The profile and the common: what offender profiling can and cannot do

On the morning of 15 July 1992, Rachel Nickell, who was 23, was walking on Wimbledon Common in south-west London with her two-year-old son and their dog. She was attacked and killed there. Her son survived. The investigation that followed became one of the most examined in British policing, not because of how it ended, but because of the method it used to get there.

The Metropolitan Police asked a clinical psychologist, Paul Britton, for a profile of the offender. A suspect emerged, Colin Stagg, a local man. There was no forensic evidence connecting him to the killing. Police then used Britton's advice in designing a covert operation, later known as Operation Edzell. An undercover officer, using the name "Lizzie James", wrote to Stagg over a period of months, inviting him to share sexual fantasies, in the hope that what he wrote would match the profile and amount to a confession.

It produced no confession. Stagg was charged anyway. In September 1994, at the Old Bailey, Mr Justice Ognall ruled the evidence from the operation inadmissible, describing it as deceptive conduct of the grossest kind. The prosecution offered no evidence, and Stagg was acquitted (R v Stagg, 1994).

Fourteen years later, in December 2008, Robert Napper, a man already detained for other killings, pleaded guilty to the manslaughter of Rachel Nickell on the grounds of diminished responsibility (R v Napper, 2008). What led police to him was not a profile. It was a cold-case review of the forensic evidence using newer DNA techniques.

A profile is a prediction

An offender profile is a set of claims about the unknown person who committed a crime, inferred from the crime: age, background, habits, personality, where they live. Fiction has made it look like a special kind of seeing. Research treats it as what it is, a prediction, and predictions can be tested.

Brent Snook and colleagues did that in 2007. They reviewed the profiling literature and ran two meta-analyses of the studies that compared profilers with other groups, such as detectives, psychologists, students. Their narrative review found that the case for profiling rested largely on commonsense justifications: anecdotes, testimonials and the reputation of practitioners. Their first meta-analysis found that self-labelled profilers and experienced investigators did not outperform comparison groups at predicting offenders' cognitive processes, physical attributes, offence behaviours or social history, and were only marginally better on overall offender characteristics. A second analysis, with profilers alone, was more favourable on some measures, but the authors flagged the methodological weakness of the data underneath (Snook et al., 2007).

Claims that cannot be checked

Laurence Alison and colleagues took a different approach. They collected 21 real profiles, written for major investigations in the UK and elsewhere, and broke them into individual claims: nearly 4,000 of them. They then asked of each claim whether the profile gave any grounds or reasoning for it, using a standard framework for analysing arguments (Alison et al., 2003).

Nearly 80% of the claims came with no grounds, no warrant, no backing and no rebuttal at all. They were simply asserted. And fewer than 31% of the claims were falsifiable, meaning that if the offender were caught, the claim could be shown right or wrong (Alison et al., 2003). A claim such as "the offender will have difficulty with relationships" cannot fail. It also cannot help.

How a profile shapes an investigation

The research on tunnel vision explains why a profile can do harm even when nobody intends it. Once investigators commit to a picture of the offender, they tend to read new information in its light and discount what does not fit (O'Brien, 2009). A profile is exactly such a picture, delivered early, with the authority of an expert. If it is vague enough to fit many people, it will fit whoever the investigation is already looking at, and it will seem to confirm that suspect.

That is the mechanism in the Wimbledon Common case, as the court saw it. The operation did not test whether Stagg was the killer. It tested whether Stagg could be drawn into producing words that resembled a profile, and treated resemblance as evidence. Findley and Scott describe this pattern as the institutional side of tunnel vision: once a case is organised around a suspect, the system's effort goes into building that case rather than testing it (Findley & Scott, 2006).

What responsible advice looks like

None of this means behavioural science has nothing to offer an investigation. It means the useful parts look less like a portrait and more like statistics and logistics: which offences are likely to be linked, which areas to prioritise, which lines of inquiry the base rates support. The standard Alison and colleagues proposed is a reasonable test for any profile (Alison et al., 2003):

• Every claim states its grounds: the data or case facts it rests on.
• Every claim states its reasoning: why those grounds support it.
• Claims are falsifiable: they could be shown wrong once the offender is identified.
• The profile says how confident it is, and what would change its mind.

Advice that meets that standard can narrow a search without choosing a suspect. Advice that does not meet it is a story, and an investigation can follow a story a long way down the wrong path. On Wimbledon Common, it took sixteen years for the path to reach the right man.

Short answers

Does criminal profiling work?

The evidence is thin. A meta-analysis found that self-described profilers did not clearly outperform comparison groups at predicting offender characteristics, and that the profiling literature rests largely on commonsense arguments rather than tested evidence (Snook et al., 2007).

What is wrong with a typical profile?

In a study of 21 profiles used in real investigations, nearly 80% of almost 4,000 claims came with no supporting grounds or reasoning, and fewer than 31% could be checked against the facts at all (Alison et al., 2003).

What happened in the Rachel Nickell investigation?

Police built a covert operation around a suspect, Colin Stagg, with advice from a psychologist who had profiled the offender. In September 1994 the trial judge excluded the evidence from that operation and the prosecution offered no evidence. In December 2008 another man, Robert Napper, pleaded guilty to her manslaughter on the grounds of diminished responsibility.

References

1. Alison, L. J., Smith, M. D., Eastman, O., & Rainbow, L. (2003). Toulmin's philosophy of argument and its relevance to offender profiling. Psychology, Crime & Law, 9(2), 173–183. doi.org/10.1080/1068316031000116265
2. Findley, K. A., & Scott, M. S. (2006). The multiple dimensions of tunnel vision in criminal cases. Wisconsin Law Review, 2006(2), 291–397.
3. O'Brien, B. (2009). Prime suspect: An examination of factors that aggravate and counteract confirmation bias in criminal investigations. Psychology, Public Policy, and Law, 15(4), 315–334. doi.org/10.1037/a0017881
4. R v Napper (Central Criminal Court, 18 December 2008) (unreported).
5. R v Stagg (Central Criminal Court, September 1994, Ognall J) (unreported).
6. Snook, B., Eastwood, J., Gendreau, P., Goggin, C., & Cullen, R. M. (2007). Taking stock of criminal profiling: A narrative review and meta-analysis. Criminal Justice and Behavior, 34(4), 437–453. doi.org/10.1177/0093854806296925