Climate, weather, and political behavior

This dissertation explores the extent to which weather and climate systematically affect political behavior. The idea that weather (and other elements of the natural world) exercise a fundamental influence on politics has long been a theme in classical and modern political thought. As political science moved from pure description to a more social-scientific form of analysis, scholars became less interested in understanding the impact of climate. If mentioned at all, weather typically is referred to as one of the various elements making up the "error term" in our statistical analyses. Recent work in the natural and social sciences, however, has suggested there are systematic and important links between weather, climate, and behavior. This work (which I review) not only inspires a return to a traditional focus of political analysis, but more importantly provides a number of hypotheses to guide our analysis of politics. Inclement weather increases the costs of moving from place to place. Sunlight enhances while extreme temperature depresses mood. Finally, hot weather is associated with enhanced aggression.

These correlates of climate have implications for a variety of subfields across political science, including comparative politics and international relations. This dissertation concentrates primarily, however, on American politics, particularly from a behavioral perspective. To see if weather has a significant effect on politics, then, I explore behavior in four settings that have been especially important in mainstream studies: Presidential approval; social capital; Election Day voting; and finally elite participation (in the form of abstention on roll call voting). In terms of the first, if (as Zaller argues) a response to a telephone survey indeed entails a summing up of `considerations' regarding an issue rather than expression of a `true' attitude, then it is likely sunlight should stimulate positive responses to questions because it encourages the release of serotonin, which makes people more positive in general. Controlled logistic regression of sunlight on Presidential approval reveals that, in spring, sunlight boosts approval. The next chapter explores how hot climates and rain may reduce levels of social capital. This is because heat boosts levels of aggression, which should diminish helping behavior, and because rain makes it more difficult to volunteer and associate with other people. Analysis of state-level social capital data and city-level volunteer data provides some evidence that these propositions are correct. The third empirical chapter focuses upon voting on Election Day. While it finds that rain does have a depressive effect upon voting rates among the poor due to raising the costs associated with voting, there is little evidence that vote choice is affected by the weather. The final empirical chapter examines how weather conditions may affect voting rates among members of the United States House of Representatives, which seems possible because, like regular citizens during Election Day, House members pay costs when visiting the Capital to vote, and unpleasant weather could comprise a real if minor cost. OLS regression at the vote-level and logistic regression at the legislator level reveals that in the winter and spring, sunlight boosts voting, while summer humidity depresses voting and heat in winter has a positive effect. While these conclusions are interesting in themselves and meaningfully contribute to contemporary academic discussions, they further suggest some things about how we thing about political science. In particular, analyses of political topics could often be enhanced by reflectively considering the contents of the error term, as this exercise can offer new and useful perspective on current scholarship. Further, this dissertation also suggests that political science (and research in general) could benefit from taking a more comprehensive view of the environmental context of human behavior.