Social networks — the webs of social relationships that connect individuals to each other and through which information, resources, influence, and behavior flow — have emerged over the past two decades as one of the most productive analytical frameworks in criminology. The social network approach does not replace the structural, institutional, and normative frameworks reviewed in the preceding articles but extends them by specifying the relational pathways through which structural conditions, institutional experiences, and normative influences reach specific individuals. Where ecological research asks why crime concentrates in specific neighborhoods, network research asks whose relationships within those neighborhoods concentrate the risk. Where social learning theory asks how pro-criminal definitions are transmitted through groups, network research asks through which specific ties those definitions flow and how network position shapes their reception. The social network perspective makes the relational micro-structure of criminal behavior visible in ways that both aggregate structural analysis and purely individual analysis cannot achieve.
Sociology and Criminology has engaged network analysis most consequentially through the epidemiological work of Andrew Papachristos and colleagues, who documented that gun violence concentrates in specific co-offending network clusters in ways that reveal both the causal mechanisms of violence transmission and the intervention leverage points that targeted approaches can exploit. But the network perspective’s applications in criminology extend well beyond violence research to encompass co-offending network analysis, organized crime structure, the social network dimensions of drug market organization, the network transmission of criminal careers, and the network mechanisms of desistance. Each application demonstrates both the distinctive analytical leverage that network methods provide and the importance of integrating network analysis with the structural and behavioral frameworks that criminology has developed through its longer research traditions.
Introduction
Social network analysis (SNA) brings to criminology a set of concepts and methods developed primarily in sociology and organizational studies but productively applicable to the study of crime and criminal justice. The basic unit of analysis in SNA is the relationship rather than the individual: rather than characterizing individuals by their attributes (demographics, attitudes, criminal history) and predicting behavior from those attributes, network analysis characterizes the pattern of relationships among individuals and asks how network position — where in the web of social relationships each individual sits — affects behavior and outcomes.
Key network concepts include: nodes (individuals or organizations in the network); ties (relationships between nodes, which may be directed or undirected, weighted or unweighted); degree (the number of connections each node has); centrality (various measures of how central each node’s position is in the overall network structure); clustering (the tendency for one’s contacts to also be connected to each other); and structural holes (positions in networks where an individual bridges otherwise disconnected subgroups, creating unique information and influence advantages).
These network structural properties predict behavior in ways that individual attribute analysis cannot capture. An individual who is highly central in a criminal network — connected to many others, many of whom are themselves well-connected — faces different opportunities and risks than an individual who is peripheral, regardless of their individual attributes. An individual who bridges criminal and conventional networks — occupying a structural hole between two otherwise disconnected social worlds — faces unique pressures and opportunities that neither purely conventional nor purely criminal network embeddedness produces. And an individual embedded in a tightly clustered, heavily armed criminal network faces different violence risks than one embedded in a loosely connected, lightly armed network, regardless of their individual propensity for violence.
Co-Offending Networks
The Structure of Criminal Partnerships
Co-offending — the commission of crimes by multiple individuals acting together — is far more common than official statistics suggest, with research estimating that 30–50% of all offenses involve multiple participants. The network of co-offending relationships — who has committed crimes with whom, how often, and across what crime types — has been analyzed using network methods in multiple jurisdictions with consistent findings about the structural properties of criminal networks. Research by Sarnecki in Stockholm, by Warr in the United States, and by Bouchard and colleagues in British Columbia has documented that co-offending networks tend to be sparse (relatively few ties among potential co-offenders), clustered (existing ties tend to be localized in small groups), and scale-free (a small number of highly connected individuals participate in a disproportionate share of all co-offending events).
The developmental implications of co-offending network position are significant for criminal career theory. Research by McGloin and Piquero found that co-offending with more experienced offenders — occupying a position as a junior partner to a higher-centrality criminal network member — substantially elevated subsequent solo and group offending, establishing the network mechanism of criminal career escalation. This finding is consistent with social learning theory’s prediction that exposure to experienced criminal partners transmits criminal techniques and definitions, but it specifies the mechanism precisely as a network position effect rather than merely a social exposure effect. Research on the network dynamics of criminal career onset and desistance has found that entry into criminal networks occurs through the same relationship formation processes that shape all social networks — through geographic proximity, shared institutional experiences, and existing social ties — while exit from criminal networks is often blocked by the same relational processes, as established network ties continue to create criminal opportunity and social pressure long after individual motivation to desist has developed.
