9422272 Kohfeld The overall objective of this research is to explain both long- and short-run changes in the temporal and spatial distribution of homicide in St. Louis over the period 1960-1994. Two hypotheses frame the specific aims of the project: the diffusion hypothesis, which accounts for variability in homicide on the basis of interpersonal processes that directly link homicide events to one another, and the independent-generation hypothesis, which accounts for variation in homicide in terms of changes in social-ecological conditions. Individual homicide events are coded in detail from police records for the 35-year period of study to provide a highly differentiated classification of homicide events. These events are aggregated for selected time intervals, for varying spatial units, and for varying homicide types. The analytic strategy is to conduct repeated cross-sectional analyses over systematically varied units of time and space for homicide events classified according to characteristics of victims and offenders (race, gender, age) and incidents (gun. drug, gang, victim-offender relationship). An innovative feature is the use of geographic information system software as a fundamental component of the measurement of both homicide events and other measures used in prediction.. The analyses are carried our with statistical models appropriate for the highly skewed distributions of events such as homicide. The completed data set will include more than 5.000 homicides that have occurred in St. Louis since 1960, data from the census years 1950-1990, and data from other sources designed to measure related behavioral processes. The research will produce a comprehensive account of the spatial and temporal structure of urban homicide that can advance scientific understanding of a serious national problem. %%% This is a study of the distribution of homicide in St. Louis over the period 1960-1994. It examines that distribution over both time and space to test two differing hypotheses advanced to explain the distribution and occurrence of murders. The diffusion hypothesis explains homicide as the result of interpersonal behaviors that directly link one homicide with another. The independent-generation hypothesis explains homicide as the result of changes in socioeconomic conditions. To investigate the hypotheses, this study will code the characteristics of St. Louis homicides in great detail and then aggregate homicide and their characteristics across various periods of time and units of space. The analysis of these aggregated homicide patterns will involve repeated cross-sectional analyses for the most salient characteristics of the homicides, characteristics such as traits of victims and offenders (e.g., race, gender, age) and of the homicide incident (e.g., whether the incident involves guns, drugs, gangs, or certain victim-offender relationships). Geographic information software will play an important role in organizing, displaying, manipulating, and analyzing the homicide data. The analysis will be based on more than 5,000 homicides. The data set to be produced will include the data on the 5,000 homicides plus geographically- and temporally-organized data from censuses and other sources relevant to measuring behavioral processes related to explaining homicide. The statistical analyses will be tailored for skewed distributions such as those characterizing homicide data. The research results will provide a comprehensive account of the structure or urban homicide that will advance scientific understanding of this serious national problem. ****
