Everyday legal compliance

Why do most people obey laws when the chance of ever being caught is tiny?

Consider what you did not do today. You did not pocket anything from a shop with no visible camera, did not misreport a number on a form nobody will audit, did not park across a driveway on an empty street at midnight. For most such acts the chance of being caught is very small, and for some effectively zero. If obedience were a calculation about getting caught, the calculation would frequently come out the other way — and yet it almost never does. So either people are very bad at arithmetic, or obedience is not arithmetic.

Take the calculating account seriously first, because it is respectable. In Gary Becker's 1968 formulation, an offender weighs the gain against the punishment discounted by the chance of receiving it. It is a good model of some behaviour, and it makes a testable prediction: raise the expected cost and offending should fall.

Now put real numbers against it. Most crime is never reported; of what is reported, only a fraction is cleared with a suspect identified; of those, fewer still are convicted. Multiply the stages and the expected penalty for many everyday offences is small — and for the mundane ones we are asking about, smaller still, because nobody is looking at all. Yet compliance is not merely common, it is overwhelming. The model does not fail because it is wrong about incentives; it fails because the incentives it names are far too weak to carry the observed result.

There is a second problem, and it is the more revealing one. The calculating account assumes a decision is being made. Ask yourself honestly whether you considered and rejected stealing from a shop this week. Almost certainly the option never entered the frame. Whatever is doing the work happens before deliberation, which is why an account built entirely out of deliberation cannot reach it.

So what is doing the work? Tom Tyler's answer, from panel studies of Chicago residents in the 1980s, is legitimacy: people comply because they regard legal authority as entitled to be obeyed. His second finding is where the money is — legitimacy tracks not whether people liked the outcome they got, but whether they judged the procedure fair. The ingredients turn out to be having a voice, being treated with respect, and believing the decision-maker was neutral and well-motivated. A driver stopped and ticketed but treated courteously comes away with legitimacy intact; one who won their case but was treated with contempt often does not.

That reframing changes what enforcement is for. If a police force's job were deterrence, its priority would be being feared; if legitimacy carries most of the load, its priority is being seen to act fairly — and a tactic that raises fear while reducing perceived fairness could easily produce less compliance overall.

One more ingredient, distinct from both: people read the situation for what others are doing. Cialdini's distinction helps here — an injunctive norm says what is approved, a descriptive norm says what is actually done, and the descriptive one moves behaviour surprisingly hard. Field experiments found that a sign reporting that many visitors take petrified wood increases theft, presumably by advertising the practice as normal. Obedience is partly the reading of a room.

The three accounts are not rivals so much as different layers, and the interesting part is how they interact.

Deterrence has not vanished; it sets a floor. Criminological work consistently finds that certainty of detection matters far more than severity, which is why visible enforcement in a specific place at a specific time can change behaviour while longer sentences often do not. But a floor is not the building.

Legitimacy and descriptive norms are also entangled. Widespread visible non-compliance — everyone evading a tax, nobody queuing — does more than change the perceived odds; it undermines the sense that the rule is genuinely the community's rule, so the two mechanisms feed each other in both directions.

And the honesty caveat matters here more than usual. Most legitimacy research rests on surveys: people report their attitudes and report their own compliance, in the same instrument. That leaves two problems unresolved. Self-reported compliance is a soft measure. And the causal arrow is genuinely disputed — it is entirely plausible that people who do not offend come to see the system as legitimate as a way of making sense of their own behaviour, rather than the legitimacy producing the obedience. Experimental and longitudinal work has tried to separate these, with mixed success, and replications outside the United States have found the relationship varies by context. The claim that legitimacy matters more than deterrence is well supported; the claim that it is the cause is supported less firmly than its popularity suggests.

• Why certainty of detection outperforms severity of punishment, and what that implies for sentencing policy.
• What happens to compliance when descriptive norms collapse — tax evasion, or traffic rules in cities where nobody follows them.

Every term the collection defines is gathered in the glossary.