Abstract

Co-offending is the act of committing a crime in a group. While it involves some risks, criminals engage in co-offending to achieve specific goals and needs, such as receiving physical assistance during the crime or learning new skills to commit more sophisticated crimes. The interactions between co-offenders can be represented by social networks and one of the mechanisms of criminal network formation is linked to homophily, which is the tendency of individuals with similar characteristics to connect with each other. Offenders who commit crimes together tend to be similar in terms of age, sex, and race. However, it is still not clear how partner selection and homophily affect co-offending outcomes. The aim of this study is to explore how similarity-based partner selection affects the structure of criminal networks and, consequently, the co-offending outcomes. To achieve this, we have developed an Agent-Based Model that simulates a population of offenders. By manipulating the level of homophily in the environment, we are able to track the differences in the number of connections in the network, the strength of these connections, along with the total number of co-offenses, and the average number of skills acquired.