John Schiemann offers a considerate (and generous) review of my book. I am grateful for the opportunity to respond to some of the many valuable points he has raised, both empirical and theoretical.

On the empirical side, Schiemann expresses concern over “possible undue influence of measurement error” in the main outcome variable of prisoner abuse—across all levels, but particularly medium-level violations—that I created for the quantitative portion of the book. On the one hand, measurement error in the dependent variable is generally known to reduce the efficiency of causal estimates, even if not biasing them. The presence of measurement error would thus actually make it harder to identify statistically significant relationships for the various determinants of prisoner abuse.

On the other hand, Schiemann’s comments raise deeper questions about the challenges of gathering data on wartime conduct, and political violence more broadly. Contrary accounts, propaganda, and oftentimes a dearth of reliable information make the collection and interpretation of conflict data daunting at best. Recent work reminds us that gradually shifting standards in what is viewed as abuse can skew evaluations of political violence even by well-intentioned monitors and researchers (Christopher J. Fariss, “Respect for Human Rights has Improved Over Time,” American Political Science Review, 108(2), 2014). In his review, Schiemann kindly notes the various steps I took to cross-validate codings for prisoner treatment, but this remains an endeavor on which conflict scholars should continually seek to critically self-evaluate and improve.

In his comments, Schiemann’s attention to moderate-level abuses against captured combatants, and how patterns with certain explanatory factors may vary, is an important one. While I focus on high-level abuses given their egregious nature, my argument is intended to apply to the entire range of severity in violations. One of the main points of the book is that even though democracies may be less prone to prisoner abuse overall, they can nonetheless engage in violations (sometimes enthusiastically). The propensity to resort to more moderate abuses, as he observes, may parallel a more general dynamic evident in recent research on repression, where democracies engage in more subtle and indirect forms of political violence—an interesting implication that was not fully drawn out in the book.

On the theoretical side, Schiemann’s main criticism is that the book’s overall “argument is more a collection of things that matter to states and democratic publics, rather than a set of more or less brittle propositions that might be proven false.” One of the overall objectives of this book was indeed to delineate both costs and benefits of prisoner abuse. I then developed an argument centered on the role of domestic political attributes and the nature of the conflict that combine to yield several falsifiable expectations about the conditions under which belligerents would emphasize the costs versus benefits of abuse when deciding about how to treat captured combatants. In addition to various statistical tests, the book seeks to offer a different understanding to conventional explanations for particularly salient cases, including the seemingly benevolent conduct of the Western democracies in World War II, as well as placing the Katyn massacre in a broader context of wartime conduct.

Expressing the desire for more “brittle” propositions is understandable in light of Schiemann’s own important work looking specifically at torture. The dialogue between our two books highlights the complementary and divergent ways to study such controversial topics, and I hope that our exchange encourages further research on these questions.