A Preliminary Assessment of Small World Scholarship Networks in Criminology and Criminal Justice

Abstract

The present study applies small world network analysis to 727 articles from Criminology, Justice Quarterly, and the Journal of Research in Crime and Delinquency to assess scholarship collaboration trends within criminology and criminal justice (CCJ). Findings indicate that CCJ scholars tend to collaborate (versus sole author) on a great number of peer‐reviewed efforts, with several scholars collaborating with well over 10, and as many as 40, unique (non‐redundant) co‐authors. Consistent with the structure of scientific collaboration networks where scientists are separated by short paths of intermediates, Alex R. Piquero is found to be the most collaborative scholar in this sample and also the best center (central vertex) in the collaboration graph by linking to other scholars, on average, by only 3.6 degrees of separation. To further illustrate combinatorial patterns among CCJ scholars, this study also offers a descriptive and graphical analogue to the “Erdös number” from mathematics (here, the “Piquero number”), focusing on co‐authorship branching from first‐tier collaborators. In our view, small world analysis holds promise in better understanding far‐reaching collaboration patterns by the CCJ professoriate.

Acknowledgments

This research was supported by a student research assistantship from the College of Arts & Sciences, Seattle University; the authors thank Laura Polson for her assistance.

Notes

1. The articles cited above inform several of these underexamined questions, albeit generally in a narrative (non‐empirical) format. Also see Boyer (1990) and Kunselman, Hensley, and Tewksbury (2003).

2. Milgram was not without his critics. Of particular note is Kleinfeld (2002), who documents low completion rates and other potential biases based on her investigation of Milgram’s earlier unpublished experiments (from the Yale archives) and casts serious doubt on the validity of the reported results and conclusions. Replications of the Milgram experiment were actually not as numerous as one might think, limited to one by Milgram himself (Korte and Milgram 1970) and four or five other studies in the 1970s using limited populations (see Kleinfled 2002 for a review), and generally suffering from similarly low completion rates. A more “modern” approach using email instead of postal mail, while also suffering from low completion rates, did corroborate Milgram’s reported average chain length (Dodds, Muhamad, and Watts 2003).

3. While all of these studies relied on additional common indicators of the nature and scope of social networks, Moody’s (2004) analysis is particularly instructive as he used the comparison random networks suggested by Watts and Strogatz to evaluate the values of C and L. Moody (2004) investigated scholarship networks in sociology during the periods 1963–1999 and 1989–1999, and found there was not empirical support for a small world: the actual value of C was less than C for a random network in both cases, suggesting a lack of clustering, and actual L was higher than L for a random network in both cases, suggesting higher average chain lengths.

4. As Cormen, Leiserson, Rivest, and Stein (2001) explain BFS is among the most straightforward algorithms for searching a graph (here, a database of journal article entries). In the present context, this entails searching all neighboring nodes (collaborators) from a root node (specific publication authors), out to all unexplored neighboring nodes (other authors, linked by degrees of separation).

5. Additional co‐authors have been omitted for readability. They are Michael G. Turner in “Testing Moffitt’s Account of Delinquency Abstention,” Paul Mazerolle in “Student Anger and Aggressive Behavior in School,” Nancy A. Morris in “Risk‐Focused Policing at Places,” Laura A. Wyckoff, John E. Eck, Joshua C. Hinkle, and Frank Gajewski in “Does Crime Just Move Around the Corner?,” and Cynthia Lum and Sue‐Ming Yang in “Trajectories of Crime at Places.”