Courtenay Conrad and Will Moore have an important new article (ungated version) in the American Journal of Political Science that analyzes why states, if ever, stop using torture. The use of torture is remarkably persistent: 93% of states that are accused of using torture in one year will face similar accusations in the following year. Conrad and Moore find that liberal institutions matter, especially in the absence of campaigns of violent dissent. If violent dissent persists, however, even countries with strong liberal institutions are highly likely to carry on torturing.
Conrad and Moore’s theoretical innovation is to take seriously that it is not the state but agents of the state (e.g. police, military, prison guards) who engage in torture. Executives do not always have perfect control over what these agents do. In order to explore this further, they received an NSF grant to collect data on accusations of torture against specific state agents across the globe. A beta version of this data is now available. Below is a first graph they created, which shows the regional distribution of allegations of torture against specific agencies. As you can see, the Middle East stands out in the relative prominence of the military and prison guards as perpetrators. There is much more detail in the data and I look forward to seeing what we can learn from it.