Tactical (S)NA: Using Affiliation Networks to Aid Serial Crime Investigations Deputy Lim Gisela Bichler Edgar Larin Steven Lim (with the assistance of Lisa Chavez & Pedro Martinez) ______________ 2nd Lt. Larin Department of Criminal Justice California State University, San Bernardino 5500 University Parkway, SBDO, 92407 francesco calderoni What utility does the network approach have for an ongoing investigation? completed cases Italian mafia wire taps; court documents (natarajan, bright, leuprecht) investigatory documents (morselli, bouchard) aggregates combined intelligence (malm & bichler) LE & other sources (kleemans, van de bunt, de poot, sarnecki) Calderoni, Francesco. 2012. Roundtable at ASC, D.C. kim rossmo investigatory failure cognitive biases e.g., tunnel vision, satisficing errors in probability e.g., coincidences, conjunction fallacy organizational traps e.g., groupthink, organizational momentum Rossmo, D. Kim (2009).Criminal Investigative Failures. CRC Press: New York Rossmo, D. Kim (2014). Criminal Case Rethinking. Presentation Western Society of Criminology, HI Organizational momentum, the inability to change direction in the midst of a major investigation… shifting focus away from working hypothesis or “good” suspect suggestion: projected networks Rossmo, D. Kim (2009).Criminal Investigative Failures. CRC Press: New York Rossmo, D. Kim (2014). Criminal Case Rethinking. Presentation Western Society of Criminology, HI simon demers When is there enough information to warrant a criminal inquiry? find the investigatory tipping point suggestion: network evolution models LePard et al. (2014). Challenges in Serial Murder Investigations involving Missing Persons. Presentation Western Society of Criminology, HI purpose test if (S)NA has tactical utility in an investigation • identify information or change thresholds • find central actors/places start with serial case with known offender green river killer 18 lengthy case (1982-2001) shift in investigative focus w/ temporal variation 16 number of victims 14 12 10 8 6 a conviction 4 2 0 before 6 months 12 months 18 months 24 months post 30 bodies months discovered timeline vice crackdown on SEATAC stroll “john” arrest & recent divorce decoys resume search home got DNA questioned key source published moved & married questioned &“passed” polygraph Gary Ridgway dating bodies found arrested 1982 1983 25 members 1984 1985 1986 43 member task force 1987 1988 1989 1990 1991 1992 1993 …. 2001 sources published 1991 (before offender known) independent public account (re-interviewed witnesses & visited places) followed case from beginning Kim said it was good did not use… inclusion criteria could construct a dyad April 1983 j person associated with a case important place/site unknown time stamp for place association X 70% dyads usable (151 / 217 ) geographic aggregation obtained coordinates aggregated to census blocks (2000) Washington temporal aggregation All info JAN-85 – DEC-03 30 Months JUL-84 – DEC-84 24 Months JAN-84 – JUN-84 18 Months JUL-83 – DEC-83 12 Months JAN-83 – JUN-83 T1 multiple bodies found in the Green River JUL-82 DEC-82 6 Months temporal aggregation All info JAN-85 – DEC-03 30 Months JUL-84 – DEC-84 24 Months JAN-84 – JUN-84 18 Months JUL-83 – DEC-83 T2 multiple bodies found in the Green River 12 Months JAN-83 – JUN-83 JUL-82 DEC-82 6 Months temporal aggregation All info JAN-85 – DEC-03 30 Months JUL-84 – DEC-84 24 Months JAN-84 – JUN-84 T3 18 Months JUL-83 – DEC-83 12 Months JAN-83 – JUN-83 multiple bodies found in the Green River JUL-82 DEC-82 6 Months temporal aggregation All info JAN-85 – DEC-03 30 Months JUL-84 – DEC-84 T4 24 Months JAN-84 – JUN-84 18 Months JUL-83 – DEC-83 12 Months JAN-83 – JUN-83 multiple bodies found in the Green River JUL-82 DEC-82 6 Months temporal aggregation All info JAN-85 – DEC-03 30 Months T5 JUL-84 – DEC-84 24 Months JAN-84 – JUN-84 18 Months JUL-83 – DEC-83 12 Months JAN-83 – JUN-83 multiple bodies found in the Green River JUL-82 DEC-82 6 Months temporal aggregation All info T6 JAN-85 – DEC-03 30 Months T5 JUL-84 – DEC-84 24 Months T4 JAN-84 – JUN-84 18 Months