Police Decision-Making, Criminal Profiling, and Geographic Analysis

Course Administration & Announcements

• Test Marking & Exam Structure:

  • Test marking is nearly complete, with grades expected to be released by the end of the week.
  • Overall performance on multiple-choice questions was very strong.
  • The test question difficulty distribution was roughly:
    • $40\%$ Easy
    • $40\%$ Moderate
    • $20\%$ Hard
  • The final exam balance will shift slightly, featuring fewer easy questions and slightly more moderate questions.
  • The long-answer replacement will be detailed by the end of the week. The final format will resemble a case presentation requiring the integration and application of learnings from across the entire course.

• Upcoming Guest Lecture & Requirements:

  • A guest lecturer will present on Māori in the criminal justice system in two weeks.
  • Students must complete two mandatory readings prior to the lecture.
  • Questions covering these mandatory readings are included in the weekly homework quiz.
  • Homework Quiz Password: Capone (named after American gangster Al Capone, who influenced the implementation of expiration dates on milk bottles because he detested spoiled milk).

Introduction to Police Decision-Making & Criminal Profiling

• Conceptual Overview:

  • Formerly referred to as investigative decision-making, the topic is more accurately designated as police decision-making to cover applications beyond initial investigations.
  • Proverb: "The corners of a house can be seen, the corners of the heart cannot."
    • Application: While traditionally highlighting the need for empathy and the hidden nature of internal emotions and intentions, in profiling it represents the attempt to observe external crime scene characteristics to infer internal, unobservable offender traits.

• Definition of Criminal Profiling:

  • Criminal profiling is an investigative method that uses information from the crime scene and the offense itself to infer behavioral, personality, and demographic characteristics of an unknown offender.

• Information Desired by Law Enforcement from Profiles:

  • Offender Characteristics: Demographics, physical descriptions, likely age, appearance, and background.
  • Behavioral & Threat Assessment: Understanding psychological motives, behavioral patterns, and assessing the risk or likelihood of future reoffending.
  • Identifying Prior Criminal History: Determining whether behavioral features suggest a suspect with an existing official record to narrow the search pool.
    • Case Example: Green River Killer (Gary Ridgway). Investigators identified a pattern showing the killer returned to crime scenes to interact with disposed bodies. Setting up surveillance at known body disposal sites contributed to his eventual identification.
  • Offense Linkage: Connecting a series of separate offenses (e.g., a series of burglaries) to a single perpetrator to consolidate geographical and temporal parameters.
  • Interview & Interrogation Strategy: Advising investigators on tailored communication styles best suited to elicit a confession from a specific psychological profile.
  • Operational Resource Allocation: Prioritizing suspect lists and determining the urgency and deployment of police resources based on public threat.
  • Evidentiary Status: Profiles cannot be used as evidence in court and do not explicitly pinpoint a specific individual; their sole operational purpose is directing police investigations and narrowing suspect lists.

Historical Case Study: The Mad Bomber (1940–1956)

• Background & Offense Pattern:

  • Over a 16-year period (1940–1956) in New York City, 33 explosive devices were planted across the city; 22 detonated, injuring 15 people (0 fatalities).
  • Initial targets were facilities belonging to Con Edison (a major utility power company). Targets later expanded to public locations including movie theaters, phone booths, and transportation hubs.
  • The perpetrator sent neat, handwritten letters signed FP ("Fair Play") to newspapers and Con Edison, articulating explicit grievances.
  • Text Characteristics: Handwriting was fastidiously neat, lacking cross-outs, corrections, or spelling errors. Excerpted text included quotes such as: "I will make no more bomb units for the duration of the war. My patriotic feelings have made me decide this. Later, I will bring the Con Edison to justice. They will pay for their dastardly deeds."
  • Escalation: In December 1956, a bomb detonated in a Brooklyn movie theater, injuring 6 people and creating intense public pressure to solve the case.

• James Brussel's Psychiatric Profile:

  • New York police engaged Dr. James Brussel, a psychiatrist with counterintelligence experience in WWII and the Korean War, who utilized "reverse psychology" and the "laws of probability."
  • Brussel's Retrospective Profile (as published in his memoir):
    • Male, foreign-born, Eastern European/Slavic (specifically Lithuanian/Yugoslavian descent).
    • Age 40–50.
    • Single, living with female relatives.
    • Clean-shaven, neatly dressed, athletic/stocky build.
    • Suffering from severe paranoia/schizophrenia.
    • Wearing a double-breasted suit with all buttons done up.

