Systematic Social Observation of Public
Spaces: A New Look at Disorder in Urban
Neighborhoods
1
Robert J. Sampson
University of Chicago and American Bar Foundation
Stephen W. Raudenbush
University of Michigan
This article assesses the sources and consequences of public disorder.
Based on the videotaping and systematic rating of more than 23,000
street segments in Chicago, highly reliable scales of social and physi-
cal disorder for 196 neighborhoods are constructed. Census data,
police records, and an independent survey of more than 3,500 resi-
dents are then integrated to test a theory of collective efficacy and
structural constraints. Defined as cohesion among residents com-
bined with shared expectations for the social control of public space,
collective efficacy explains lower rates of crime and observed disor-
der after controlling neighborhood structural characteristics. Collec-
tive efficacy is also linked to lower rates of violent crime after ac-
counting for disorder and the reciprocal effects of violence. Contrary
to the “broken windows” theory, however, the relationship between
public disorder and crime is spurious except perhaps for robbery.
The answer to the question of how city life was to be possible,
then, is this. City life was made possible by an “ordering” of the
urban populace in terms of appearance and spatial location such
that those within the city could know a great deal about one
another by simply looking.
(Lyn Lofland, A World of Strangers: Order and Action in
Urban Public Space, 1973, p. 22; emphasis in original)
Visual signs of social and physical disorder in public spaces reflect power-
fully on our inferences about urban communities. By social disorder, we
refer to behavior usually involving strangers and considered threatening,
1
We thank Tony Earls, Albert J. Reiss Jr., Steve Buka, Jeffrey Morenoff, Richard
Congdon, and Matheos Yosef for their help in this project, and the NORC team led
by Woody Carter, Cindy Veldman, Jody Dougherty, and Ron Boyd for heroic efforts
in data collection. John Laub and the AJS reviewers provided helpful comments on
an earlier draft. The long-standing interest of Albert J. Reiss, Jr., in systematic social
1999 by The University of Chicago. All rights reserved.
0002-9602/2000/10503-0001$02.50
AJS Volume 105 Number 3 (November 1999): 603–51 603
American Journal of Sociology
such as verbal harassment on the street, open solicitation for prostitution,
public intoxication, and rowdy groups of young males in public. By physi-
cal disorder, we refer to the deterioration of urban landscapes, for exam-
ple, graffiti on buildings, abandoned cars, broken windows, and garbage
in the streets. Visible evidence of disorder, or what Albert Hunter (1985)
calls “incivilities,” have long been noted as central to a neighborhood’s
public presentation (Goffman 1963). Jane Jacobs’s classic observation of
urban life in the 1950s even then evoked a concern with the threats of
disorder to neighborhood civility (1961, pp. 29–54), especially the negotia-
tion of public encounters in the “world of strangers” (Lofland 1973).
2
The streets, parks, and sidewalks still belong to no one and therefore
to everyone. Disorder continues to be of theoretical interest precisely be-
cause of its visual salience and symbolism regarding the use of such
spaces. Even if we wish it were not so, disorder triggers attributions and
predictions in the minds of insiders and outsiders alike. It changes the
calculus of prospective home buyers, real estate agents, insurance agents,
and investors and shapes the perceptions of residents who might be con-
sidering moving. Evidence of disorder also gives a running account of the
effectiveness of residents seeking neighborhood improvement, and that
record may encourage or discourage future activism. Physical and social
disorder in public spaces are thus fundamental to a general understanding
of urban neighborhoods.
