SOCIAL NETWORK ANALYSIS 2. CRIME APPLICATION
Growing (late) application of SNA to crime (Carrington 2011, Calderoni 2014a): 1. Personal network influence on delinquency, 2. Neighborhood influence on delinquency and 3. Criminal networks a. Gangs b. Terrorism c. Organized crime or criminal organizations
Alien conspiracy (1950s and 1960s) OC as a heavily structured bureaucratic organization OC as a natio-wide conspiracy of Italian American families Limited evidence, but very successful Cressey, Theft of a nation (1968)
Empirical analysis (ethnography), limited success Reactions against Alien Conspiracy OC is not typical of Italian-Americans (Bell, Crime as an American Way of Life, 1953) OC is not an organization (Hess, Mafia and Mafiosi, 1970 in German, 1973 in English) OC as a network of patron-client relations (Albini, The American Mafia; genesis of a legend, 1971 OC as system of cultural and kinship relations(ianni and Reuss-Ianni, A family business, 1972)
OC as an enterprise OC provides illicit goods and services: security over efficiency (Reuter, Disorganized Crime, 1983) All OC activities are within a market (Gambetta, The Sicilian Mafia, 1993) Empirical analysis (case studies, interviews), good success
Caution Alien conspiracy, ethnic, cultural, kinship relations and enterprises are all paradigms OC has not (probably) changed (much) OC perception has
Since 1990s-2000s: OC as networks New paradigm shift OC has not (probably) changed (much) OC perception has Increasing claims that OC has become global, transnational, glocal Increasing claims that OC has adopted network structure Distinction between paradigm and reality
Application of SNA to OC Different methodological approach to the topic Different methods Different results Phenomenon is the same! SNA is suited because, relations, influence and flows are important in OC
Applications to OC are relatively few and recent (2000s) Limited SNA acceptance in the discipline Limited knowledge in SNA methods of OC scholars Limited data availability Limited data validity and reliability Growing field, interesting innovations
Data Judicial files Intercepted communications Co-offending Friendship, association, kinship Issues Inevitably incomplete pictures of network relations Need to control Triangulation with other methods
IMPLICATIONS OF SNA AND OC Research: Structure of criminal groups (Hyerarchy vs loose cooperation) Dynamics and evolution of criminal networks Roles and activities Law enforcement: Targeted intervention, more effective/efficient disruption strategies (myth or reality?) Evaluation of law enforcement activity Bridging different operations
OPERATION CAVIAR (MORSELLI & PETIT 2007) Analysis of the effects of law enforcement activity on drug trafficking organization 11 drug seizures (4 hashish, 7 cocaine), no arrests over 2 years 11 investigative phases Source: Morselli & Petit 2007
OPERATION CAVIAR (MORSELLI & PETIT 2007) Source: Morselli & Giguere 2006
OPERATION CAVIAR (MORSELLI & PETIT 2007) The structure 3 nodes with central positions Node Degree c. Betweennes c. N1 55% 64% N3 25% 11% N12 25% 29% Accused subjects have higher centrality than non accused subjects Source: Morselli & Petit 2007
OPERATION CAVIAR (MORSELLI & PETIT 2007) The network decentralizes As the law enforcement seizes the drug loads: Betweenness and degree centrality decrease The network decentralizes The leaders loose their role Source: Morselli & Petit 2007
OPERATION CAVIAR (MORSELLI & PETIT 2007) New leaders emerge The law enforcement activity provokes the emergence of new leaders (N3 and N12) Source: Morselli & Petit 2007
OPERATION CAVIAR (MORSELLI & PETIT 2007) The boss is increasingly criticized The law enforcement activity impacts upond the role of N1 (the boss ) N1 looses his central role Source: Morselli & Petit 2007
OPERATION CAVIAR (MORSELLI & PETIT 2007) Findings 1 Leaders have high centrality scores (both degree and betweenness) 2 With intense law enforcement activity, the network decentralizes 3 - With intense law enforcement activity, new leaders emerge 4 - With intense law enforcement activity, the leaders are increasingly criticized Highly centralized networks (few very central nodes) are vulnerable to targeted attacks Poorly centralized networks (a number of relatively central nodes) are more vulnerable to random attacks Consequence for law enforcement strategies