• Rationale Behind Brussel's Inferences:

  • Neat Handwriting: Inferred a fastidiously neat, clean-shaven individual wearing buttoned double-breasted attire.
  • Age & Build: Statistical probabilities of mental illness at the time placed schizophrenia onset around age 30. Adding the 16-year duration yielded an age range in the 40s or 50s. Psychiatric theories of the era tied stocky builds to schizophrenia.
  • Slavic Descent & Foreign Origin: Cold War tensions, frequent Yugo-Slav bombings during that period, and unusual turns of phrase ("dastardly deeds") suggested non-native language origins.
  • Living Situation: Freudian psychoanalytic theory interpreted planting bombs inside movie theater seats as sexual repression and unresolved Oedipal dynamics (living with female relatives).

• Apprehension & Discrepancies:

  • Apprehension: George Metesky was identified not via the profile, but by a clerk searching Con Edison employee records for workers terminated following an industrial injury.
  • Match to Profile: Metesky was Lithuanian, lived with his two unmarried sisters, had never married, had a history of paranoia and neighbor disputes, was 54 years old, and possessed a fastidiously clean home.
  • Arrest Detail: Arrested in his pajamas, Metesky requested permission to change before going to the station; he emerged wearing a fully buttoned double-breasted suit.
  • Original vs. Retrospective Profile Discrepancies: Brussel's published newspaper profile differed from his memoir. The original newspaper release claimed the bomber was born in Germany (not Eastern Europe), was an active skilled mechanic working night shifts, and resided in White Plains (Metesky lived in Connecticut and had been unemployed since the 1930s).

The FBI Top-Down Profiling Approach

• Origins:

  • Developed in the 1970s by the FBI's Behavioral Sciences Unit (BSU) at Quantico.
  • Based on semi-structured interviews with 36 incarcerated sexual murderers (e.g., Ed Kemper, David Berkowitz, John Wayne Gacy, Ted Bundy).
  • Coined the term serial killer.

• Methodological Framework (Top-Down):

  • Establishes overarching categories derived from known, incarcerated offenders and applies these theoretical frameworks top-down to analyze unsolved crime scenes.

• Methodological Limitations of the FBI Sample:

  • Sampling Bias: Small ($N = 36$), non-representative sample restricted strictly to caught, incarcerated, sexual homicide offenders.
  • Uncaptured Population: Excludes unsolved homicides or successful, uncaught serial offenders.
  • Deception & Ego: Incarcerated offenders often possess massive egos, leading to exaggeration, minimization, or outright manipulation of interviewers.
  • Lack of Self-Insight: Assumes offenders possess accurate psychological insight into their own complex motivations.
  • Static Data Base: The FBI theoretical framework has rarely been empirically updated with modern datasets since its initial inception.

• Core FBI Concepts: Modus Operandi vs. Signature:

  • Modus Operandi (MO):
    • Functional and practical behaviors necessary to execute the crime and avoid detection.
    • Dynamic and learned; changes over time as the offender gains experience or adapts to obstacles.
  • Criminal Signature:
    • Behaviors not required to commit the crime, executed to satisfy emotional or psychological needs.
    • Static and stable across time, serving as the primary basis for building a profile.

• Applied Example: Jack the Ripper (1888, London):

  • MO: Targeting female sex workers in London's West End, operating past midnight, utilizing a sharp knife. Shifted from outdoor attacks to indoor locations after being interrupted during his fourth attack (demonstrating MO adaptation/learning).
  • Signature: Excessive post-mortem violence (overkill), face mutilation, deliberate posing and degradation of bodies for discovery, and signs of picquerism (sexual gratification derived from stabbing or cutting).

• Organized vs. Disorganized Offender Typology:

• Case Example: Ted Bundy:

  • Organized Characteristics: Targeted young women across broad geographic ranges; used deceptive ruses (fake arm slings, crutches, posing as police); drove modified vehicles; moved bodies to remote sites; maintained long-term relationships; worked on political campaigns and crisis phone lines; closely followed media coverage (even offering unsolicited profiling advice to aid in the Green River Killer investigation).
  • Disorganized Exception: Bundy's 1978 attack on the Chi Omega sorority house was chaotic, sudden, and opportunistic, attacking multiple women in their sleep after breaking in during the early morning hours. This highlights that organized offenders can demonstrate disorganized behaviors depending on circumstances.

