From agent to action: The use of ethnographic social simulation for crime research

Abstract

This paper proposes a methodology for grounding agent-based social simulation in ethnographic data, using the example of crime research. The application of computational tools in crime research typically entails a removal of the “intelligible frame” of criminal behaviour and, hence, of meaningful evidence. Ethnography is a microscopic research tradition geared towards the preservation of contextualized meaning deemed essential for the exploration of the variety of prospective alternative scenarios and, hence, of plausible futures. On the basis of exemplary empirical material from a qualitative study on the transit trade of cocaine in the Netherlands, this paper looks into the complementarity and potential integration of the research traditions of ethnography and agent-based modelling. That is to say, it explores the compatibility of the formal languages of both these domains and the mutual benefit of “stitching together” these at first sight very different methods. The ethnographic approach to social simulation specifies the what-if relations of traditional/conventional ABM modelling into condition-action sequences. As we contend, it is exactly this more microscopic level of condition-action sequences that is needed to facilitate ”thick description” and, in turn, enable the grounding of ABM in meaningful evidence.