Last Updated on September 14, 2026 by Karl Thompson
Broken windows theory holds that visible disorder in a neighbourhood — litter, graffiti, vandalism, public drinking — causes serious crime, because it signals that nobody is in control. It has been called the most influential article on crime prevention ever written, and it is the basis of zero tolerance policing and environmental crime prevention generally.
The evidence for it is much weaker than its influence suggests, and the way it fails is instructive. This post sets out what the theory claims, the evidence on both sides, and why the question turns out to be so hard to answer.
Broken windows is part of the Right Realist approach to criminology within A-level sociology crime and deviance, examined on Paper 3 (7192/3).
What the Theory Actually Claims
James Q. Wilson and George Kelling set the argument out in ‘Broken Windows’, published in The Atlantic in March 1982.
The date matters and it is frequently given wrongly, including previously on this page. Broken windows is 1982. Kelling and Coles’ book Fixing Broken Windows is 1996 — a later, different argument by different authors.
The theory is a sequence, and most of the argument about it is really an argument about which step in the sequence fails:
- A neighbourhood shows signs of physical disorder — a broken window left unrepaired, graffiti, litter, abandoned cars
- Residents read this as a sign that nobody cares and that informal social control has broken down
- Residents withdraw from public space, which removes the informal control that was there
- Offenders read the same signs and conclude that offending in this area is low risk
- More serious crime moves in, and the neighbourhood declines further
The policy conclusion follows from step 1: fix the windows, remove the graffiti, and arrest the fare dodgers, and the sequence never starts.
Notice that steps 2 and 3 are about residents and steps 4 and 5 are about offenders. The theory needs both to be true. Much of the research below tests only one of them, which is part of why the picture is so muddled.
Evidence Supporting Broken Windows Theory
The disorder experiments
The strongest experimental support comes from a series of field experiments conducted in the Netherlands by Keizer and colleagues and published in Science in 2008.
In the best-known of these, researchers left an envelope visibly containing a five-euro note protruding from a postbox. In the control condition the postbox and its surroundings were clean. In the experimental condition the area was littered or the postbox covered in graffiti.
In the tidy condition, 13% of passers-by took the envelope. In the disordered condition, 25% did. The disorder roughly doubled the rate of theft.
This is genuine support for a causal link between visible disorder and rule-breaking, in a controlled setting. It is also worth being precise about what it shows: that disorder increases minor opportunistic dishonesty in the moment. It does not show that disorder produces burglary, robbery or violence over months and years, which is what the theory claims.
Meta-analysis of disorder policing
Braga, Welsh and Schnell (2015) conducted a meta-analysis of thirty studies of policing strategies aimed at disorder, and found an overall statistically significant reduction in crime.
HOWEVER, the detail of that finding is the single most useful thing on this page for an essay. The reduction is driven by community and problem-solving approaches — police working with residents to identify and fix recurring local problems. Strategies based on aggressive order maintenance, meaning arresting people for minor offences, showed little or no effect on their own.
So the meta-analysis supports a version of broken windows that almost nobody implemented. The version governments adopted — zero tolerance — is the version that does not work.
Evidence Not Supporting Broken Windows Theory
Moving to Opportunity
The Moving to Opportunity programme was a large American housing experiment in which low-income families in five cities were randomly assigned housing vouchers enabling them to move out of high-poverty neighbourhoods.
Because assignment was random, the study can do something observational research cannot: separate the effect of the neighbourhood from the effect of the people in it. If broken windows theory is right, families moving from disordered to orderly areas should show reduced offending.
Reviews of the programme found little evidence of reductions in criminal behaviour among those who moved. Adults’ offending did not fall; effects on young people were mixed, and in some analyses offending among boys increased.
The conclusion the original version of this post drew — that the root cause of crime lies with individuals rather than places — goes further than the evidence supports. What Moving to Opportunity shows is that changing the physical and social environment around someone does not by itself change their offending, which is a serious problem for broken windows but not a demonstration that environment is irrelevant.
Disorder and crime may share a cause
The most damaging finding for the theory is that the relationship between disorder and crime may not be causal at all.
Research on Chicago neighbourhoods by Sampson and Raudenbush used systematic observation — researchers recording visible disorder street by street — combined with survey and crime data. Once the structural characteristics of a neighbourhood were taken into account, particularly concentrated poverty and what the authors called collective efficacy (the willingness of residents to intervene for the common good), the apparent link between disorder and serious crime largely disappeared.
Their conclusion is that disorder and crime are both consequences of the same underlying conditions rather than one causing the other. Deprived neighbourhoods with weak social ties have more litter and more burglary, and the litter is not causing the burglary.
