Data-Driven Approach

The Bottom-Up Approach

The bottom-up approach to offender profiling is data-driven: it begins with the specific evidence from a crime scene and uses statistical analysis to identify patterns, from which inferences about the offender are built. No prior typology is imposed. The approach originated in the UK and is associated primarily with David Canter, who applied psychological science to the analysis of criminal behaviour.

Investigative Psychology

Canter and Heritage (1990) conducted a content analysis of 66 sexual assault cases in the UK. Using smallest space analysis (a statistical technique that identifies co-occurring features), they identified five core behavioural themes consistent across offences:

• Interpersonal coherence — how the offender relates to the victim reflects how they relate to people in general
• Significance of time and place — offence location reveals information about the offender's home area and routine
• Criminal career — offence characteristics suggest prior criminal history
• Forensic awareness — precautions taken indicate knowledge of police procedures
• Criminal characteristics — style of offending suggests offender type

These themes form the basis of investigative psychology: by mapping crime scene behaviour onto statistical patterns, investigators can make evidence-based inferences about the likely offender.

Geographical Profiling

Canter and Larkin (1993) proposed the circle hypothesis: serial offenders tend to live within the circle formed by their two furthest crime scenes. They found that approximately 87% of serial offenders in their UK sample were marauders — operating from a fixed base and committing crimes in the surrounding area — rather than commuters (who travel to a different area to offend). Rossmo (2000) developed this into a mathematical model (implemented in software such as Rigel and CrimeStat) that generates a probability map of where the offender is likely to live, significantly narrowing police search areas.

Evaluation

The bottom-up approach is more scientifically credible than the top-down: it is data-driven, uses replicable statistical methods (smallest space analysis), and does not presuppose a typology that may be invalid. Geographical profiling in particular has practical utility — Rossmo's model has been used by police forces internationally to narrow search areas in serial offence investigations.

However, Alison, Smith, Eastman and Rainbow (2002) raised the Barnum effect concern: they gave the same offender profile to two groups of detectives, one with a matched case and one with an unmatched case. Detectives in both groups rated the profile as accurate for their case — suggesting profiles are often so vague and general that they appear to fit almost any suspect. This undermines the discriminative validity of profiling. Additionally, a profile that places an offender in the wrong geographical area or assigns incorrect characteristics can actively misdirect investigations, with potentially serious consequences for both the investigation and wrongly suspected individuals.