• Empirical Critiques of FBI Profiling:

  • Canter et al. (100 US Serial Homicides): Analyzed the 3rd offense in a series to account for early trial-and-error learning. Found that crime scene behaviors do not cleanly separate into organized and disorganized clusters.
  • Ubiquity of Organized Features: Organized characteristics were found to be typical features of almost all serial homicides; because they are universal baseline traits, they lack diagnostic power to differentiate offenders.
  • Lack of Clustering in Disorganized Traits: Disorganized traits failed to co-occur consistently in specific individuals.
  • Homology Assumption Failure: Studies on stranger rapes ($N = 100$) revealed no statistical correlation between behavioral similarity at the crime scene and background demographic similarity among offenders. Offender behavior is overwhelmingly situation-dependent.

• High-Profile Profiling Failures:

  • Richard Jewell (1996): Security guard who discovered a bomb at the Atlanta Olympics and saved lives. FBI profilers asserted he fit the "lone bomber seeking hero status" profile. The resulting investigation upended his life until physical evidence (fingerprints/DNA) identified the actual perpetrator (Eric Rudolph).
  • Washington Snipers (2002): FBI profiles predicted a single white male with military training and gun expertise. The actual shooters were John Allen Muhammad and Lee Boyd Malvo (two Black males, an adult and a teenager, operating from a modified vehicle trunk with no military background). Police previously stopped and released the pair at a cordon because they did not match the white-male profile.
  • Colin Stagg / Rachel Nickell Murder (1992): Rachel Nickell was murdered on Wimbledon Common. Police targeted Colin Stagg based on a matching profile, executing a 6-month undercover operational "honeytrap" operation to extract a confession. Stagg denied involvement throughout. The judge threw the case out of court as blatant police coercion. DNA evidence later identified the actual killer, Robert Napper, who had committed additional murders while police focused on Stagg.

Statistical / Bottom-Up Profiling & Linkage Analysis

• Origins & Definition:

  • Developed by David Canter in the 1980s in the United Kingdom.
  • Bottom-Up Approach: Starts with empirical data from large samples of solved offenses to identify statistical co-occurrences of crime scene behaviors and predict probability distributions of offender characteristics.
  • Applicable across broad offense types: arson, burglary, robbery, fraud, sexual assault, and homicide.
  • Eliminates subjective intuition and unvalidated expert guesswork.

• Foundation Case: The Railway Murders:

  • Perpetrated by John Duffy and David Mulcahy in the UK.
  • Canter generated a statistical profile for Duffy that reduced his rank on a 2,000-suspect list down to the top 5 key suspects.

• Multidimensional Scaling / Cluster Analysis (Sexual Assault Example):

  • Crime scene behaviors are coded (e.g., binding limbs, kissing, facial injury, explicit humiliation, surprise attacks, clothing removal).
  • Statistical clustering identifies behavioral sub-types mapping to four quadrants:
    1. Aggressive
    2. Sadist (Statistically less likely to have prior criminal records compared to the criminality quadrant)
    3. Intimacy
    4. Criminality
  • Spatial Centrality: Behaviors located near the center of a multidimensional scaling plot (e.g., vaginal penetration, surprise attack, clothing removal) represent high-frequency behaviors common to almost all sexual offenses. Behaviors on the outer edges are low-frequency, specific acts that define unique offender sub-types.

• Linkage Analysis:

  • The practice of linking multiple offenses to a single perpetrator based on behavioral consistency and distinctiveness (uniqueness of actions beyond standard crime features).
  • Linkage Blindness: The failure of police agencies to link related crimes due to isolated, non-interoperable jurisdictional databases across state or international borders.
  • New Zealand Context: Operates a centralized national jurisdiction using the ViCLAS (Violent Crime Linkage Analysis System) database, primarily maintained for sexual offenses.
  • Evidentiary Status: Behavioral linkage can be admitted in court as evidence if the judge determines its probative value outweighs its prejudicial impact.
  • Case Example: Malcolm Rewa / Susan Burdett:
    • Susan Burdett was raped and murdered in 1992. Teina Pora (aged 17) falsely confessed to gain a monetary reward and was wrongfully convicted.
    • DNA later identified Malcolm Rewa. Linkage analysis using ViCLAS linked the scene to 43 sexual offenses across 25 complainants. Rewa was convicted of the sexual assaults, but the jury hung on the murder charge. After the Privy Council quashed Pora's conviction, Rewa was retried and convicted of Burdett's murder based heavily on distinct crime scene signature linkages matching his prior offenses.

