Testing a Bayesian learning theory of deterrence among serious juvenile offenders

The effect of criminal experience on risk perceptions is of central importance to deterrence theory but has been vastly understudied. This article develops a realistic Bayesian learning model of how individuals will update their risk perceptions over time in response to the signals they receive during their offending experiences. This model implies a simple function that we estimate to determine the deterrent effect of an arrest. We find that an individual who commits one crime and is arrested will increase his or her perceived probability of being caught by 6.3 percent compared with if he or she had not been arrested. We also find evidence that the more informative the signal received by an individual is, the more he or she will respond to it, which is consistent with more experienced offenders responding less to an arrest than less experienced offenders do. Parsing our results out by type of crime indicates that an individual who is arrested for an aggressive crime will increase both his or her aggressive crime risk perception as well as his or her income-generating crime risk perception, although the magnitude of the former may be slightly larger. This implies that risk perception updating, and thus potentially deterrence, may be partially, although not completely, crime specific.