Table 1. Social Network Concepts and Their Criminological Applications
| Network Concept | Definition | Criminological Application | Key Findings |
|---|---|---|---|
| Degree centrality | Number of direct connections | Identifies prolific co-offenders; central gang members | High-degree nodes disproportionately involved in crime |
| Betweenness centrality | Position bridging otherwise disconnected nodes | Drug market brokers; violence network interrupters | Brokers disproportionately influential; vulnerable to disruption |
| Clustering coefficient | Extent to which neighbors are also connected | Gang cohesion; delinquent clique tightness | Dense clustering amplifies peer influence |
| Structural hole | Gap between otherwise disconnected subgroups | Opportunity for novel criminal innovation; enforcement leverage | Brokers of structural holes are high-influence targets |
| Network reach | Indirect connections through network path | Distance from violence events; future victimization risk | Network proximity to shootings predicts own victimization |
| Tie strength | Frequency, duration, and emotional intensity of relationships | Strong ties for learning criminal skills; weak ties for information | Strong ties predict criminal socialization; weak ties predict opportunity |
Network Analysis and Enforcement Strategy
The network perspective has direct implications for enforcement strategy that have been translated into operational police practice through the crime analysis units that incorporate network analysis into deployment decisions. Research by Papachristos and colleagues demonstrating that a small number of high-centrality individuals in criminal co-arrest networks account for a disproportionate share of violent events has supported the targeted enforcement component of focused deterrence strategies that identify and communicate with specific high-risk network members rather than saturating neighborhoods with undifferentiated patrol. The Group Violence Intervention’s notification meeting model — which brings the specific individuals identified as highest-risk network members to a meeting where law enforcement, social services, and community partners communicate focused messages — directly implements the network targeting logic.
Research evaluating network-targeted enforcement strategies has found that the precision of network targeting substantially improves the crime reduction per unit of enforcement resource deployed compared to either geographic hot spots targeting or demographic profiling approaches. A randomized trial by Weisburd and colleagues comparing network-targeted and geographically targeted intervention found that the network-targeted approach produced larger violence reductions per individual contacted, establishing the efficiency advantage of network-precise intervention over geographic approximations that include many low-risk individuals in their targeting.
Gun Violence as Social Contagion
Papachristos and the Network Epidemiology of Violence
Andrew Papachristos’s research program on the social network structure of gun violence — developed in collaboration with colleagues at Yale, Harvard, and Chicago — has produced what is arguably the most significant methodological innovation in violence research of the past two decades. His foundational studies in Boston and Chicago documented that gunshot victimization concentrates in specific co-offending network clusters — groups of individuals connected by prior co-arrest — in patterns suggesting that violence transmits along network ties rather than being randomly distributed across the neighborhood population. The key finding: individuals in the highest-risk network positions — those with the most prior co-arrests with individuals who had been shot — faced gunshot victimization risks 900 times higher than the general city population, a concentration so extreme that geographic targeting alone cannot explain it (Papachristos et al., 2012).
The network transmission mechanism — violence spreads along social ties through retaliation, escalation, and the co-offending relationships that produce shared criminal exposure — provides the most precise causal account of gun violence concentration available in urban criminology. Rather than treating violence as a neighborhood-level phenomenon produced by ecological conditions, the network perspective reveals it as a relational phenomenon produced by specific social relationships — relationships that are partly determined by ecological conditions (concentrated disadvantage concentrates the risky network positions) but that have their own causal logic at the network level. A person’s risk of being shot in Chicago is better predicted by their network distance to prior shooting victims than by their neighborhood of residence, their demographic characteristics, or their own prior criminal history.
Contagion, Retaliation, and Network Amplification
The epidemiological framing of violence as social contagion — proposed by Gary Slutkin and elaborated by Papachristos — treats violence as spreading through social contacts in ways analogous to the transmission of infectious disease, with individual “exposures” to violence in one’s social network elevating the probability of subsequent violent involvement. Research testing this framework using time-series analysis has found that violent events produce predictable clustering of subsequent violence events in the same network clusters, consistent with the contagion mechanism and inconsistent with purely ecological or deterrence-based accounts that do not predict temporally clustered, network-localized violence patterns.
