The Journal29 August 20267 min read
Fewer Than One Per Cent of All Murders
What the numbers say about serial offenders — and what the cinema invents
In 2005 the FBI sat down with investigators, researchers and prosecutors for a symposium on serial murder. The report that came out of it in 2008 is not a thriller but a clearing-up document. The figure that puts everything else in proportion stands right at the start:
serial murder is a relatively rare event, estimated to comprise less than one percent of all murders committed in any given year
Hardly any offender figure is culturally larger and statistically smaller. It fills podcasts, streaming series, novels and university seminars, and has done so for decades. The imbalance follows a heuristic that Amos Tversky and Daniel Kahneman described in 1973: people judge how often something happens by how easily examples come to mind. Rare, vivid, often-told cases are effortlessly available — and therefore seem more typical than they are. The serial killer is the model case of this bias, because it brings everything that produces availability: a name, a face, a story with an arc.
What gets counted and what does not
Even the definition is less tidy than the genre. The 2005 symposium settled on a very spare formula: the unlawful killing of two or more victims by the same offender or offenders in separate events. That was a lowering. Older definitions required three victims.
More important still is what was struck out. For a long time the so-called cooling-off period counted as a defining mark — the pause between offences in which the offender supposedly returns to ordinary life. The symposium discarded the criterion as useless for investigative practice. It can neither be established reliably nor cleanly distinguished from accidents of opportunity. Anyone reading figures on serial offenders is therefore always reading a definitional decision as well. Change the threshold from three victims to two, and the population changes.
No psychological kind called “the serial killer” follows from any of this. Motives, offence patterns, social circumstances and psychological characteristics are heterogeneous. What links the cases is a counting criterion, not a diagnosis.
The catalogue of myths
The main body of the FBI report is a list of what is not true, and it hits the popular figure at almost every point. Serial offenders are not uniformly white — their distribution roughly matches the population. They are not uniformly loners; many are socially embedded, with work, family, neighbours. They are not uniformly sexually motivated; control, profit, anger, or several motives side by side occur just as much. They are not uniformly mobile, but frequently act within a narrow geographical radius. And they are neither reliably highly intelligent nor reliably mentally ill.
Women appear markedly less often. That is not an invention of the cliché but the finding. Yet the rarity is amplified by the cliché: whoever looks for spectacular sexualised violence overlooks offence patterns that look different — and thereby overlooks female offenders too. A picture cut to fit one type of motive works like a filter on the investigation.
The most tenacious myth is that of genius, and it is also the most instructive, because it contains a confusion. That a series goes long undetected is read as an achievement of the offender. The report names the more sober reasons: missing data exchange between agencies, jurisdictional boundaries, earlier investigative errors — and victims from marginalised groups whose disappearance is reported late or not at all. The underlying problem has a technical name, linkage blindness: two cases sit in two registers, and nobody lays them side by side. Whoever reads non-detection automatically as offender superiority turns institutional gaps into personal brilliance.
| The myth | What the report says about it |
|---|---|
| always white | the distribution roughly matches the population |
| a loner | many are socially embedded — work, family, neighbours |
| sexually motivated | also control, gain, anger, several motives at once |
| travelling | often a narrow geographical range |
| highly intelligent | neither regularly gifted nor regularly mentally ill |
| going undetected is his achievement | missing data sharing, jurisdictional borders, victims reported late |
What crime scenes actually yield
The field’s most famous ordering is the distinction between “organised” and “disorganised” offenders: on one side the planned offence, the tool kit brought along, the cleaned-up scene — on the other the chaotic, the spontaneous, the left behind. A rule of thumb became a typology, and the typology a narrative pattern.
David Canter and colleagues tested it in 2004. They took 39 crime-scene features from 100 killings by 100 American serial offenders and put them through multidimensional scaling — a procedure that shows which features occur together. If there were two offender types, two clouds of features would have to emerge. They did not. Organised features characterise most serial killings in general; disorganised ones are rare and form no group of their own but occur scattered. As rigid offender types the distinction is therefore untenable.
The objection reaches beyond this one typology. Snook and colleagues drew up the balance sheet for the whole field in 2008, under the title “The Criminal Profiling Illusion”. Offender profiling, they argued, lacks a load-bearing theoretical foundation, and strong empirical support is likewise absent. Laurence Alison and colleagues counter that investigators do benefit from expert advice — only it has to answer narrowly framed questions rather than deliver a portrait of a person.
