Putting Severity of Punishment Back in the Deterrence Package

Studies of criminal deterrence usually show an effect of certainty of punishment but often fail to find an effect of the severity. This is a serious threat to the theoretical underpinnings of deterrence theory. Through both a survey of 39 analyses in 33 published studies and our own reanalysis of an often-used data set, we show the problem rests not with the theory but with the analysis of the theory. Finding no severity effect can be traced to “unbundling the theoretical package” when moving from the theory to the statistical models used to represent the theory.