The retaliation mechanism through which violence contagion operates has been documented in research on dispute escalation in gang-involved communities. Research by Jacobs and Wright on street robbery retaliation, and by Papachristos and colleagues on sequential shooting patterns, has documented that a substantial proportion of gang-involved shootings occur in retaliation for prior shootings involving network-connected individuals — establishing a specific social mechanism of violence transmission that community violence intervention programs specifically target by interrupting retaliatory cycles before they produce additional shootings. Research documenting that CVI interruptions of retaliatory sequences reduce subsequent shootings provides the most direct experimental-style evidence for the network contagion mechanism as a target for intervention.
Dark Networks and Intelligence Analysis
The application of social network analysis to the investigation and disruption of criminal and terrorist organizations — what researchers call “dark network analysis” — has become an important interface between academic criminological research and operational intelligence and law enforcement practice. Research on terrorist network structures, drug trafficking organizations, and money laundering networks has documented that these organizations face a specific organizational dilemma: they need to maintain enough connectivity to coordinate operations but must minimize connectivity to reduce detection and enforcement risk. Research by Milward and Raab, Krebs on terrorist networks, and Coles on drug trafficking organizations has found that the most resilient criminal networks achieve this balance through redundant connectivity (multiple pathways between key actors) combined with cellular compartmentalization (limiting any individual member’s knowledge of the broader network).
Law enforcement disruption strategies targeting dark networks have been evaluated for their effectiveness in network fragmentation and crime reduction. Research finds that enforcement strategies targeting high-centrality nodes (leaders) sometimes fragment networks and reduce crime but sometimes produce resilient reorganization as remaining members reconfigure connections around new brokers. Research by Bouchard and Morselli has found that the most effective enforcement strategies combine leadership targeting with targeting of the brokers who connect otherwise disconnected network segments — the structural holes positions through which coordination across network segments occurs — producing more comprehensive network disruption than leadership targeting alone. These findings have been incorporated into the multi-agency criminal intelligence analysis frameworks that federal and state enforcement agencies use for complex conspiracy investigations.
Organized Crime and Criminal Network Structure
From Hierarchy to Network: Organized Crime Analysis
The traditional sociological analysis of organized crime — developed through ethnographic and historical research on the Italian-American Mafia, Sicilian Cosa Nostra, and similar hierarchically organized criminal enterprises — emphasized hierarchical organizational structure, strong territorial organization, and ethnic solidarity as the defining features of organized crime. Contemporary research using network analysis methods has substantially revised this picture, finding that most organized criminal enterprises are more loosely networked than the hierarchical models suggested, with varying degrees of centralization, formalization, and ethnic homogeneity depending on the specific criminal market and historical moment.
Research by Morselli on criminal network structures, by Varese on the Sicilian Mafia, and by Kleemans and colleagues on Dutch organized crime has documented that criminal organizations adopt the network structures that their specific operational environments select for: hierarchical structures provide coordination efficiency but create vulnerability to enforcement decapitation; flat, distributed network structures are resilient to enforcement but face coordination challenges; and the specific balance between efficiency and resilience that different criminal organizations strike reflects both the enforcement environment they face and the specific criminal market in which they operate. Research on how law enforcement disruption affects criminal networks — finding that the removal of high-centrality network members (leaders) sometimes produces network fragmentation and crime reduction but sometimes produces resilient reorganization — has informed the network-targeted enforcement strategy that identifies specific high-leverage network positions for intervention.
Criminal Networks and Desistance
Network Ties as Barriers and Resources
The social network perspective illuminates the desistance process in ways that life course theory’s emphasis on individual turning points does not fully capture. For individuals embedded in active criminal networks, desistance requires not only the individual motivation for change and the institutional opportunities (employment, marriage) that life course theory emphasizes but also the social network changes — the weakening of criminal network ties and the strengthening of conventional network ties — through which the relational infrastructure of criminal career maintenance is dismantled. Research by Giordano and colleagues found that “hooks for change” — the social opportunities that enable desistance — required not only individual cognitive transformation but active engagement with new social relationships that provided alternative network memberships incompatible with continued criminal involvement.
