Geographic Profiling Study Notes

Geographic profiling (GP) defined as a technique for predicting an offender's home location based on crime locations.
Focus on understanding:
- How geographic profiles are constructed.
- The perceived usefulness and accuracy of GP.
- Examination of core GP conditions prior to profile construction.
- Types of cases where GP is employed.
Results show GP's usage spans a broad range of crimes and its high accuracy perceived especially with computerized systems.
Research suggests a gap in understanding of GP's operational use.
Keywords: Geographic profiling, serial offenders, crime investigation, investigative psychology.

GP defined as: "a criminal investigative methodology that analyses the locations of a connected series of crimes to determine the most probable area of offender residence" (Rossmo, 2012: 144).
Primarily a tool for prioritizing suspects by proximity to predicted offender locations (Rossmo, 2000).
Investigative gap noted in the procedures and operational settings of GP.

Two primary assumptions:
- Distance decay: offenders do not venture far from home to commit crimes.
- Domocentricity: offenders reside within the area of their crimes.

GP methods classified into:
1. Spatial Distribution Strategies:
- Use crime site distribution to calculate central location (offender's predicted residence).
- Examples include:
  - Centroid: average location based on x- and y-coordinates of crimes.
  - Center of the Circle: Based on the farthest crime points.
  - Center of Minimum Distance: Minimizes distances to crime sites.
1. Probability Distance Strategies:
- Use mathematical functions (linear, lognormal, etc.) applied to linked crime sites to predict residence, creating a probability surface for finding the home.
Computerized systems employing probability distance strategies (e.g., Rigel, CrimeStat, Dragnet) predominantly used in GP today.

Various studies indicate varying accuracy levels.
Rossmo (2000): FBI case study on Rigel yielded an average hit score percentage of 6% (6% of the search area needed to be examined).
Canter et al. (2000): Study of Dragnet yielded an average hit score of 11%.
Snook et al. (2005): No relationship between complexity of GP strategies and accuracy.
- Comparison of 11 GP strategies showed simpler methods can be as effective as complex ones.
Research by Paulsen (2006a & b) indicated heuristic methods could equal machine performance in accuracy with training.

Hit Score Percentage: area that needs to be searched before finding the offender's residence.
Error Distance: distance between predicted and actual residence of offender.
Debates exist on measurement reliability, with some arguing hit score percentage reflects practical GP use better.

Five conditions for GP feasibility (Rossmo, 2000 & 2005a):
1. Minimum of five crimes committed.
2. Crimes linked to a single offender with a complete series.
3. Offender is not a commuter.
4. No movement of anchor points during crime series.
5. Uniform distribution of suitable targets around the offender's home.
Difficulty verifying conditions in real-time investigations highlighted.
Commuters vs. marauders (Canter and Larkin, 1993) illustrated to demonstrate common investigative challenges.

Paulsen's (2007) study: 60% accuracy in commuter/marauder predictions improved to 81% with knowledge of geographic variables used.

Originally for serial murder; now broadly applied to:
- Rape, arson, robbery, bombings, burglary, fraud, auto theft, kidnappings (Rossmo, 2012).
Accuracy varies by crime type (Paulsen, 2006a): e.g., auto theft yields better GP accuracy than commercial robbery.
Offenders of interpersonal crimes found more likely to be marauders.

Online survey through SurveyMonkey distributed to police professionals globally for feedback on GP usage.
Recruitment included direct contact, mailing lists, and conferences.

47 total questions assessing:
1. Construction of geographic profiles.
2. Usefulness and accuracy perceptions.
3. Examination of GP conditions.
4. Case types applying GP.
Variability in responses due to branching questions within the survey.

Majority (91%) of profiles constructed individually.
77% used computerized systems, 27% spatial methods, and 27% educated guesses to create profiles.
Notable differences in method usage related to training received noted.

GP seen as valuable in half of the cases (average M = 53.2%).
Computerized systems rated the highest in accuracy (75%) relative to other methods (spatial techniques and educated guesses).

Most respondents apply GP even when conditions are unmet.
Particularly for conditions regarding commuter status and uniform target distribution.
Majority agreed on importance of validating crime series linked to the same offender (74.1%).

Majority applied GP to burglary, robbery, and murder, with notable usage reported in series involving multiple crime types.

Computerized GP systems dominate usage with positive perceptions.
Notably used even when core GP conditions violated, implying the field's need for further research and training emphasis.

Notable small sample size and dropout rates may limit broad applicability of findings.
Future studies should streamline methods to enhance participant engagement and include feedback from profilers for tailored survey design.

Findings provide insights into GP's operational use, reinforcing its prominent position in law enforcement despite unmet conditions. Further research is warranted to understand effectiveness under diverse scenarios, particularly considering the violation of GP conditions in real-world applications.