Abstract

A relationship between trends in the age structure of the population and homicide rates has been systematically observed only in the United States since World War II. This study analyzes data from five nations over 70 years to examine the generalizability of this relationship. Using two time-series techniques, we find that changes in the proportion of young males in a population do not exert consistent effects on homicide rates. There is no evidence of an age effect in Scotland and Japan, and strong evidence of a postwar age effect only in the United States and Italy. Based on similarities in the character of homicide in the United States and Italy, we suggest a set of conditions under which an age effect is most likely to obtain. Debates over the invariance of the age distribution of crime and the relative effects of age, period, and cohort on crime rates continue to generate a steady stream of criminological research and are far from resolved. Yet one type of age-crime relationship has remained relatively uncontroversial, namely, that between trends in the age structure of a population and that population s crime rates. As the proportion of young adults in the population increases, so do rates of common crime. Several analysts have referred to the apparently self-evident nature of this relationship before tackling the more complex ways in which age and crime may be related (O'Brien 1989; Wilson & Herrnstein 1985:140). Easterlin (1987:6), for example, briefly cites the "widely recognized " fact that "changes in the age distribution of the population have important consequences for national crime rates " and then turns to an examination of the criminogenic effects of relative cohort size. This article demonstrates that the presumed "well-established " (O'Brien 1989:57) temporal relationship between age structure and at least one type of crime is neither ubiquitous nor unproblematic. To do this, we analyze lon-gitudinal data on homicide and age structure from five nations beginning early *Earlier versions of this article were presented at the 1988 annual meetings of the American