We are most grateful to John Donohue, Justin Wolfers, and Carol Steiker for their valuable and illuminating responses to our article. Donohue and Wolfers explore empirical questions, on which we have little to say. Steiker investigates the moral issues, and here our Reply must be more extensive.

Donohue and Wolfers believe that, with respect to the death penalty, “existing evidence for deterrence is surprisingly fragile.” They attack the peerreviewed empirical work of a number of social scientists, including Hashem Dezhbakhsh, Paul Rubin, Joanna Shepherd, H. Naci Mocan, R. Kaj Gittings, and Paul Zimmerman. They highlight theoretical claims by Lawrence Katz, Steven Levitt, and Ellen Shustorovich, who emphasize the infrequency of capital punishment and who thus doubt the claim of deterrence. (Interestingly, Katz, Levitt, and Shusterovich do find that prison deaths have massive effects in deterring murders and other crimes.) Most importantly, their own work, using existing data, suggests that deterrence has not been shown.

Donohue and Wolfers misunderstand the point of our article. Let us distinguish among three purposes for which one might discuss the recent deterrence evidence: (1) to argue that, in fact, capital punishment deters murder; (2) to argue that the evidence has reached a threshold of reliability such that policymakers should change laws now, adopting capital punishment; and (3) to argue that the evidence has reached a threshold of reliability, much lower than in (2), such that it is worthwhile to consider the moral implications of the evidence. We do not mean to take a stand on either (1) or (2). We do not know whether deterrence has been shown; and contrary to Donohue and Wolfers’s suggestion, we do not insist “that it would be irresponsible for government to fail to act upon the studies.” Nor do we conclude that the evidence of deterrence has reached some threshold of reliability that permits or requires government action upon it right now. . . .