and control of violence. Because there appears to be little consensus in the social science literature on how multilevel factors should be defined, we describe in more detail our conceptualization of these terms.
Our use of the terms ''individual," "situational," and "community" corresponds closely to Short's (1985) definition of the "individual," "micro-," and "macro-" levels of analysis (see also Short, 1990:11). The individual level of explanation typically inquires as to characteristics of individuals that explain behavior. In our discussion of individual-level risk factors for violence, we specifically focus on the ascribed and achieved characteristics of individuals that are statistically associated with violent victimization and offending. For example, we examine how the risks of violent victimization and offending are distributed across characteristics such as age, sex, race, marital status, lifestyle, and socioeconomic status. Our intention, though, is not only to provide a descriptive summary of statistical findings, but to suggest how, and in what ways, individual characteristics are causally related to the risk of violent victimization and offending.
By situational-level risk factors, we are referring to those factors, broadly defined, that influence the initiation or outcome of a violent event. This conceptualization corresponds to Short's (1990:11) definition of the microlevel, where attention is focused on the unfolding of events and the interaction of parties involved in events. Situational-level analyses usually treat the violent incident, or event, as the unit of analysis. Included in our review is a discussion of such factors as the presence and type of weapon, the presence of drugs or alcohol, the role of bystanders or third parties during violent events, and victim resistance and retaliation. However, we expand the traditional conceptualization of situational-level analyses to include a discussion of the victim-offender overlap and how victim-offender relationships are related to violence. These factors are included because we believe that the simultaneous consideration of victims, offenders, and their past and present interactions provides a more complete context for studying the initiation and outcomes of violent events.
The macrosocial or community level of explanation asks what it is about community structures and cultures that produces differential rates of crime (Byrne and Sampson, 1986; Bursik, 1988; Short, 1990:11). For example, what characteristics of communities are associated with high rates of violence? Are communities safe or unsafe because of the persons who reside in them or because of community properties themselves? Can changes in community