OPERATION CIEL (MORSELLI 2009) Contrast law enforcement traditional view of criminal groups hyerarchy Structure of the network from SNA perspective Hyerarchy The boss should be at the top of the structure: Minimize risk Giving order to lieutenants Low centrality Source: Morselli 2009
OPERATION CIEL (MORSELLI 2009) Partnership Players with brokerage positions Sharing sets of contacts High centrality Source: Morselli 2009
OPERATION CIEL (MORSELLI 2009) Analysis of centrality scores Two nodes with central positions N2 has higher scores than the formal boss (N1) Evidence supports the partnership model Confirm research on drug trafficking Source: Morselli 2009
OPERATION CERVO (MORSELLI & ROY 2008) Analysis of two car theft and smuggling organizations Social network analysis and crime script analysis (Cornish 1994). Operation SIREN: 44 nodes, 35 cars in 4 months Operation TOGO: 33 nodes, 20 cars in 3 months
OPERATION CERVO (MORSELLI & ROY 2008) Creating the scripts for Siren and Togo Source: Morselli & Roy 2008
OPERATION CERVO (MORSELLI & ROY 2008) Assessing the two groups The product of the available facets (number of possible alternatives) and the number of scenes (number of phases of the criminal activity) provides the theoretical number of available combinations for the perpetration of the crime. Operation Siren: 252 possible combinations Operation Togo: 72 possible combinations The Siren network is more flexible because it has more alternative disposal choices
OPERATION CERVO (MORSELLI & ROY 2008) The Siren network Analysis of the relations between the nodes according to their role within the criminal activity
OPERATION CERVO (MORSELLI & ROY 2008) The Togo network
OPERATION CERVO (MORSELLI & ROY 2008) Analysis of the effects of targeted vs. random law enforcement interventions Removal of the most important nodes (brokerage) is more effective than the randoremoval of 15 nodes Random intervention Siren: from 275 to 36/72 combinations Togo: from 72 to 8/16 combinations Source: Morselli & Roy 2008
OPERATION CERVO (MORSELLI & ROY 2008) Targeted intervention Siren: from 275 to 24 combinations Togo: from 72 to 0 combinations Source: Morselli & Roy 2008
OPERATION CERVO (MORSELLI & ROY 2008) Findings Added value of script analysis: verification of the impact of law enforcement on the functioning of the criminal network A network approach alone would give no information about the resistance of the remaining fragments of a network
2 La Ndrangheta nel traffico di cocaina OPERATION CHALONERO AND STUPOR MUNDI (CALDERONI 2012 & 2014B) Operazione Chalonero Cocaine trough Spain. Phone interceptions 2003-2004. Court order 2007 In Italy: Milano, Firenze, Bologna, Roma, Calabria Operazione Stupor Mundi Cocaine trough Spain and the Netherlands Phone interceptions 2002-2004. Court order 2007 In Italy: : Torino, Milano, Roma, Calabria
2 La Ndrangheta nel traffico di cocaina OPERATION CHALONERO AND STUPOR MUNDI (CALDERONI 2012 & 2014B) Network analysis Analysis of phone interceptions and meetings Information flows and role of different nodes Degree centrality Betweenness centrality Operation Chalonero n Density Degree centr. Betwee nness centr. Cluster ing coeff.* Operation Stupor Mundi n Density Degree centr. Betwee nness centr. Cluster ing coeff.* entire group 107 2.9% 38.4% 53.6% 14,5% 134 3.0% 41.9% 49.0% 19,2% at least 2 contacts 61 7.2% 56.3% 73.5% 21,0% 73 8.3% 64.6% 49.4% 24,7% core group 39 11.5% 68.4% 73.1% 24,3% 48 14.9% 71.1% 43.5% 32,2%
2 La Ndrangheta nel traffico di cocaina OPERATION CHALONERO AND STUPOR MUNDI (CALDERONI 2012 & 2014B) Status analysis (Natarajan 2000) Analysis of a sample of conversations 6 codes of status: give orders, ask information, give information, dare del lei / voi, express (dis)satisfaction, ask clarifications Status score measures the social status of a subject
2 La Ndrangheta nel traffico di cocaina OPERATION CHALONERO AND STUPOR MUNDI (CALDERONI 2012 & 2014B) Status analysis (Natarajan 2000) Operation Chalonero Operation Stupor Mundi Nodes 39 Nodes 48 Min 2 Min 1,7 Max 4 Max 4,5 Mean 2,8 Mean 2,6 St. dev. 0,62 St. dev. 0,7
2 La Ndrangheta nel traffico di cocaina OPERATION CHALONERO AND STUPOR MUNDI (CALDERONI 2012 & 2014B) Task analysis Analysis of a sample of conversations Identification of the main task of each subject Operation Chalonero Operation Stupor Mundi Task no Task no Supplier 14 Trafficker 9 Trafficker 16 Support activity 15 Support activity 26 Courier 12 Buyer 15 Buyer 19 Dealer 7
OPERAZIONE STUPOR MUNDI NODE SIZE: DEGREE C.
