Marshall Burke and Sol Hsiang wrote a short response to a recent editorial in Nature. The editorial was itself a response to an article in Nature Climate Change that argued that climate-conflict research contained systematic sampling bias. The published Correspondence is pretty succinct:
“Your Editorial on sampling bias in studies linking climate change with civil unrest (Nature 554, 275–276; 2018) is based on an analysis that in our view provides no evidence for biased results (C. Adams et al. Nature Clim. Change 8, 200–203; 2018).
We disagree with your contention that it is “undesirable” to study risk factors for populations with a high likelihood of conflict on the grounds that it could “stigmatize” these regions as politically unstable. The same logic would argue against studying risk factors for people who have a high chance of developing cancer for fear of stigmatizing patients. In our view, such recommendations could create bias in the literature by inhibiting research.
Studies of connections between climate and conflict should instead be motivated to identify causes of human suffering so that it can be alleviated (see, for example, M. Burke et al. Annu. Rev. Econ. 7, 577–617; 2015). We do not believe that shying away from investigations in this field is an effective path towards this goal.”