weather on election day.  This, it seems to me, is the crux of the whole phenomenon - finding an answer to why those who vote Republican are more willing to venture out into the "snow, rain, heat or darkness", to borrow Herodotus' (or the US Postal Service's) words, in order to cast their ballots.  The answer could be a highly complex aggregate of innumerable small factors.  Is it some sort of atavistic rugged individualism thing?  Are Republicans more likely to own gumboots and a slicker?  Or is it something simpler? 

According to research cited in a 2005 article in the New York Times, for example, 46% of Jeep Grand Cherokee drivers identified as Republicans, while only 28% identified as Democrats.  Not surprisingly, the split was more pronounced for Hummers (52% Republican, 23% Democrat); while the "Number 1 vehicle" for "farmers, ranchers, contractors, [and] independent businesspeople" - a solidly Republican group - is the “Ford F-Series pickup truck".  Meanwhile, ownership statistics for minivans and for smaller vehicles like the Ford Escort, the Dodge Neon, and the Toyota Prius - none of which are noted for their handling in inclement weather - all skewed significantly Democrat. Saturns are the most unbalanced; they tend to be driven by Democrats by nearly a 4:1 ratio.

Perhaps the uneven impact of bad weather on voter turnout is less the result of staunch, frontiersman-like determination to exercise one's franchise than it is of the fact that, ceteris paribus, people who own 4WD vehicles are statistically more likely to vote Republican.

It's a funny old world, isn't it?  Now, if only I could find some hard data on the political preferences of people who buy umbrellas and rubber boots.



P.S.  Full disclosure: I've driven a Jeep Grand Cherokee for the past 12 years.  And I've never, ever failed to cast my vote.