NCJ Number
206359
Date Published
July 2004
Length
23 pages
Annotation
This working paper, examines the methodological and analytical problems of research exploring the deterrent effect of capital punishment.
Abstract
Research on the deterrent effect of capital punishment has a long history by many diverse disciplines. Recently, papers by economists have called into question previous econometric analyses that found deterrent effects because of fatal specification errors found in these studies. This article examines the data frequently used by researchers to examine the deterrent effects of capital punishment. The data were a pooled cross-section time series of 50 States during the period 1977 through 1997; there were a total of 1,050 observations, with each observation representing 1 year within a State. Key response variables were the number of homicides per State per year and the homicide rate per 1,000 residents. The author empirically analyzes the data with a 1 year lag, leaving 1,000 observations to analyze. As a result of various bivariate and multivariate analyses, the author argues that these data are not empirically able to deliver the answers asked by researchers. No negative relationship was observed between the number of executions lagged by 1 year and either the homicide rate or the number of homicides. Thus, for the majority of States, for the majority of years, there was no evidence that capital punishment deterred homicide. Only the appearance of deterrence was reached by imposing a linear relationship between executions and either the number of homicides or the homicide rate, and by including the most extreme observations in the analyses. However, the author argues that the linear relationship is contradicted by the data and the deterrent effect disappears when the extreme cases are excluded. The author concludes that the available data are insufficient to be able to determine the deterrent impact of capital punishment. Figures, references