Classifying Drug Markets by Travel Patterns: Testing Reuter and MacCoun's Typology of Market Violence

Abstract

Research to date has demonstrated significant relationships between the presence of outdoor drug markets and violent crime. Scholars have neglected however, to consider the role of travel distance on the drugs/violence nexus. The current study examines whether features of the distributions of travel distance to markets of drug buyers, drug sellers, or the interaction between the two distributions predicts drug market violence levels net of surrounding community demographic structure. Reuter and MacCoun's (1992) as yet untested model about the connections between drugs and violent crime, predicts that the interaction of drug seller and buyer distance distributions from varying distances more powerfully drug market violence levels than buyer and average distance averages. This suggests that how the travel patterns of the two major participants in drug markets intersect is key to understanding differences. That model is tested here. In addition, for comparison purposes, impacts of buyer and seller travel median distances are modeled separately. This work uses 5 years (2006-2010) of incident and arrest data from the Philadelphia Police Department. Reuter and MacCoun's model will be tested using the following analytical techniques. First, a methodology for locating and bounding drug markets using a nearest neighbor, hierarchical clustering technique is introduced. Using this methodology 34 drug markets are identified. Second, hierarchical linear models examining buyers and sellers separately predict travel distances to drug markets. Arrestees are nested within markets. This technique separates influences on distance arising from arrestees from drug market distance differences. Third, how market level median travel distance affects within drug market violence is considered. Specifically, the main effects of median buyer travel distance and median seller travel distance on drug market violence are captured using separate Poisson hierarchical linear models. Finally, impacts of the interaction between buyer and seller distance, Reuter and MacCoun's (1992) focus, are explored in another series of generalized hierarchical linear models. The main findings from the dissertation are as follows: 1. Results provide partial support for Reuter and MacCoun's drug market-violence model using multiple operationalizations. Public markets--those in which buyers and sellers travel from outside their own neighborhoods--are expected to be the most violent. 2. Separate raw distance measures for buyers and sellers correlate with within-drug market violence, after controlling for community demographics. 3. A negative effect of socioeconomic status and violence holds even when modeled with drug market variables. 4. As the proportion of crack cocaine sales within drug markets increases so too does within-market violence. Conceptual implications highlight the need to investigate social ties as an intervening variable in the travel distance »» drug market violence relationship. It is not clear from this research whether the travel distances of drug offenders in some way explains the amount or strength of social ties in a drug market, which in turn serves to suppress or elevate within-drug market violence. Policy implications suggest that Reuter and MacCoun's drug market types may connect with specific policing responses. Policing efforts may not receive much support from community residents because dense social networks may discourage reporting illicit activity. Markets drawing dealers and customers from farther away, and located around commercial and recreational centers may be amenable to place-based policing initiatives and coordinated intervention strategies with multiple city agencies.