Research by Decker and Pyrooz on gang exit has documented that the network embeddedness of gang membership creates specific barriers to desistance that policy must address: gang ties are not simply peer relationships that can be replaced by making better choices but are embedded in territorial, economic, and protective functions that require the development of alternative institutional ties before gang exit becomes feasible. Programs designed to support gang desistance — providing employment, housing, social services, and alternative social networks through credible messengers — address these network-level barriers to desistance rather than simply providing incentives for individual behavioral change.
Digital Networks and Crime
Social Media and Online Criminal Networks
The emergence of social media and digital communication platforms as primary environments for adolescent and young adult social life has created new network contexts for criminal behavior whose analysis requires extending the network analysis framework developed for offline social networks to the specific properties of online social network environments. Research documenting that gang members use social media platforms — primarily Instagram, Facebook, Twitter, and TikTok — to claim territory, display weapons, taunt rivals, and coordinate violent retaliation has established that online networks are not separate from offline criminal networks but are digital extensions of them, with the same network dynamics of rivalry, status competition, and violence transmission operating in both contexts simultaneously.
Research by Patton and colleagues on social media and gang violence in Chicago found that content posted on social media — including threats, provocation, and memorial posts for killed gang members — predicted subsequent shootings in the offline network, establishing a bidirectional relationship between online and offline network dynamics that both law enforcement and violence interrupters must address. The implications for violence reduction are significant: monitoring social media for escalating conflict in high-risk networks provides early warning of violence that traditional surveillance cannot detect, while social media interruption — connecting with high-risk individuals through social media platforms — extends the reach of community violence intervention beyond the street contexts where traditional outreach operates.
Research by Desmond Patton and colleagues using machine learning to identify conflict-escalating social media content among high-risk youth networks in Chicago has documented both the feasibility and the effectiveness of AI-assisted network monitoring for violence prevention — an application that raises important privacy and civil liberties concerns alongside its potential for violence reduction that criminological research must engage with as digital network analysis becomes an increasingly important dimension of the field.
Peer Networks and the Prevention of Criminal Career Onset
The social network dimension of criminal career prevention — keeping at-risk individuals out of criminal networks before they become embedded — is as important as the desistance research on how to exit criminal networks, and it is addressed through different policy mechanisms. Research on mentoring programs, after-school programming, and youth development organizations — all reviewed in the urban crime prevention article — addresses the network prevention function by providing prosocial network memberships that compete with the delinquent network memberships that developmental criminology and social learning theory identify as primary mechanisms of criminal career onset. The network logic of these programs — that providing access to conventional social networks reduces the probability of delinquent network embeddedness through simple opportunity competition — is rarely made explicit but is implicit in their design and consistent with the empirical evidence for their effectiveness.
Research by Haynie on network centrality and delinquency in adolescent friendship networks documented that high-centrality adolescents — those most central in their peer networks — showed both higher own delinquency and higher influence over their network neighbors’ delinquency, establishing that network position effects on both individual behavior and peer influence amplify each other. This finding suggests that prevention interventions that identify and engage high-centrality adolescents in prosocial activities — through the same network-targeting logic that focused deterrence uses for violence prevention — may produce network-level prevention effects larger than their direct effects on targeted individuals would predict.
Weak Ties, Bridging Capital, and Reintegration
Mark Granovetter’s foundational distinction between strong ties (close relationships with frequent interaction and high emotional investment) and weak ties (acquaintance-level relationships with infrequent interaction and limited emotional investment) has important applications to the criminological study of reintegration and desistance. Research by Granovetter found that weak ties were more valuable for job searching than strong ties — because weak tie contacts have access to different information networks than one’s close circle, providing novel opportunities that strong ties’ network overlap cannot provide. Applied to reintegration, this finding suggests that returning citizens seeking legitimate employment may benefit from weak ties to conventional networks outside their immediate circle — precisely the ties that criminal network embeddedness limits and that reintegration programs can build through networking events, employer connections, and community integration activities.
Research on the social capital dimensions of reintegration has documented that formerly incarcerated individuals typically have deep criminal network ties (strong ties to former criminal associates) but shallow conventional network ties (few weak ties to employers, mentors, and conventional community members) — a social capital profile that reflects both the criminal network embeddedness that incarceration interrupts only incompletely and the stigma-driven exclusion from conventional networks that criminal records produce. Programs that explicitly address this network capital imbalance — through job placement with employers who provide not only employment but professional social networks, through mentoring that connects returning citizens to mentors in conventional networks, and through civic reintegration activities that build weak ties to conventional community members — address the network dimension of reintegration barriers alongside the practical skill and housing dimensions that most reentry programs address.
