It’s Not Who You Know—It’s What You Know About People You Don’t Know That Counts

Abstract

This paper raises two difficulties with an earlier contribution to this Journal (Coles 2001) and proposes solutions. While we strongly support his suggestion that neglect of Social Network Analysis (SNA) has hampered criminology, and welcome his opening up of the debate, we argue that Coles has failed to make an important distinction between types of SNA data collection and presented a flawed 'theoretical framework'. In the first section of this article, we highlight the difference between egocentric and non-egocentric network data collection. In the former, respondents are asked about their own network ties, while in the latter, they are also asked what they know about the ties of others. It transpires that egocentric data suit qualitative surveys while non-egocentric data harmonize with ethnography. While Coles endorses 'qualitative' SNA, his failure to make a clear distinction between egocentric and non-egocentric data means he confuses criticisms appropriate to two different data collection techniques. In the second section of the article, we argue that the 'theoretical framework' which Coles presents is problematic. He discusses a number of extant findings about the structure of non-criminal networks but fails to support the claim that they can be applied to criminal networks. Such support could either be empirical - showing that the two types of networks were alike - or model based - showing that relevant social interaction mechanisms actually produce similar structural properties in networks. In the absence of either kind of evidence, however, there are ethnographic reasons to expect that criminal networks will have both distinctive generative mechanisms and structural properties. These reasons are discussed in the context of examples provided by Coles. The final section of the paper uses computer simulation to address the two problems raised. The first model shows the conditions under which collection of qualitative (non-egocentric) and quantitative (egocentric) data will produce a more accurate picture of the underlying network. The second (outline) model shows how simulation might link ethnographic data on social interaction to the aggregate properties of networks. The conclusion discusses the consequences of this approach for novel research and criminal intelligence.