Weathering unrest

Abstract Over the last several years research has examined anew the potential for climate change to induce political conflict and potentiate social unrest. Several explanations for the relationship between weather and social unrest have been proposed, including the idea that temperature, acting through a physiological response mechanism, gives rise to collective aggression. This proposition first appeared in the aftermath of the 1960s US riots, which occurred primarily in the heat of summer, and has re-emerged within the contemporary literature on conflict and climate, in addition to explanations rooted in political economic processes. Building on both bodies of work, this article utilizes a case-crossover time-series design to explore the relationship between meteorological factors derived from high resolution spatial data of temperature and precipitation and social disturbances occurring in 50 major cities in Africa and Asia between 1960 and 2006. Poisson regression and generalized additive modeling are utilized to model linear and non-linear effects, respectively. A significant, but qualified, association between heat and urban social disturbances is found. The general relationship is non-linear, with peak levels of unrest occurring in the upper 20s (°C). The relationship between temperature and social unrest within individual cities is linear. In addition, there are differential effects of heat on lethal versus non-lethal episodes of unrest. The non-linear response to temperature is much more pronounced among lethal events than it is among non-lethal episodes. The conclusion taken from this research is that heat is associated with urban social conflict, but generally does not trigger episodes and instead acts to supplement aggression while other factors govern the primary timing of social unrest.