Networks, Trust, and Collective Efficacy
The collective efficacy framework — reviewed in the urban criminology articles — can be understood through a network lens that specifies the relational properties that enable community members to exercise informal social control. Collective efficacy requires sufficient social cohesion among community members that they trust each other enough to intervene in observed problems — but this trust is not uniformly distributed across community members. It is embedded in the network of relationships through which community members know each other, have had positive prior interactions, and have developed the mutual expectations that make intervention feel like a community obligation rather than an individual risk.
Research by Browning and colleagues on neighborhood networks and collective efficacy found that the density of neighborhood social ties — measured through survey data on neighbor relationships — predicted collective efficacy and crime rates above and beyond the structural characteristics (poverty, instability, heterogeneity) that traditional social disorganization research emphasized. Dense neighborhood networks — where many community members know each other and have repeated interactions — provide the relational substrate within which the mutual trust and normative consensus that collective efficacy requires can develop. This network perspective on collective efficacy specifies the micro-level relational mechanisms through which the neighborhood-level concept operates, connecting the macro-level research on ecological conditions to the micro-level research on social network structure in a way that enriches both.
Network Analysis Methods: Advances and Limitations
The methodological development of network analysis in criminology has progressed substantially since the early co-arrest network studies, with advances in both data collection and analytical methods that have expanded what network criminology can establish empirically. Administrative data linking — connecting arrest records, court records, probation records, and other administrative databases to construct co-offending and co-association networks without relying on self-report or ethnographic methods — has produced the largest and most complete criminal network datasets available, enabling the kind of large-scale network epidemiology that Papachristos’s research exemplifies. Social media data mining has added the digital dimension of criminal networks to the offline administrative data that prior network research relied on, enabling the integrated offline-online network analysis that contemporary gang research requires.
The methodological limitations of network criminology deserve the same transparency that its innovations deserve. Network data from administrative records are limited by the enforcement patterns that generate them — co-arrest records are biased toward heavily policed communities and visible crimes, systematically underrepresenting white-collar crime, corporate crime, and the criminal networks of socially advantaged groups. The causal identification of network effects — demonstrating that network position independently affects behavior rather than that selection into networks confounds the relationship — remains methodologically challenging even with the quasi-experimental designs that the best network criminology employs. And the translation of network research findings into ethical intervention practice raises civil liberties concerns about the use of network position (rather than individual conduct) as a basis for differential treatment by law enforcement — concerns that the evidence-based criminology movement must engage with honestly rather than treating as secondary to evidence of effectiveness.
Conclusion
The social network perspective has transformed criminological understanding of violence, criminal careers, organized crime, and the relational mechanisms of desistance in ways that neither purely structural nor purely individual analytical approaches could have achieved. By making the relational micro-structure of criminal behavior visible — by showing that it is not only who you are or where you live but who you are connected to that determines your crime risk — network criminology has provided the most precise available account of how structural conditions and social learning processes produce criminal behavior at the level of specific relationships rather than abstract categories.
Social network analysis has become an indispensable component of the criminological toolkit — not because it replaces the structural, institutional, and behavioral frameworks that the discipline has built over a century of research but because it specifies the relational pathways through which those frameworks operate at the micro-level where criminal decisions are made. The finding that gun violence concentrates in specific network clusters; that co-offending network position predicts criminal career trajectories; that gang exit requires network transformation alongside individual motivation; and that digital networks extend offline criminal dynamics into new analytical territories — each represents a genuine theoretical contribution that network analysis makes to the understanding of crime that no other analytical approach can provide.
The policy implications of network criminology are being translated into practice through the focused deterrence and community violence intervention programs that use network analysis to identify the highest-risk individuals, target prevention resources precisely, and communicate deterrence messages and service offers to the specific network positions that violence analysis identifies. The evidence for these network-targeted interventions — reviewed in detail in the urban criminology articles — establishes network analysis as not merely a theoretical framework but a practical tool for violence reduction whose criminological development continues to inform its operational application. The trajectory from Papachristos’s foundational network epidemiology research to the American Rescue Plan’s community violence intervention investment — which explicitly incorporated network-targeting logic into its program requirements — illustrates both the translational potential of network criminology and the multi-year research-to-policy process through which academic insights become operational investments.