If this is right, then cleaning up the litter is treating a symptom. It is the criminological equivalent of noticing that people in hospital are often in bed and concluding that beds cause illness.
Harcourt’s work reanalysing the New York data reached a similar conclusion: the statistical support for the disorder–crime link is much weaker than the theory’s influence would suggest.
The New York confounders
New York is the case always cited in support of broken windows, and crime there did fall sharply from the mid-1990s. But several other things were happening at the same time:
- The crack cocaine epidemic was declining, having driven much of the violence of the late 1980s
- The population in the 16–24 age group, which commits most street crime, was shrinking
- Incarceration rates were rising sharply
- Vehicle and household security was improving nationally
- Crime was falling across American cities that had not adopted the approach
Any of these could account for a large part of the decline, and the last is the most difficult to explain away: if crime fell everywhere, the thing that only happened in New York cannot be the explanation.
Why Broken Windows Is So Hard to Evaluate
This is a methods point as well as a crime point, and it is worth making in an essay because it demonstrates AO3 in a way that listing studies does not.
Disorder is hard to operationalise. How do you measure how much disorder a street contains? Is one abandoned sofa worse than twenty crisp packets? Is graffiti disorder if it is good? Skogan’s work on disorder has been criticised on exactly this ground: different researchers measuring the same streets produce different numbers, and the measure tends to pick up how deprived an area looks as much as how disordered it is.
Disorder and deprivation are almost impossible to separate. The areas with most visible disorder are the areas with least money, which are also the areas with most crime for reasons that have nothing to do with litter. Untangling the three requires either randomisation, as in Moving to Opportunity, or careful statistical control, as in the Chicago research — and both of those approaches have produced results unfavourable to the theory.
Perception is part of the theory and is rarely measured. The mechanism runs through what residents and offenders believe about an area. Almost all the research measures disorder as observed by researchers, not as perceived by the people whose behaviour the theory is about.
It is politically attractive, which distorts the evidence base. A theory that says crime can be reduced without spending money on deprivation is extremely convenient. Its popularity may owe more to that than to its evidential record, and a great deal of the research has been conducted by people with a stake in the answer.
| Broken windows theory — evaluation | |
|---|---|
| Supporting | Criticising |
| ✓ The Dutch field experiments show disorder roughly doubled minor theft (Keizer et al. 2008) | ✗ Those experiments show momentary dishonesty, not serious crime over time |
| ✓ Meta-analysis finds disorder policing reduces crime overall (Braga et al. 2015) | ✗ The effect comes from community problem-solving, not aggressive order maintenance |
| ✓ New York crime fell sharply after the approach was adopted | ✗ Crime fell in cities that never adopted it |
| ✓ Intuitively plausible and matches residents’ own accounts of decline | ✗ Moving to Opportunity: changing the neighbourhood did not reduce offending |
| ✓ Cheap to act on compared with tackling deprivation | ✗ Disorder and crime may share a cause — concentrated poverty and low collective efficacy (Sampson and Raudenbush) |
| ✗ Disorder is very difficult to operationalise consistently | |
| ✗ Its popularity may be political rather than evidential |
Signposting and Sources
This post is written for students studying the crime and deviance topic as part of the AQA’s A-level sociology course, examined on Paper 3 (7192/3).
Broken windows theory is introduced within Right Realist Criminology, which covers the right realist account of the causes of crime in full.
It is the basis of environmental crime prevention and of zero tolerance policing, both of which evaluate the policies rather than the theory.
For all three crime prevention strategies together, see Crime Prevention and Control Strategies.
The rival explanation — that crime comes from relative deprivation and marginalisation rather than from disorder — is Left Realist Criminology.
Please click here to return to the homepage – ReviseSociology.com
Sources
Wilson, J.Q. and Kelling, G.L. (1982) ‘Broken Windows’. The Atlantic, March 1982.
Keizer, K., Lindenberg, S. and Steg, L. (2008) ‘The Spreading of Disorder’. Science 322. [LINK NEEDED — the existing ppsw.rug.nl link needs checking and probably replacing]
Braga, A.A., Welsh, B.C. and Schnell, C. (2015) ‘Can Policing Disorder Reduce Crime? A Systematic Review and Meta-Analysis’. Journal of Research in Crime and Delinquency 52(4).
Sampson, R.J. and Raudenbush, S.W. (1999) ‘Systematic Social Observation of Public Spaces: A New Look at Disorder in Urban Neighborhoods’. American Journal of Sociology 105(3).
Harcourt, B.E. (2001) Illusion of Order: The False Promise of Broken Windows Policing. Harvard University Press.