14.9 — How a Criminal Is Made

Two facts frame everything else.

One: crime is overwhelmingly a young man's activity. Plot offending against age in any country, in any era, and you get the same curve — rising sharply in adolescence, peaking in the late teens to early twenties, falling steeply afterwards. This is the age-crime curve and it is one of the most reproducible findings in criminology.

Two: a very small number of people commit most of it. In the long-running Cambridge study of London males and in similar cohorts elsewhere, a small percentage of the sample accounted for around half of all convictions.

So the useful question is not why people commit crimes. Most young men do something at some point and stop. The question is what separates the small group who continue.

The two paths

Terrie Moffitt's distinction organises this better than anything else, and it has held up well.

Adolescence-limited offending. Starts in the teens, involves ordinary-looking young people, is mostly minor, is heavily social, and stops by the mid-twenties without intervention. The usual explanation is a gap between physical and social maturity — old enough to want adult things, not yet permitted them — combined with peer influence.

Life-course-persistent offending. Starts early, in childhood. Continues across the life span. A small group, and it accounts for a disproportionate share of serious and violent crime.

And the two have different origins, which is the important part.

The persistent group is characterised by early neuropsychological difficulties — problems with impulse control, language and attention — interacting with an adverse environment. Neither alone is sufficient; the combination is what predicts.

Note what this means practically: most young offenders are the first kind, and treating them as though they were the second — through severe punishment and criminal records that close off legitimate routes — is likely to convert some of them into the second.

What actually predicts persistent offending

From longitudinal studies that followed children into adulthood. Sorted by strength, and none is destiny.

Early behavioural problems — aggression and impulsivity in childhood — is the strongest individual predictor.

Harsh, inconsistent or neglectful parenting. Inconsistency appears to matter as much as harshness: rules that are enforced unpredictably teach that consequences are random.

Poor supervision. Simply knowing where a child is and who they are with is one of the most protective factors identified.

A criminal parent, which operates through several routes — modelling, supervision, poverty, and the effects of imprisonment on a family.

Low family income and disadvantaged neighbourhood.

School failure and truancy. Frequently downstream of undiagnosed learning difficulties, hearing problems or attention difficulties — which is why they are so consequential and so addressable.

Delinquent peers, which is the strongest situational factor in adolescence.

Childhood maltreatment (13.10).

And substance use, which is both a cause and a consequence.

The mechanisms underneath

Low self-control is the most-cited individual factor — difficulty delaying gratification, poor impulse regulation, preference for immediate reward. Frequently traceable to early environment plus temperament (9.4).

Weak social bonds. Travis Hirschi's account: people do not commit crime when they have attachments they would lose, commitments they have invested in, activities that occupy them, and a belief in the rules. Crime is what happens when there is nothing to lose — which is why employment, marriage and stable housing predict desistance.

Strain. Robert Merton's version: a society that promotes goals — money, status — while restricting legitimate routes to them for some people produces pressure towards illegitimate routes.

Learning. Crime is largely learned socially, from peers and family — the techniques, the justifications, and the norms.

And situation. From 11.2, and this is the most underweighted factor. A very large share of crime is opportunistic — it happens because a specific opportunity existed at a specific moment. Which is why situational changes reduce crime more reliably than changes to people do.

What stops it

Desistance research is more useful than causation research and much less known.

Age. The single largest factor. Most people simply stop.

Employment, particularly stable and reasonably paid.

A stable relationship. Marriage is associated with desistance, and the association appears to be partly causal in some analyses — attachment produces something to lose and changes daily routine.

Distance from delinquent peers, frequently a consequence of the two above.

And identity change. This is the one that matters most in the qualitative research: people who desist typically describe coming to see themselves as somebody who does not do this (8.7) — not as somebody resisting the temptation, but as a different kind of person.

What does not work

Severe punishment as deterrence. The evidence consistently finds that certainty of being caught deters far more than severity of punishment. Longer sentences have little marginal deterrent effect.

Prison, for most offenders. Custody is generally associated with higher reoffending than community sentences for comparable people. It removes employment, housing and relationships — the three things that predict desistance — and supplies delinquent peers.

Scared-straight style programmes. Repeatedly evaluated and found to increase offending in participants. This is one of the clearest cases in criminology of a popular intervention doing the opposite of its intention.

Boot camps. No reliable effect on reoffending.

What does work

Early years support — parenting programmes, home visiting, quality preschool. The evidence here is among the strongest in the field, and the effects show up decades later.

Treating the addressable childhood problems: hearing, vision, learning difficulties, attention difficulties. A child who cannot follow lessons because of an untreated problem fails at school, and school failure is a major pathway.

Consistent supervision and consistent rules. Not harsh — consistent.

Keeping people in education and employment, including after an offence.

Cognitive behavioural programmes for offenders have real, if modest, evidence.

And situational prevention — better lighting, better locks, design, supervision of places rather than of people. Unglamorous and among the most effective things available, because it removes opportunity rather than trying to change dispositions.

The honest boundary

None of this excuses anything. Explanation and responsibility are different questions (13.9), and a person who understands exactly why they became who they are is still responsible for what they do.

And none of it is deterministic. Most children with every risk factor do not become persistent offenders. The predictors are probabilistic, and a great many people with terrible starts live entirely ordinary lives.

What the evidence does establish is that this is largely a developmental and situational phenomenon rather than a matter of individual wickedness — and that the interventions with the best returns happen twenty years before the crime.

What to do with this

When you read about a crime, notice the two questions. What should happen to this person is one question. What would have prevented this is a different one, and the answers rarely point at the same policy.

Treat childhood risk factors as addressable rather than as fate — hearing, learning, supervision, consistency.

And remember the certainty-versus-severity finding the next time somebody argues for longer sentences. The evidence points at being caught, not at how long afterwards.

Next: 14.10 closes Part 14 with the construct everybody has heard of and almost nobody has right.