Survivability not Superiority: A Critique of Kroenig (2013)

Kroenig (2013b) finds that in crises between nuclear-armed states, countries possessing nuclear arsenals larger than those of their opponents tend to be victorious. After correcting for coding errors in the dataset and for finite-sample bias in clustered standard error estimates, we show that the original conclusion is too over-confident. We further demonstrate that the association between nuclear superiority and crisis victory is extremely sensitive to the author’s variable coding decisions and model specifications. Using a new method for evaluating coding uncertainty, we find that under reasonable alternative choices, the nuclear superiority finding is no longer statistically significant at all conventional significance levels. We find instead that the possession of an assured nuclear second-strike capability is consistently and robustly associated with positive crisis outcomes among nuclear states. Survivability, rather than superiority, appears to be the element of a state’s nuclear arsenal that has the most significant bearing on its ability to win nuclear crises. However, we urge caution in drawing any strong conclusions from the Kroenig dataset due to sample size and model dependence issues. ∗Harvard University, Department of Government, [email protected] †Harvard University, Department of Government, [email protected] ‡Corresponding Author, Harvard University, Department of Government [email protected]