T3 JUL-83 – DEC-83 T2 12 Months JAN-83 – JUN-83 T1 multiple bodies found in the Green River JUL-82 DEC-82 6 Months study units Study Units Descriptive Categories People Victims Investigatory involvement (e.g., witnesses, body finders) Suspects Family and Associates Places Body dumpsites Last seen locations Material to investigation or combination Percent (N) (88) 56.8 29.6 8.0 5.7 (58) 37.9 44.8 17.2 network generation affiliation networks (2 different units) prostitute #2 prostitute #1 Shared activity space/node body dumpsite prostitution stroll suspect projecting a network (deriving a 1-mode network) actor-to-actor place-to-place prostitute #2 prostitute #1 body dumpsite prostitution stroll suspect projecting a network (deriving a 1-mode network) actor-to-actor place-to-place prostitute #2 prostitute #1 body dumpsite prostitution stroll suspect results network stability change in positional importance stability jaccard coefficient time 2 time 1 D B C A D B C A repeated for the entire network proportion the same < 20% new network forms …radical shift in focus 20-60% natural evolution … filling in gaps > 60% too much stability … investigatory stagnation network stability actor-to-actor T3 actors T2 places place-to-place (18 months) (12 months) note: jaccard coefficients converted to % betweenness centrality standardized probability of sitting between others or joining different subsets of the network – go between along most efficient path C B (ni ) C (ni ) ( g 1)( g 2) / 2 ' B more likely to be associated with pairs (of victims) linked people Other Victim Suspect Period 1 6 months 24 people 102 unique ties 19% density T2 12 months T3 18 months T4 24 months T5 30 months T6 entire investigation linked people Other Victim Suspect T1 6 months 39 people 220 unique ties 15% density Period 2 12 months T3 18 months T4 24 months T5 30 months T6 entire investigation linked people Other Victim Suspect T1 6 months 59 people 424 unique ties 13% density T2 12 months Period 3 18 months T4 24 months T5 30 months T6 entire investigation linked people Other Victim Suspect T1 6 months 66 people 556 unique ties 14% density T2 12 months T3 18 months Period 4 24 months T5 30 months T6 entire investigation linked people Other Victim Suspect T1 6 months 76 people 686 unique ties 12% density T2 12 months T3 18 months T4 24 months Period 5 30 months T6 entire investigation linked people Other Victim Suspect T1 6 months 88 people 1,304 unique ties 18% density T2 12 months T3 18 months T4 24 months T5 30 months Period 6 entire investigation freeman’s betweenness Suspects T1 T2 T3 T4 (standardized) T5 T6 07/82–12/82 07/82–06/83 (12 months) 07/82–12/83 07/82–06/84 07/82–12/84 07/82–12/03 (18 months) (24 months) (30 months) (entire period) (6 months) Gary Ridgway . 0.17 2.54 1.53 1.21 1.16 Bobby Woods . . 1.54 0.17 . 0.64 0.20 . 1.08 0.26 1.19 1.45 0.27 0.87 1.35 0.30 0.79 1.31 24 39 59 66 76 88 James M Tindal Melvyn Foster People in network prime suspect Sept. 1982 to 1985 If we ignore Melvyn Foster as a suspect…. Suspects T1 T2 T3 T4 T5 T6 07/82–12/82 07/82–06/83 (12 months) 07/82–12/83 07/82–06/84 07/82–12/84 07/82–12/03 (18 months) (24 months) (30 months) (entire period) (6 months) Gary Ridgway . 0.048 0.725 0.393 0.238 0.179 Bobby Woods . . . 0.048 . . 0.053 . . 0.063 0.316 . 0.063 0.153 . 0.038 0.123 0.267 24 39 59 66 76 88 James M Tindal Melvyn Foster People in network prime suspect Sept. 1982 to 1985 place connectivity 21 nodes 32 ties 8% density Last Seen Body Dumpsite Symbol size shows betweenness centrality. (derived place‐to‐place network) Period 1 6 months T2 12 months T3 18 months T4 24 months T5 30 months T6 entire investigation place connectivity 32 nodes 64 ties 7% density Last Seen Body Dumpsite Symbol size shows betweenness centrality. (derived place‐to‐place network) T1 6 months Period 2 12 months T3 18 months T4 24 months T5 30 months T6 entire investigation crackdown on prostitution stroll by SEATAC place connectivity 44 nodes 104 ties 6% density Last Seen Body Dumpsite T1 6 months T2 12 months Period 3 18 months T4 24 months T5 30 months T6 entire investigation