A debate among scientists over climate change and conflict has turned ugly. At issue is the question of whether the hotter temperatures and chaotic weather produced by climate change are causing higher rates of violence. A new analysis refutes earlier research that found a link, and the two lead researchers are exchanging some pointed remarks.
Last year, a team of U.S. researchers reported a robust connection between climate and violence in Science. But in a critique published online yesterday in Climatic Change, a team of mostly European researchers dismissed the connection as "inconclusive." The Science authors are hitting back, claiming that the critics are fudging the statistics and even manipulating their figures. The new analysis "is entirely based on surprisingly bold misrepresentations of our article, the literature, basic statistics, and their own findings," says Solomon Hsiang, the lead author of the Science paper and an economist at the University of California, Berkeley.
Numerous past studies have found a correlation between heat waves and violence, manifesting as conflicts between individuals and between groups. Demonstrating a direct connection between climate change and violence on a global scale, however, is tricky. It requires a meta-analysis of hundreds of already published studies that have slightly different techniques and measurement scales. Hsiang's team performed just such a meta-analysis and grabbed headlines with their findings that a changing climate appeared to be amping up conflict.
The Science paper was met with some skepticism, however, and some of those skeptics have been building their case. The Climatic Change critique is authored by Halvard Buhaug, an economist at the Peace Research Institute Oslo, and co-signed by 25 of his colleagues. The problem, Buhaug wrote in an e-mail to ScienceInsider, is that the meta-analysis "blends all sorts of actors at all sorts of spatial and temporal scales. … [They] draw sweeping conclusions that, supposedly, are robust and apply across scales and types of violent conflict. Of course that doesn’t make sense. But it works if you seek attention." He also accuses Hsiang's team of "severe bias in sample selection” and says that his analysis of the same data did not support the climate-conflict link.
Why critique the research now? The study "appears to have had some influence on policy thinking," Buhaug wrote, citing a recent U.S. Department of Defense road map on addressing climate change in military planning and another report by the CNA Corporation on climate change and security. Such official statements "reinforce the impression that the climate-conflict link is considered uncontroversial in policy circles,” Buhaug wrote. “As scientists and experts on this issue, we see it as our duty to provide a more balanced message."
Hsiang in standing by his analysis. In a detailed, blow-by-blow blog post responding to the new paper, Hsiang charges Buhaug with basic mathematical errors that undermine his conclusion. In an e-mail to ScienceInsider, Hsiang also accuses Buhaug’s group of "doctoring the display of their figures." (The evidence of that alleged doctoring is laid out in Hsiang’s blog post.)
The spat has other researchers exasperated. "What is frustrating is that they can't work together," says Andrew Solow, a statistician at the Woods Hole Oceanographic Institution in Massachusetts, who is associated with neither side. "Why can't they get together and thrash these issues out? Even if they don't come to an agreement, they could explore alternative modeling choices and their implications."
Solow adds that he is "not a big fan" of meta-analysis, in part because the technique sparks disputes like these. Rather than directly addressing the scientific question of whether climate change is causing an increase in conflict, he says, "this disagreement is over the degree to which studies of the climate-conflict link agree [with each other]."