Understanding criminal groups as social networks has led to the design of powerful systems for decision support in criminal investigative work. Tools using the methods of social network analysis have proven particularly effective in the identification of associations between individuals whose relationships are not otherwise evident. This identification is typically based on the links between individuals and does not account for other relevant information, such as individual attributes. The present study proposes a new model for identifying criminal associations that incorporates this type of data. Built around a linear association model, this approach identifies the principal association between two individuals. Assuming one of the individuals as the crime planner, the approach can be used to maximize his/her utility function. The model is compared with an existing algorithm for identifying associations using a real dataset provided by the Public Prosecutor's Office of Region del Biobio-Chile. The results demonstrate the proposed model's effectiveness and flexibility in generating different association alternatives, a particularly useful feature that contributes to the more efficient use of criminal investigation resources.