A subsequent version of this paper was published in American Law and Economic Review 11, no. 2 (2009).

The reintroduction of capital punishment after the end of the Supreme Court moratorium has permitted researchers to employ state level heterogeneity in the use of capital punishment to study deterrent effects. However, no scholarly consensus exists as to their magnitude. A key reason this has occurred is that the use of alternative models across studies produces differing estimates of the deterrent effect. Because differences across models are not well motivated by theory, the deterrence literature is plagued by model uncertainty. We argue that the analysis of deterrent effects should explicitly recognize the presence of model uncertainty in drawing inferences. We describe methods for addressing model uncertainty and apply them to understand the disparate findings between two major studies in the deterrence literature, finding that evidence of deterrent effects appears, while not nonexistent, is weak.

Also listed as Wisconsin SSRI WPS 2007-3; Columbia Public Law Research Paper 07-138;

This paper was revised in July 2007.