Co-offending in Burglary: A Network Approach

Open Access
Lantz, Brendan Gregg
Graduate Program:
Crime, Law and Justice
Master of Arts
Document Type:
Master Thesis
Date of Defense:
October 25, 2013
Committee Members:
  • Barry Ruback, Thesis Advisor
  • Burglary
  • Co-Offending
  • Networks
Burglars commonly offend in groups, but even when they work alone they often learn about possible targets through networks of friends and relatives. The present analysis employs a network analysis perspective to investigate co-offending in burglary. Three questions are considered. First, who commits burglary offenses alone, and who chooses to offend in groups? Second, what are the conditions associated with group offending, and what types of burglary offenses are committed by co-offending groups? And finally, how do characteristics of the co-offending group influence burglary offending and target selection? These questions are investigated using data from the Pennsylvania Commission on Sentencing (PCS) and the Centre County Probation and Parole Department. Information on offenders is recorded, as well as ties between offenders. Offender and offense networks are visualized and analyzed. Finally, several of the network measures are then incorporated into Hierarchical Linear Models (HLM) predicting three offending outcomes: value of goods stolen, successful offense completion, and total time until arrest. Consistent with past research, the results demonstrate that young offenders are more likely to offend with others, and with more total partners. While the typical burglary offender is young, the typical co-offender is about 23 years old, roughly 4 years younger than offenders who commit crimes alone. The instrumental perspective on co-offending, which posits that offenders choose to offend with others in order to increase offense success and reduce offense risk, was supported by the results. Offenses involving more than one offender are more successful than offenses involving solo offenders. Theft value is significantly increased for offenses involving multiple offenders. Co-offenses are also significantly more likely to be completed successfully (i.e., without interruption by residents or police). Group density, a measure of group cohesiveness, plays an important role in offense outcome. Subgroups of higher density are more likely to be successful than groups of lower density. Dense groups are more likely to complete offenses than other groups, as well as to steal goods of higher value. These effects of density are probably due to the likelihood that subgroups with more cohesion and trust are able to function as a group more successfully. Finally, the use of a two-mode network structure permits a unique examination of repeat victimization offenses. The results demonstrate that group processes within the network are particularly important in the occurrence of multiple victimization offenses. Three possible mechanisms of repeat victimization are tested: (1) characteristics of a location make the target attractive to all motivated offenders, (2) the same offender returns to the offense location multiple times, or (3) the primary offender notifies other offenders, to whom he is tied, about the target. The results are consistent with the third explanation. Group processes appear to play a significant role in multiple victimizations. Multiple victimization offenses are significantly more likely to be committed by offenders from within a subgroup, compared to isolates. Further, most repeat victimization offenses are committed by offenders within the same subgroup, rather than from different subgroups. Taken together, the results suggest that group processes are a particularly important mechanism of repeat victimization. The results suggest two implications for theory. First, the instrumental perspective of co-offending is supported. Offenders who offend with others are more likely to be successful, suggesting that the offender decides whether or not to commit co-offenses based on a rational choice framework. Second, the prevalence of subgroups across burglary offenders suggests that interaction is important. A significant network is shown to be operating in Centre County, suggesting the prevalence of even larger networks in areas of higher population density and higher crime rates. The network was also constructed using sentencing data, and might be much larger if additional data on non-sanctioned offending were considered. Finally, the present study has three policy implications. First, co-offending is important, and is conceptually different from solo offending. Co-offenses appear to be more serious, as higher valued goods are stolen, and are more likely to be completed successfully in the short term. Thus, from a crime control perspective, criminal justice resources need to be directed at co-offending groups. Second, the prevalence of group offending makes clear the importance of examining interaction processes in offending. Understanding how information is shared among group members and transmitted to other groups would be important for intervening in order to prevent crimes. Third, in terms of apprehension, arrests are made in a shorter amount of time for multiple offender offenses compared to solo offenses. But, in addition to the arrest, group membership must also be considered. If group membership is not considered, arresting only one offender may lead to little to no effect on the overall crime output of the group.