On the robustness of key player assessment in co-offending networks based on police registry data

• In Event: Advances in quantitative methods 2

Abstract

When offenders work together in teams of varying composition, over time elaborate co-offending networks can arise. Both criminologists and law enforcement agencies use these networks to pinpoint the key players in the network, typically by using measures of centrality. These key players are than ascribed importance in the network or are prioritized for law enforcement intervention. Information on co-offending however is usually derived from police registries, which raises questions on the level register data are representative of actual offending behavior. Due to the covert nature of criminal collaboration, register data are likely to underrepresent actual co-offending ties. However, register data may also overrepresent co-offending ties when in actuality one or more suspected offenders is not involved in a particular crime. While criminologists are well aware of these potential biases in police data, their potential impact on centrality measures derived from these networks is often ignored. Using police data on 40 individuals suspected of involvement in serious (drug)crime by local police, and their registered co-offenders, we estimate the robustness of key player assessment under different levels and processes of potential bias.

Authors

• Arjan Blokland, NSCR

• Ida Adamse, NSCR