References
- Bouchard, M., & Nguyen, H. (2010). Is it who you know, or how many that counts? Criminal networks and cost avoidance in a sample of young offenders. Justice Quarterly, 27(1), 130–158. https://doi.org/10.1080/07418820902759694
- Braga, A. A., Papachristos, A. V., & Hureau, D. M. (2014). The effects of hot spots policing on crime: An updated systematic review and meta-analysis. Justice Quarterly, 31(4), 633–663. https://doi.org/10.1080/07418825.2012.673632
- Browning, C. R., Feinberg, S. L., & Dietz, R. D. (2004). The paradox of social organization: Networks, collective efficacy, and violent crime in urban neighborhoods. Social Forces, 83(2), 503–534. https://doi.org/10.1353/sof.2005.0006
- Decker, S. H., & Pyrooz, D. C. (2011). Leaving the gang: Logging off and moving on. Justice Quarterly, 28(2), 292–316. https://doi.org/10.1080/07418825.2010.509779
- Giordano, P. C., Cernkovich, S. A., & Rudolph, J. L. (2002). Gender, crime, and desistance: Toward a theory of cognitive transformation. American Journal of Sociology, 107(4), 990–1064. https://doi.org/10.1086/343191
- Granovetter, M. (1973). The strength of weak ties. American Journal of Sociology, 78(6), 1360–1380. https://doi.org/10.1086/225469
- Kennedy, D. M., & Braga, A. A. (2020). The Group Violence Intervention: An implementation guide. National Network for Safe Communities.
- Kleemans, E., & de Poot, C. J. (2008). Criminal careers in organized crime and social opportunity structure. European Journal of Criminology, 5(1), 69–98. https://doi.org/10.1177/1477370807084265
- McGloin, J. M., & Kirk, D. S. (2010). Social network analysis. In A. R. Piquero & D. Weisburd (Eds.), Handbook of quantitative criminology (pp. 209–224). Springer.
- McGloin, J. M., & Piquero, A. R. (2010). On the relationship between co-offending network redundancy and offending versatility. Journal of Research in Crime and Delinquency, 47(1), 63–90. https://doi.org/10.1177/0022427809350166
- Morselli, C. (2009). Inside criminal networks. Springer.
- Papachristos, A. V. (2020). The city as a social network. In P. Sharkey (Ed.), The Oxford handbook of urban sociology (pp. 345–368). Oxford University Press.
- Papachristos, A. V., Braga, A. A., & Hureau, D. M. (2012). Social networks and the risk of gunshot injury. Journal of Urban Health, 89(6), 992–1003. https://doi.org/10.1007/s11524-012-9703-9
- Papachristos, A. V., Wildeman, C., & Roberto, E. (2015). Tragic, but not random: The social contagion of nonfatal gunshot injuries. Social Science & Medicine, 125, 139–150. https://doi.org/10.1016/j.socscimed.2014.01.056
- Patton, D. U., Lane, J., Leonard, P., Macbeth, J., & Smith Lee, J. R. (2017). Gang violence on the digital street: Case study of a South Side Chicago gang member’s Twitter communication. New Media & Society, 19(7), 1000–1018. https://doi.org/10.1177/1461444815625949
- Pyrooz, D. C., & Decker, S. H. (2011). Motives and methods for leaving the gang: Understanding the process of gang desistance. Journal of Criminal Justice, 39(5), 417–425. https://doi.org/10.1016/j.jcrimjus.2011.07.001
- Sarnecki, J. (2001). Delinquent networks: Youth co-offending in Stockholm. Cambridge University Press.
- Slutkin, G., Ransford, C., & Zvetina, D. (2018). How the health sector can reduce violence by treating it as a contagion. AMA Journal of Ethics, 20(1), 47–55. https://doi.org/10.1001/journalofethics.2018.20.1.pfor2-1801
- Warr, M. (2002). Companions in crime: The social aspects of criminal conduct. Cambridge University Press.
- Weisburd, D., Telep, C. W., Hinkle, J. C., & Eck, J. E. (2010). Is problem-oriented policing effective in reducing crime and disorder? Criminology & Public Policy, 9(1), 139–172. https://doi.org/10.1111/j.1745-9133.2010.00617.x