How easily a portrait deceives Alison had already demonstrated in 2003 with Matthew Smith and Keith Morgan. They presented police officers and forensic professionals with one and the same profile, once alongside the real offender, once alongside an invented one. Both times the majority found that it fitted: more than 75 per cent rated the profile as at least somewhat accurate. After an arrest what stands out is above all what matches — the same mechanism this archive describes under the Barnum effect. More useful than a dramatic total personality are therefore concrete behavioural traces and the linking of cases to one another, which is exactly what linkage blindness lacks.
Where the danger actually lies
The figures on lethal violence as a whole say something other than the genre does. In the FBI’s expanded homicide data for 2019, victims were acquaintances of the offender in 28.3 per cent of cases and family members in 13.0 per cent; strangers accounted for 9.9 per cent. A qualification belongs with this that is rarely quoted alongside: in 48.9 per cent of cases the relationship remained unknown. No rate of stranger killings can therefore be derived from these numbers. What can be derived is a ratio: among the cases where the relationship was known, around four fifths of victims knew their killer.
That is the point at which the bias becomes practical. If the picture of danger is set by the exotic stranger, attention — private as well as political — is directed at the rarest case. The more frequent risks lie in partnerships, families and known social relationships.
The storytelling itself has costs too. True crime can inform, but attention is a limited resource. When an offender’s name, face and childhood are reconstructed endlessly while the victims shrink to a number, the story rewards precisely the visibility some offenders sought.
What remains
What is secure is the order of magnitude, not the portrait. That serial murder accounts for fewer than one per cent of murders is in the FBI report; that the popular marks of the figure — white, solitary, sexually motivated, travelling, highly intelligent — do not hold up against the cases is in there too. What is disputed is what investigators can derive from a crime scene instead: the organised/disorganised typology is refuted, and the usefulness of profiling work overall is contested. And much stays unknown for the plain reason that the data are missing — the barely under half of unresolved victim–offender relationships in the 2019 statistics is the most honest example.
In everyday life the error of thinking shows up in a simple test. If a danger can be illustrated at once with a name and a face, the odds are high that you are overestimating how often it happens — and underestimating the dangers that come without pictures. Rare crime deserves explanation. It does not deserve the power to design our entire image of evil.
Sources, and why they are here
Federal Bureau of Investigation (2008). Serial Murder: Multi-Disciplinary Perspectives for Investigators. Behavioral Analysis Unit, National Center for the Analysis of Violent Crime.
The clearing-up document the quotation comes from — and the catalogue of myths on which the popular figure fails at nearly every point.
Federal Bureau of Investigation (2019). Expanded Homicide Data, Crime in the United States.
The figures on homicide overall, together with the qualification rarely quoted alongside them: in almost half the cases the relationship remained unclear.
Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and probability. Cognitive Psychology, 5(2), 207–232.
The mechanism behind the imbalance: how easily examples come to mind determines the estimated frequency. The serial killer brings everything that produces availability.
Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
The working-out of the same heuristic for a general readership — and the place where most readers have met it.
Canter, D., Alison, L. J., Alison, E., & Wentink, N. (2004). The organized/disorganized typology of serial murder: Myth or model? Psychology, Public Policy, and Law, 10(3), 293–320.
The test of the field's most famous ordering against 39 crime-scene features from 100 killings — and its result: no two clusters of features, and so no two types of offender.
Alison, L., Smith, M. D., & Morgan, K. (2003). Interpreting the accuracy of offender profiles. Psychology, Crime & Law, 9(2), 185–195.
The same profile, once beside the real offender and once beside an invented one — both times the majority found that it fitted. The Barnum effect in the case file.
Snook, B., Cullen, R. M., Bennell, C., Taylor, P. J., & Gendreau, P. (2008). The criminal profiling illusion: What's behind the smoke and mirrors? Criminal Justice and Behavior, 35(10), 1257–1276.
The balance sheet of the whole field: neither a sound theoretical foundation nor strong empirical support.
Alison, L., Goodwill, A., Almond, L., van den Heuvel, C., & Winter, J. (2010). Pragmatic solutions to offender profiling and behavioural investigative advice. Legal and Criminological Psychology, 15(1), 115–132.
The opposing position, and the more usable one: advice helps when it answers narrowly framed questions rather than supplying a portrait of a person.
Hickey, E. W. (2016). Serial Murderers and Their Victims. Cengage Learning.
The field's most extensive case collection — and one of the few sources that does not carry the victims merely as a count.