place connectivity 50 nodes 154 ties 7% density Last Seen Body Dumpsite Symbol size shows betweenness centrality. (derived place‐to‐place network) T1 6 months T2 12 months T3 18 months Period 4 24 months T5 30 months T6 entire investigation place connectivity 55 nodes 196 ties 7% density Last Seen Body Dumpsite Symbol size shows betweenness centrality. (derived place‐to‐place network) T1 6 months T2 12 months T3 18 months T4 24 months Period 5 30 months T6 entire investigation place connectivity 58 nodes 284 ties 10% density Last Seen Body Dumpsite Symbol size shows betweenness centrality. (derived place‐to‐place network) T1 6 months T2 12 months T3 18 months T4 24 months T5 30 months Period 6 entire investigation research findings brantingham2 influence of social network on choice of activity nodes (behavioral settings) and thus, spatial behavior utility of graph theory to understand connectedness (aggregated activity space) Brantingham, P. and P. Brantingham (1984). Patterns in Crime. Macmillan. Ch. 12; 9. Brantingham, P. and P. Brantingham (2008). Crime pattern theory. In Environmental Criminology and Crime Analysis, R. Wortley and L. Mazerolle (Eds.) Willan. Ch. 5. practical findings actor network • • • • • information threshold 18 months radical shift in suspect centrality some people never central (outliers) structurally important victims additional relationships and actors obscures key suspect location network • stable at 12 months • variation in centrality of hunting locations • shift in centrality of body dumpsites • some places not connected • central places limitations issues next steps geographic aggregation lost spatial precision street segments? missing time reduced no. actors person-to-person (w/o time) descriptive statistics repeated measures ANOVA case study w/ public info. collaborate & replicate take home betweenness / projected networks might be useful rethinking cases test for organizational momentum select sources Calderoni, Francisco (forthcoming) Predicting Leadership from Meeting Attendance in Criminal Networks. In Gisela Bichler & Aili Malm eds. Using Social Network Analysis to Prevent Crime, Crime Prevention Studies Vol. 28, Lynne Rienner Publish: Boulder, Colorado Reichert, David (2004) Chasing the Devil: My Twenty-year Quest to Capture the Green River. Little, Brown and Company: New York, NY Rossmo, D. Kim (2000). Geographic Profiling. CRC Press: London New York Rossmo, D. Kim (2009).Criminal Investigative Failures. CRC Press: London New York Smith, Carlton and Thomas Guillen (1991). The Search for the Green River Killer: A True Crime Shocker. Penguin Putnam: New York, NY State of Washington vs. Gary Leon Ridgway. Prosecutor’s Summary of the Evidence. No. 01-1-10270-9 SEA Superior Court of Washington for King County. plugs Disrupting Criminal Networks Crime Prevention Studies Vol. 28 Gisela Bichler & Aili Malm March 2015 Contributors Topic Jean M. McGloin & Zachary Rowan Gangs and Delinquency Michael Sierra-Arevalo & Andrew V. Papachristos Carlo Morselli, Thomas U. Grund, & Rémi Boivin David A. Bright Francesco Calderoni Robert R. Faulkner & Eric R. Cheney David Décary-Hétu & Dominique Laferrière Kila Joffres & Martin Bouchard Stacy Bush & Gisela Bichler Lucia Summers & Shane D. Johnson Group Audits and POP Co-offending Drug Distribution Organized Crime Leaders Enterprise Conspiracy Online Criminal Markets Online Child Exploitation Terrorist Communications Space Syntax Journal of Contemporary Criminal Justice Aili Malm, Gisela Bichler & Carlo Morselli March 2015 Contributors Topic Jacob Young & Justin Ready On-officer Cameras (Diffusion) Evan McCuish, Martin Bouchard & Raymond Corrado Peter Carrington David A. Bright et al. Thomas Grund & James Densley Gisela Bichler, Andrea Schoepfer & Stacy Bush Homicide Co-offending Age & Co-offending Drug Trafficking Gangs Corporate Offenders Criminology, Criminal Justice, Law and Society Christian Leuprecht & Andrew Aulthouse [International Gun Trafficking] Fall 2014, Volume 15, Issue 3 Sean Everton & Dan Cunningham [Terror Group Structure] Forthcoming March 2015, Volume 16, Issue 1 thank you