Neighborhood disorder has more specific bearing on the study of crime
as well. Research has established connections between disorder and both
fear of crime and crime rates (Skogan 1990; Kelling and Coles 1996). In
fact, a reigning theory posits that minor disorder is a direct cause of serious
crime. Originators of the “broken windows” thesis, Wilson and Kelling
(1982) argued that public incivilities—even if relatively minor as in the
case of broken windows, drinking in the street, and graffiti—attract pred-
atory crime because potential offenders assume from them that residents
are indifferent to what goes on in their neighborhood. The metaphor of
observation provided the inspiration and practical guidance for our work, and the
Center for Advanced Study in the Behavioral Sciences provided a most hospitable
intellectual environment for its fruition. Data collection was funded in part by the
John D. and Catherine T. MacArthur Foundation and the National Institute of Jus-
tice. Direct correspondence to Robert J. Sampson, Department of Sociology, Univer-
sity of Chicago, 1126 East Fifty-ninth Street, Chicago, Illinois 60637. E-mail: rjsam
@src.uchicago.edu
2
Goffman (1963) goes back yet further, to the obligation in medieval times to keep
one’s pigs out of the streets. In this case, the norms regulating public order refer
not just to face-to-face interaction among strangers or acquaintances, but the visual
ordering of the physical landscape (1963, p. 9). For example, Goffman writes of expec-
tations regarding the maintenance of sidewalks and keeping the streets free of refuse.
Hence, disorder in public places may be conceived in physical as well as social terms.
604
Public Spaces
broken windows is apt, insofar as the theory asserts that physical signs
of disorder serve as a signal of the unwillingness of residents to confront
strangers, intervene in a crime, or call the police (Greenberg and Rohe
1986; Skogan 1990, p. 75). Proponents thus assume that both physical
disorder and social disorder provide important environmental cues that
entice potential predators. The “broken windows” thesis has gained ascen-
dancy in criminology and has greatly influenced public policy, leading to
police crackdowns in numerous cities on the manifestations of social and
physical disorder. New York City is the best-known example of aggressive
police tactics to control public incivilities (Kelling and Coles 1996, pp.
108–56; Kelling 1998).
Taking seriously the idea that visual cues matter, this article applies
the method of systematic social observation (SSO) to the study of social
and physical disorder in urban neighborhoods. We depart from prior re-
search in three ways. First, we describe novel systematic procedures for
collecting observational assessments of public spaces using videotaping
procedures that produce a permanent visual record amenable to later cod-
ing and reinterpretation based on emergent insights.
3
Second, we formu-
late a hierarchical item-response model that identifies sources of error in
aggregating across observed disorder items within block faces and in ag-
gregating across block faces within some 200 census tracts. The method
yields high tract-level reliabilities for assessing both social disorder and
physical disorder.
The third and major goal of the article is to assess the sources and
consequences of neighborhood disorder. We do so by testing the associa-
tion of systematically observed disorder with independent measures of
officially recorded and survey-reported crime, census-based sociodemo-
graphic composition, and a survey-based measure that taps the collective
efficacy of residents in achieving informal social control. A theory combin-
ing structural constraints with local collective efficacy is presented as an
alternative to the “broken windows” interpretation of the disorder-crime
link. We also assess broader implications for reinvigorating the study of
urban communities based on systematic observation and video-based ap-
proaches.
SYSTEMATIC SOCIAL OBSERVATION OF DISORDER
In the spirit of the early Chicago school of urban sociology, we believe
that direct observation is fundamental to the advancement of knowledge
3
Building on Lofland (1973, p. 19), we define public spaces as those areas of a neigh-
borhood to which persons have legal access and can visually observe—its streets and
sidewalks, its parks, its places of public accommodation, its public buildings, and the
public sectors of its private buildings.
605
American Journal of Sociology
(Park and Burgess 1921; see also Whyte 1988). As Andrew Abbott (1997)
notes, one of the hallmarks of the Chicago school was its concern with
observing public places—not just abstract variables, but the sights,
sounds, and feel of the streets. Attempting to systematize such approaches,
more than 25 years ago Albert J. Reiss, Jr. (1971) advocated systematic
social observation as a key measurement strategy for natural social phe-
nomena. By systematic, Reiss meant that observation and recording are
done according to explicit rules that permit replication; he also argued
that the means of observation, whether a person or technology, must be
independent of that which is observed. By natural social phenomena, Reiss
(1971, p. 4) meant “events and their consequences, including properties of
organization, can be observed more or less as they occur.” As his main
example, Reiss described systematic observations of police-citizen encoun-
ters. He also noted the general import of the SSO method for assessing
physical conditions and social interactions within neighborhood settings
that survey respondents may be incapable of describing accurately.
