The deterrent effect of executions on murder: effect of different ways of measuring execution rates.

Abstract

Several studies of the states of the USA have been conducted to see whether executions have a deterrent effect on murder. Results conflict, sometimes showing a deterrent effect (e.g., Black & Orsagh, 1978) and sometimes not (e.g., Bailey, 1977). These studies typically have used as a measure of execution rate the number of executions per murder in th'e state. Investigators have focused in the past on execution rates per murderer since they have been interested in whether the probability facing a murderer that he will be executed affects his decision to murder. Focusing on the execution rate per capita of the population or on the absolute number of executions (which is perhaps a crude index of the publicity given to executions) directs interest more to the demand of the public for executions to satisfy their desires for the punishment of criminals. The present note reports an examination of the relationship between different measures of executions and the murder rate. Since executions are usually publicized, the absolute number of executions in a state may be a suitable measure of the arnounr of media publicity given to executions in the state. In addition, the public demand for executions could be independent of the supply of murderers, and execution rates could be expressed instead per capita of the general population rather than per murder. The data set used was that of Cloninger (1977) who presented data for the 4 1 continental states in which there was a death penalty in effect from 1955 to 1959. Both the number of executions and the rate of executions per capita of the general population were correlated positively with the murder rates (Pearson rs = 0.35 and 0.53) but the execution rate per murder was not associated with the murder rate ( r =0.01). We replicated Cloninger's multiple regression analysis using the execution rate per murder and also repeated the analysis using the execution rate per capita instead. The deterrent effect of executions emerged from both regressions. The beta coefficient for the execution rate per murder was -0.20 ( p = .01), while the beta coefficient for the execution rate per capita was -0.18 ( p = .O6). Although the magnitude of the former beta was greater than the latter, the difference was small and suggests that the deterrent effect of executions can be illustrated with the use of either measure of the execution rate, at least in this data set.'