2 La Ndrangheta nel traffico di cocaina OPERAZIONE STUPOR MUNDI NODE SIZE: BETWEENNESS C.
OPERAZIONE STUPOR MUNDI NODE SIZE: STATUS SCORE
2 La Ndrangheta nel traffico di cocaina OPERAZIONE CHALONERO NODE SIZE: DEGREE C.
2 La Ndrangheta nel traffico di cocaina OPERAZIONE CHALONERO NODE SIZE: BETWEENNESS C.
2 La Ndrangheta nel traffico di cocaina OPERAZIONE CHALONERO NODE SIZE: STATUS SCORE
OPERATION CHALONERO AND STUPOR MUNDI (CALDERONI 2012 & 2014B) Centrality and status are different Internal hyerarchy does not correspond to operational roles Strategic positioning of the Ndrangheta Bosses are not directly involved in the traffic
OPERAZIONE STUPOR MUNDI STATUS, TASK AND CENTRALITY Task Trafficker Courier Support Buyer 3 2 Status score 4 4,5 N10 N11 N12 N120 N13 N14 N15 N16 N17 N18 N19 N2 N21 N22 N23 N24 N25 N26 N27 N28 N29 N3 N30 N31 N32 N33 N34 N35 N36 N37 N38 N39 N41 N42 N43 N44 N45 N46 N47 N48 N49 N5 N50 N6 N61 N68 N80 N9
OPERAZIONE CHALONERO STATUS, TASK AND CENTRALITY 2 La Ndrangheta nel traffico di cocaina N77 N1 N20 4 N17 N47 N4 N53 N2 Status score 3 N48 N3 N46 N41 N18 N39 N6 N19 N25 N54 N43 N38 N15 N14 N23 N45 N24 N21 N50 2 N65 N87 N66 N56 N8 N51 N31 N57 N58 N64 N90 N61 Producer Trafficker Support Buyer Retailer Task
2 La Ndrangheta nel traffico di cocaina OPERATION INFINITO (CALDERONI 2014C AND FORTHCOMING) Operation Infinito: large scale operation against the Ndrangheta in July 2010 Main charge: 416-bis Identification of locali in Lombardia Careful identification of the internal hyerarchy
2 La Ndrangheta nel traffico di cocaina OPERATION INFINITO (CALDERONI 2014C AND FORTHCOMING) Objective: 1. verify whether meeting attendance predicts leadership positions 2. Verify the robustness of the prediction throughout the investigation Relevance: Research: characteristics and behavior of leaders Law enforcement: early identification of leaders
2 La Ndrangheta nel traffico di cocaina OPERATION INFINITO (CALDERONI 2014C AND FORTHCOMING) Data: manual extraction of meeting data 2-mode matrix 1 mode matrix (coparticipation in meetings) Hyerarchy derived from the court order Table 1. Meetings per number of participants and individuals per meetings attended Participants per No of Meetings Meeting Meetings Attended No of Individuals 4 47 1 99 5 19 2 29 6 22 3-10 65 7+ 41 11+ 22 Total participants 129 Total meetings 215
blue: individuals; red: meetings 2 La Ndrangheta nel traffico di cocaina OPERATION INFINITO (CALDERONI 2014C AND FORTHCOMING)
2 La Ndrangheta nel traffico di cocaina OPERATION INFINITO (CALDERONI 2014C AND FORTHCOMING) Real roles Predictive Model black: leaders grey: others
2 La Ndrangheta nel traffico di cocaina OPERATION INFINITO (CALDERONI 2014C AND FORTHCOMING) Table 1. Results of the logistic regression for the whole investigation (dependent variable: leadership) Predictor B SE OR Valued degree Betweenness (normalised) Constant *p<.05, ** p<.01, *** p<.001.036.009 1.037***.942.280 2.565** -4.261.544.014*** Hosmer-Lemeshow test (χ 2 15.344, p<.53, df =8). Nagelkerke s R 2 is of.641 Successful prediction: 92.6% of individuals (66.7% for bosses and 97.3% for others)
2 La Ndrangheta nel traffico di cocaina OPERATION INFINITO (CALDERONI 2014C AND FORTHCOMING) Fig. 1. Number of meetings per month and period 16 14 12 10 8 6 4 2 0 I 2007-6 2007-7 2007-8 2007-9 2007-10 2007-11 2007-12 2008-1 2008-2 2008-3 2008-4 2008-5 2008-6 2008-7 2008-8 2008-9 2008-10 2008-11 2008-12 2009-1 2009-2 2009-3 2009-4 2009-5 2009-6 2009-7 2009-8 2009-9 2009-10 2009-11 2009-12 2010-1 2010-2 II III IV Table 1. Descriptive statistics of the four time periods I II III IV End (from 25 Jun 2007) 30 Jun 2008 End 2008 30 Jun 2009 28 Feb 2010 No. months 13 19 25 33 No. meetings 34 53 93 118 No. nodes 82 126 183 205 Of Which Bosses 27 31 33 33