Despite the potential yield of direct observation, the majority of re-
search studies linking signs of disorder with fear of crime and criminal
victimization have been based on residents’ subjective perceptions drawn
from survey responses.
4
The typical strategy in survey research has been
to ask residents how much of a problem they perceive disorder to be; the
standard finding is that perceptions of disorder predict fear of crime (Sko-
gan 1990; Perkins and Taylor 1996; Taylor 1997, 1999). The dearth of
independent assessments of neighborhood disorder poses a special prob-
lem for interpreting this linkage. As Taylor (1999) has argued, the high
correlation between fear and disorder may arise in part from shared sur-
vey-method variance. More fundamentally, however, the perception of
disorder seems also to reflect a psychological construct—perhaps fear it-
self (Garofalo and Laub 1978; Rountree and Land 1996). Residents fearful
of crime report more disorder than do residents who experience less fear,
even though both sets of observers are reporting on the same neighbor-
hood (Taylor 1999; Perkins, Meeks, and Taylor 1992). In this scenario,
the fear (or vulnerability) of residents might be said to induce perceptions
of disorder. Even the disorder-crime link is problematic, since victimiza-
tion experiences are usually measured in the same surveys used to assess
(perceived) disorder.
One of the primary obstacles to bringing independent and systematic
social observation to bear on this conundrum has been methodological
uncertainty, not just on how to properly conduct such observations, but
on how to properly assess their measurement properties at the neighbor-
4
Useful reviews of the empirical literature are found in Perkins and Taylor (1996),
Taylor (1997, 1999), Skogan (1990), and Skogan and Maxfield (1981).
606
Public Spaces
hood level (Raudenbush and Sampson 1999b). Another concern has been
cost, even though direct observations are potentially less expensive than
household surveys, with listing, screening, broken appointments, and re-
sponse rates eliminated. Yet another obstacle has been conceptual, stem-
ming from underappreciation of the yield of systematic observation for
one of the fundamental cleavages in sociological criminology—the reliable
and valid measurement of crime and deviance. Perhaps most important,
however, has been the psychological reductionism that flows from the
dominant theoretical and empirical focus on individuals.
An exception to the lack of independent observations of disorder at the
level of ecological units rather than persons is found in research by Taylor
and colleagues (Taylor, Shumaker, and Gottfredson 1985; Taylor, Gott-
fredson, and Brower 1984; Covington and Taylor 1991; Perkins et al.
1992; Perkins and Taylor 1996; see also Mazerolle, Kadleck, and Roehl
1998).
5
Using observations conducted by teams of trained raters walking
the streets, Taylor et al. (1985) assessed 20% of all occupied face blocks
in 66 Baltimore neighborhoods. A face block is the block segment on one
side of a street. They identified two physical dimensions that stood out
empirically: physical incivilities and nonresidential land use. These two
dimensions were reliable in terms of individual-level psychometrics (e.g.,
Cronbach’s alpha; interrater reliability) and were related as expected to
perceived disorder and fear of crime derived from neighborhood surveys.
More recently, Perkins et al. (1992) examined on-site assessments of block-
level physical incivilities in Baltimore. Controlling for social factors, phys-
ical incivilities predicted perceptions of crime-related problems. Yet, using
similar procedures in a different city, Perkins et al. (1993) report that resi-
dents’ perceptions and an independent rating of physical disorder were
not significantly correlated. Observed environmental items correlated
more strongly with multiple indicators of subsequent block crime than
did residents’ perceptions of the environment. Interestingly, residents’
perceptions of physical disorder correlated positively with fear, but not
after controls were introduced for income, stability, and racial composi-
tion.
Overall, then, the research record is mixed and curiously imbalanced.
Although specified as an ecological construct, neighborhood disorder has
been investigated mainly using individual perceptions and individual-
level research designs. The number of studies employing observational
ratings across multiple ecological contexts is small, and the correlation of
observed disorder with subjective perceptions varies by level of aggrega-
5
An early version of systematic observation based on single interviewer ratings in a
neighborhood survey was used in Taub, Taylor, and Dunham (1984) and Skogan
(1990).
607
