In 2001, Coles presented an analysis suggesting that neglect of Social Network Analysis (SNA) would hamper criminology. Coles argued that the analysis failed to make an important distinction between types of SNA data collection and presented a flawed theoretical framework. By providing no clear exposition of different data collection methods in SNA, Coles failed to identify a crucial distinction between access to egocentric and non-egocentric network data. The article argues that Coles failed to distinguish between describing what occurs in networks with given structures and what social mechanisms exist to generate networks. Simulation is then used to explore how these difficulties might be resolved by making a clear distinction between qualitative (non-egocentric) and quantitative (egocentric) approaches to network data collection. In summary, the article attempts to show that the crucial distinction between qualitative and quantitative SNA is the use that is made of non-egocentric data, that under certain circumstances, qualitative sampling performs better despite its inaccuracy and that in order to understand the generative mechanisms that produce particular network structures, a combination on ethnography and simulation shows potential. References