2 La Ndrangheta nel traffico di cocaina OPERATION INFINITO (CALDERONI 2014C AND FORTHCOMING) Table 1. Share of successful predictions in the logistic regressions for the four time periods (dependent variable: leadership) Mid08 End08 Mid09 Beg10 Constant only 67.1 75.4 82 83.9 Full model 75.6 84.9 90.7 92.2 Others 90.9 94.7 98 97.7 Bosses 44.4 54.8 57.6 63.6
2 La Ndrangheta nel traffico di cocaina OPERATION INFINITO (CALDERONI 2014C AND FORTHCOMING) Table 1. Results of the logistic regressions for the four time periods (dependent variable: leadership) Time period Variables B SE OR Mid08 Valued Degree.001.025 1.001 Betweenness (normalised).614.359 1.848! No. Meetings.116.238 1.123 Mafia Charge 1.735 1.092 5.671 Constant -3.105 1.058.045** H-L Test χ2= 3.498 p=.899 Nagelkerke s R2=.387 End08 Valued Degree.008.024 1.008 Betweenness (normalised).833.359 2.300* No. Meetings.077.217 1.080 Mafia Charge 1.820 1.084 6.170! Constant -3.866 1.026.021*** H-L Test χ2= 7.747 p=.459 Nagelkerke s R2=.515 Mid09 Valued Degree.024.018 1.024 Betweenness (normalised).691.264 1.996** No. Meetings -.068.168.934 Mafia Charge 2.173 1.080 8.787* Constant -4.638 1.016.010*** H-L Test χ2= 3.861 p=.896 Nagelkerke s R2=.559 Beg10 Valued Degree.044.019 1.044* Betweenness (normalised).898.309 2.454** No. Meetings -.132.155.876 Mafia Charge 1.731 1.108 5.648 Constant -5.180 1.033.006*** H-L Test χ2= 6.066 p=.640 Nagelkerke s R2=.657!p<.1,*p<.05, ** p<.01, *** p<.001;
2 La Ndrangheta nel traffico di cocaina SNA AND CRIME: FUTURE DIRECTIONS Replication and extension of the data (e.g. Operation Il Crimine & Minotauro, cooperation with Italian law enforcement) Cross-investigation networks Evolution in time Exploration of new visualization and analysis methods (Spectral embedding & Simmelian backbones, Skillicorn, Zheng, and Calderoni forthcoming; Cooperation with the Dept. of Mathematics of the Politecnico di Milano) New fields: Online counterfeiting (use of bots and crawlers to map websites and forums selling counterfeits) Corruption in public procurement
Baker, Wayne E., and Robert R. Faulkner. 1993. The Social Organization of Conspiracy: Illegal Networks in the Heavy Electrical Equipment Industry. American Sociological Review 58 (6): 837 60. Calderoni, Francesco. 2014. Social Network Analysis of Organized Criminal Groups. Edited by Gerben Bruinsma and David Weisburd. Encyclopedia of Criminology and Criminal Justice. New York: Springer. Calderoni, Francesco. 2014b. Strategic Positioning in Mafia Networks. In Crime and Networks, edited by Carlo Morselli, 163 81. New York: Routledge. Calderoni, Francesco. 2014c. Identifying Mafia Bosses from Meeting Attendance. In Networks and Network Analysis for Defence and Security, edited by Anthony Masys. Cham: Springer International Publishing Switzerland. Calderoni, Francesco. 2012. The Structure of Drug Trafficking Mafias: The Ndrangheta and Cocaine. Crime, Law and Social Change 58 (3): 321 49. doi:10.1007/s10611-012-9387-9. Carrington, Peter J. 2011. Crime and Social Network Analysis. In The SAGE Handbook of Social Network Analysis, edited by John P Scott and Peter J. Carrington, 236 55. London: SAGE Publications. Morselli, Carlo. 2009. Inside Criminal Networks. Studies of Organized Crime. New York, NY: Springer. Morselli, Carlo, and Katia Petit. 2007. Law-Enforcement Disruption of a Drug Importation Network. Global Crime 8 (2): 109 30. Morselli, Carlo, and Julie Roy. 2008. Brokerage Qualifications in Ringing Operations. Criminology 46 (1): 71 98. 54