The present paper examines secular trends of homicide rates by means of a systematic re-analysis of all available quantitative studies on pre-modern homicide. The results confirm, first, that homicide rates have declined in Europe over several centuries. Second, the empirical evidence shows, that unequivocal decline began in the early seventeenth century. Third, the data indicate that the secular decline begins with the pioneers of the modernization process, England and Holland, and slowly encompasses further regions. These findings corroborate much of the civilizing process framework proposed by Norbert Elias. Yet, the diffusion of self-control was sustained not only by compliance to the state monopoly of power but by a variety of disciplining institutional arrangements. This includes, for example, the early expansion of schools, particularly in Northern Europe, the rise of religious reform movements, and the organization of work in manufacturing. Second, while social disciplining certainly is the central feature of the early modern period, it also served to push forward the rise of the specifically modern individualism that Durkheim sees as the cause of the decline of individual-level violence. About 20 years ago, in an article that has become highly acclaimed among historians of crime, Ted Robert Gurr (1981: 295) examined empirical evidence on secular trends in lethal criminal violence. Gurr primarily reviewed a number of historical studies on homicide in medieval and early modern England, each based on detailed analyses of records for specific periods and specific jurisdictions. While not originally concerned with computing homicide rates, the studies did provide counts that could be used for estimating the rates. Gurr plotted the some 20 estimates for the period from about 1200 to 1800 on a graph, added the London homicide rates for the modern period, and fitted an elegant S-shaped trend curve to the data points. It starts at about 20 homicides per 100,000 population in the high and late Middle Ages and ends after an extended downswing at about one homicide per 100,000 in the twentieth century. Gurr interpreted this secular trend as 'a manifestation of cultural change in Western society, especially the growing sensitization to violence and the development of increased internal and external control on aggressive behavior' (Gurr 1981; also see Gurr 1989). Since then, an impressive amount of research in the history of crime and criminal justice has greatly enlarged knowledge about historical manifestations of lethal violence. It has shown that the history of violence has to be firmly situated in the context of social and cultural history and the long-term development of the core institutions of modern 618

Total hours of work and sleep did not change after implementation of the duty hour standards. Although fewer residents were burned out, rates of medication errors, resident depression, and resident injuries and educational ratings did not improve.

Recent studies suggest that the reported effect sizes of prevention and intervention trials in criminology are considerably larger when program developers are involved in a study than when trials are conducted by independent researchers. This paper examines the possibility that these differences may be due to systematic bias related to conflict of interest. A review of the evidence shows that the possibility of a substantial problem cannot be currently rejected. Based on a theoretical model about how conflict of interest may influence research findings, the paper proposes several strategies to examine empirically the extent of systematic bias related to conflict of interest. It also suggests that, in addition to improved standards for conducting and publishing future experimental studies, more research is needed on the extent of systematic bias in the existing body of literature.

Previous studies on cross-national patterns of crime and problem behavior have focused primarily on homicide. This article proposes that cross-national research should pay more attention to broadly based measures of various types of problem behavior. By combining different types of sources, I derive measures for four types of problem behavior, namely violent crime, property crime, alcohol abuse, and drug use for a sample of thirty-seven countries. Analysis of these data first shows that, at the level of cross-national comparison, different manifestations of problem behavior do not constitute a single underlying dimension. Rather, a cluster analysis reveals several groups of countries with similar configurations of problem behavior. Many Anglo-American countries, for example, were found to belong to a cluster with a high likelihood of various kinds of problem behavior associated with the transition from adolescence to early adulthood. High levels of violence characterize many Eastern European countries. Further analyses show that distinct types of problem behavior correlate with different contextual variables. Violence is found to be high in countries characterized by great social inequality, low levels of social control, and widespread material poverty. Drug use and alcohol abuse among young people, in contrast, is frequent in highly urbanized, highly affluent contexts where lifestyles are leisure-time oriented.