In Groningen, in the Netherlands, researchers left a stamped envelope wedged halfway out of a public mailbox with a five euro note visible through its address window. When the mailbox and the pavement around it were clean, 13 percent of passersby took the envelope. When the same mailbox was covered in graffiti, 27 percent did. When the ground around a clean mailbox was covered in litter instead, the figure was 25 percent. The researchers, led by the environmental psychologist Kees Keizer at the University of Groningen, published the result, along with five other field experiments testing the same idea, in a paper in the journal Science in 2008.
This is one paper, built around six field experiments run by one research group in one Dutch city, not a settled account of how disorder works in every setting. It is nonetheless a genuinely well-designed set of real-world experiments rather than a lab simulation, and the pattern held up, in slightly different sizes, across all six.
What the six experiments actually tested
The team’s core question was narrower than “does a messy environment cause crime.” It was whether seeing one social norm visibly broken makes people more likely to break a different, unrelated norm shortly afterward.
In an alley where a sign banned graffiti, the researchers attached a flyer to parked bicycle handlebars. When the alley walls were clean, 33 percent of returning cyclists dropped the flyer on the ground instead of taking it. When the same wall was covered in graffiti, 69 percent littered. In a supermarket car park with signs asking shoppers to return their carts, littering with the same flyer rose from 30 percent, when the car park was tidy, to 58 percent, when four abandoned carts were left in view. Near a fenced-off construction detour, only 27 percent of pedestrians squeezed through a gap to avoid the detour when the fence was undisturbed, compared with 82 percent when several bicycles had already been illegally locked to that same fence. Even an audible cue worked: cyclists near a train station littered a flyer 52 percent of the time under ordinary conditions, rising to 80 percent after they heard fireworks being set off nearby, illegal in the Netherlands in the weeks before New Year’s Eve.
Why this is not quite “broken windows policing”
It is worth separating this study from the broader, more contested idea it is often lumped in with. The original broken windows theory, proposed by James Q. Wilson and George Kelling in a 1982 magazine essay, argued that aggressively policing minor disorder in a neighborhood, graffiti, loitering, fare evasion, prevents more serious crime from taking hold, and it became the intellectual basis for aggressive order-maintenance policing in several American cities. That specific policy claim, that cracking down on minor disorder measurably reduces serious crime at the city level, has fared poorly under later scrutiny; several reviews of the broader theory have found the evidence linking disorder to serious crime largely disappears once other factors, particularly poverty, are accounted for.
Keizer’s experiments test something narrower and more tractable: not whether policing disorder prevents burglary and assault across a city, but whether an individual, in the moment, is more likely to litter or take money that is not theirs after seeing an unrelated rule already broken nearby. That is a much smaller and more testable claim, and it is the one these six experiments were actually built to answer.
The explanation Keizer’s team offered for the pattern draws on earlier work on descriptive social norms by the psychologist Robert Cialdini, whose research distinguishes what people are told is appropriate from what they can see other people actually doing. A sign banning graffiti states an injunctive norm, what is supposed to happen. Graffiti covering the wall anyway is a descriptive norm, what evidently does happen. When the two conflict, Keizer’s data suggests many people quietly follow the second one. The researchers were careful to rule out a simpler explanation, that people simply assumed fewer police were watching a disorderly area; graffiti and litter are generally tolerated by police in Groningen regardless, so a change in perceived enforcement was an unlikely driver of the effect.
What the study does not show
All six experiments took place in and around one mid-sized Dutch city, with passersby who did not know they were part of an experiment and could not be tracked afterward, so the researchers could measure what people did in the moment but not who they were or why. The effect sizes varied considerably across the six setups, from a jump of roughly fourteen percentage points in the mailbox experiment to one of more than fifty percentage points in the bicycle-fence experiment, which suggests the strength of the effect depends heavily on the specific situation rather than being a single fixed number. And a culture where people behave somewhat differently around fireworks, bicycles, or shopping carts than Dutch pedestrians do might not reproduce the same figures at all.
What is left standing
What the six experiments converge on, more modestly than “disorder causes crime,” is that visible evidence of one broken rule seems to loosen people’s sense that rules in general are being enforced or expected, at least in that moment and that place. A tidy, well-kept setting does not just look nicer. In Keizer’s data, it appears to change what a fair number of ordinary passersby are willing to do when they think no one else is watching.
The word pair “prim and proper” describes an appearance more than a rule, and that is arguably the more precise reading of what these experiments actually measured. Nobody in Keizer’s mailbox or bicycle-fence setups was handed a written code of conduct. They were reacting entirely to how the physical setting looked, whether it read as a place where things were generally kept in order or one where they evidently were not. Whatever the practical implications for city policy, a much narrower and better-supported claim survives from this paper: an environment’s visible tidiness functions as information, and a meaningful share of people appear to use it, whether they mean to or not, when deciding how closely to follow a rule nobody is actively enforcing.