• Linking Property Crimes (Burglaries):

  • Empirical studies demonstrate that geographic proximity (distance between offenses) is the single most powerful statistical predictor for linking burglaries to the same offender.
  • Geographic distance significantly outperforms physical entry method, target selection, or type of property stolen (which fluctuate situationally based on opportunistic factors like unlocked doors or visible items).

• Limitations of Statistical Profiling:

  • Data Coding Difficulties: Quantifying qualitative crime scene data is complex (e.g., defining precise operational thresholds for what constitutes "concealing a body" or "moving a body").
  • Operational Gap: Primarily remains an academic research tool with limited systematic implementation in routine police practice.
  • Offender Inconsistency: Assumes behavioral consistency across time, but empirical research shows low consistency rates in serial offenders:
    • Offense planning consistency: $42\%$
    • Injury/wound style consistency: $\approx 33\%$
    • Victim interaction consistency: $\approx 60\%$

Evaluation of Profiling Utility & Psychological Biases

• The "Coals to Newcastle" Utility Study (1990s):

  • Evaluated the operational impact of profiling advice provided to UK police detectives across major investigations.
  • $80\%$ of detectives stated the profile was "useful."
  • $60\%$ reported it deepened their understanding of the case/offender.
  • $50\%$ admitted it merely confirmed what they already believed (reinforcing confirmation bias).
  • Only $14\%$ stated the profile actively assisted in solving the case.
  • Only $3\%$ reported the profile directly led to offender identification.

• Expertise Meta-Analysis (2007):

  • Compared professional self-described profilers to untrained college students.
  • Profilers achieved $\approx 66\%$ accuracy ($2/3$) on standardized test cases compared to students at $\approx 33\%$ ($1/3$).
  • However, extremely wide confidence intervals, small sample sizes, and methodological biases favoring profilers led researchers to conclude that formal evidence for professional profiling expertise remains unproven.

• Cognitive Biases Sustaining Belief in Profiling:

  • The Barnum (Forer) Effect: The tendency for individuals to accept vague, highly generic, double-sided personality descriptions as uniquely accurate interpretations tailored specifically to their case.
  • Experimental Validation: Detectives presented with a genuine profile versus a completely fake, generic profile rated both as equally accurate and useful for the exact same case file.
  • Unfalsifiability: Many professional profiles contain contradictory or overly vague statements, making them logically impossible to prove false.
  • Other Contributing Biases: Hindsight bias (remembering accurate claims while forgetting incorrect ones), availability heuristic, and commercial incentives (lucrative true-crime books and media appearances).

• Professional Regulation:

  • The title "Profiler" is legally unregulated in most jurisdictions.
  • In the UK, the role is regulated under the title Behavioral Investigative Adviser (BIA). BIAs are formally accredited, and their reports must fulfill strict criteria: advice must be empirical, fully referenced, logical, operationalized, and strictly falsifiable.
  • The New Zealand Police Behavioral Science Unit focuses operationally on crime linkage analysis using ViCLAS, rather than offender profiling.

Geographic Profiling

• Core Assumptions:

  • Crime is not randomly distributed across space; it follows measurable, predictable spatial patterns.

• Theoretical Underpinnings:

  1. Crime Pattern Theory: Offending occurs within an offender's familiar "awareness space," derived from routine non-criminal movements (home, workplace, school, social hubs, whānau locations).
  2. Routine Activity Theory: An offense occurs when three elements converge in space and time:
     • A motivated offender
     • A suitable target
     • The absence of a capable guardian (basis for situational crime prevention).

• Offender Spatial Typologies:

  • Marauder: Offender operates and commits crimes within a spatial boundary encompassing their home base. Offending clusters around the home location.
  • Commuter: Offender travels out of their residential awareness space into a distinct secondary area (e.g., commercial zones, transit hubs) to commit offenses before returning.
  • Spatial Phenomena:
    • Distance Decay: Offending frequency declines as distance from the home base increases.
    • Buffer Zone: A small area immediately surrounding the offender's residence where offending drops off to minimize the risk of recognition by neighbors.
    • Career Progression: Earlier offenses tend to occur closer to the home base, expanding outward over time.
  • Base Rates: Approximately $80\%$ of serial offenders act as marauders.

• Geographic Profiling Systems & Algorithms:

  • Computer software packages (e.g., Rigel / Criminal Geographic Targeting - CGT) run spatial calculations across map coordinates to generate 3D probability surfaces ("heat maps") prioritizing search areas.
  • Hit Score Percentage: The percentage of the total map area that law enforcement must search before encountering the offender's home base (lower percentages indicate higher tool accuracy).
  • Error Distance: The absolute physical distance between the predicted spatial point and the actual residence.
  • Heuristic Performance: Research indicates laypeople provided with two basic heuristics—(1) the offender likely lives near the spatial center of offenses, and (2) offending decays with distance—perform nearly as accurately as complex mathematical algorithms.
  • Generalizability: Geographic profiling tools perform reliably in grid-structured North American cities, but show lower predictive accuracy in irregular, historical European urban layouts (e.g., Warsaw).

• Genetic Genealogy in Offender Spatial Identification:

  • Linking crime scene DNA to open commercial genealogy databases (e.g., GEDmatch, Ancestry, 23andMe).
  • Case Example: Golden State Killer (Joseph DeAngelo): Apprehended 30 years after his final offense by uploading crime scene DNA to GEDmatch. Investigators mapped a 1,000-person family tree back to a shared 3x great-grandfather, narrowing suspects using age, sex, and geographic location relative to the offenses. Confirmation was achieved via abandoned DNA collected from a discarded tissue and his car door handle.

Police Decision-Making & Cognitive Biases

• Dual-Process Cognition in Investigations:

  • System 1: Fast, automatic, effortless, pattern-seeking, highly vulnerable to cognitive heuristics.
  • System 2: Slow, effortful, logical, resource-intensive.
  • System 1 processing dominates police decision-making under time pressure, leading to systematic errors in suspect selection and investigative focus.

• Tunnel Vision & Early Closure:

  • Tunnel Vision: Prematurely focusing on a single theory or prime suspect, systematically ignoring, misinterpreting, or discarding disconfirming evidence.
  • Identified as a primary contributing factor in almost every documented wrongful conviction.
  • Empirical Findings: Experimental studies demonstrate that police officers are statistically more vulnerable to tunnel vision and overconfidence in suspect guilt than laypeople when evaluating case materials.

• Remediation & Debiasing Interventions:

  • Effective Strategies: Procedural system interventions, such as mandatory structured review checklists, training in generating multiple competing hypotheses, and appointing an independent Devil's Advocate to challenge investigative assumptions at every phase (e.g., the Dutch police model).
  • Ineffective Strategies: Basic educational debiasing or self-awareness training for individual officers (awareness of personal bias fails to mitigate its operational impact).

• Additional Heuristics Impacting Investigators:

  • Base Rate Neglect: Ignoring general population prevalence rates when assessing suspect traits (e.g., matching a suspect because he is a 35-year-old male, while ignoring that thousands of 35-year-old males reside in the immediate area).
  • Anchoring: Fixating on the initial piece of evidence or initial hypothesis, framing all subsequent evidence adjustments around that fixed starting point.
  • Availability Heuristic: Overestimating the likelihood of a scenario based on recent, highly vivid, or memorable cases.
  • Representativeness Heuristic: Judging a suspect's probability of guilt based on how closely they match a subjective stereotype of a typical offender.

• Systemic Failure Example: The Yorkshire Ripper (Peter Sutcliffe):

  • Police interviewed Sutcliffe 9 separate times, releasing him repeatedly because he did not match their mental stereotype of the perpetrator.
  • Anchoring Failure: Investigators anchored on a series of hoax letters and an audio cassette sent by an individual with a specific regional accent ("Wearside Jack"). Because Sutcliffe's accent did not match the tape, leads pointing to him were discounted.
  • Systemic Outcome: The massive information overload and cross-referencing failures in this case directly led to the development of centralized, computerized police database systems (such as HOLMES in the UK).