|Home | About | Journals | Submit | Contact Us | Français|
‘When they approach me they see only my surroundings, themselves or figments of their imagination – indeed, everything and anything except me’. Ralph Ellison, Invisible Man (1947:3)
Scholars of the city have long interpreted signs of disorder in public spaces in ways that constitute powerful forces of social differentiation. From observers of London in the 1800s such as Charles Booth (1889) and Henry Mayhew (1862), to authors of modern classics such as Jane Jacobs’s (1961) The Death and Life of Great American Cities, to present day concerns over ‘broken windows’ and crime, signs of disorder – especially when linked to the poor – have been viewed as central to understanding city life. By social disorder, observers commonly mean behaviour involving strangers and considered potentially threatening, such as verbal harassment on the street, open solicitation for prostitution, public intoxication, and rowdy groups of young males in public. Traditional conceptions of physical disorder typically refer to markers such as graffiti on buildings, abandoned cars, garbage in the streets, and the proverbial broken window.
Booth’s detailed investigations and resulting maps of Victorian London served as an early illustration of disorder’s role in the social ranking of places. His painstaking portrayal of this great city included colour codes for the economic and social make-up of its many streets (LSE 2008). The lowest classes, coded in black, were described as not just poor but living in ‘squalor’ with public displays of alcoholism. Expressing a view that many today probably still hold (if silently), Booth unabashedly labeled the lowest-class category as ‘vicious, semi criminal,’ with the lowest grade ‘inhabited principally by occasional labourers, loafers, and semi-criminals—the elements of disorder’ (Pfautz 1967:191). What Booth thus accomplished was a merger of ecological classification and spatial difference with a subtle yet potent moral evaluation based on behaviour, what might be considered the precursor to contemporary notions of the ‘underclass’. This was a consequential intellectual move, for the designation of areas as disreputable and disordered, as I shall argue, can set in motion long-term processes that reinforce the initial stigmatized state and thereby contribute to the social reproduction of inequality.
The relative stability of concentrated disadvantage is a rather remarkable and surprising phenomenon. In 2006 The Economist provided intriguing examples of the similar nature of neighbourhoods in London from Booth’s day to the present, even at a micro ecological level. Figure I reproduces a map of a section of the Chelsea neighbourhood, comparing 1898 with 2001 on the dimension of residential wealth. There is evidence of change, of course, with a general upgrading along the southernmost edge of the neighbourhood and on the west. We also know that cultural changes swept through the area, especially during the swinging sixties replete with Mick Jagger in tow. As The Economist notes, today Booth would notice designer clothing shops and a high concentration of Porches and BMWs (Economist 2006b). Yet considering the gap of over a century, one would not have gone wrong predicting that pockets of poverty in the area bounded roughly by Fulham, Sloane, and Cale streets would remain robust. ‘Evil looking drink-sodden old Irish women’ (Economist 2006b:52) have been replaced by the merely down and out or struggling.
Seven years later, I set out on October 19th, 2008 to observe Chelsea myself and update Figure I. Camera in hand, I took note of the internal differentiation of the area. Consistent with Figure I, there was a ‘middling’ character and evidence of graffiti on the interior streets (Figure II–Figure III; all photos by author), and on one transitional street I spotted an abandoned car door and litter near a sharp Mercedes Benz (Figure IV). Although a poster outside the police station across the street boasted that crime was down (Figure V), I came across a car metres away that had just been broken into on the driver’s side (Figure VI). But there was clear evidence of gentrification in the neighbourhood (e.g., see Figure VII) and near Fulham Road I observed all the trappings of wealth in the form of chic shopping outlets with apparently well-to-do shoppers stocking up to the point of needing assistance (Figure VIII) and posh residences replete with BMWs lined up in front (Figure IX).
Moving south of the Thames to Stockwell and nearby Brixton lie some of the most racially mixed neighbourhoods in London. Home of the ‘Stockwell Strangler’ in the mid 1980s and several police raids after the 2005 London bombings, racial tensions have flared in these neighbourhoods over the years. A century ago, just east of Stockwell Road, Booth and his research team ‘found a pocket of filth and squalor, with rowdy residents and broken windows’. It was, ‘far the worst place in the division’. The Economist reports that since then the area has been transformed but in a way that replicates important features of the past:
Dismal two-storey cottages have been swept away and replaced by grass and the apartment blocks of the Stockwell Park Estate. But the appearance of the neighbourhood has changed more than its character. Julie Fawcett, who lives in one of the blocks, characterises her neighbours as ‘the mad, the bad and the sad’‥
Unemployment is double the city average and ‘heroin alley’ hides around a corner.
Perhaps Booth’s crude distinctions are not so antediluvian after all. In the current era of globalization, the alleged ‘placelessness of place’ (Relph 1976), and place even as ‘phantasmagorical’ (Giddens 1990), the apparent durability of place exposes a puzzle. As the quote by Ms. Fawcett reveals, it is not just economic or racial status but identities and moral evaluations that in many cases remain sticky. What mechanisms sustain the hierarchy of places, or what the late Charles Tilly (1998) termed ‘durable inequality’? Anecdotes notwithstanding, how stable are disorder and its related neighbourhood dimensions such as concentrated poverty? What predicts perceptions of disorder – that is, what are its sources? What are its consequences?
My general thesis is that perceptions of disorder constitute a fundamental dimension of social inequality at the neighbourhood level and perhaps beyond. At first this might seem an odd thesis considering that dominant stratification theorists take a structuralist stance in analysing the materialistic constituents of inequality. Demographers do likewise in thinking about urban change. Whether expressed by Charles Booth or our Stockwell contemporary, perceptions about disorder are likely to be dismissed in favor of presumed weightier causes. By contrast, I argue that the grounds on which perceptions of disorder are formed are contextually shaped by social conditions that go well beyond the usual suspects of observed disorder and poverty, a process that in turn molds reputations, reinforces stigma, and influences the future trajectory of an area. ‘Seeing’ disorder, like seeing Ralph Ellison’s narrator, is intimately bound up with social meaning at the collective level and ultimately inequality. This conceptualization turns around the usual strategy in the study of crime of reifying disorder as part of the environment.
To assess my general position I present a set of empirical evidence based mainly on the social laboratory of Chicago. Along with colleagues I have spent over a decade intensively studying this quintessentially American of cities and the site of much seminal work. My hope is that the current analysis will spur new insights into stability and change in urban inequality in a way that has relevance for cities around the world. The implications of my analysis are in the end optimistic, owing to increases in immigration and diversity that are eliding racial (and other) group categories and upending dominant conceptions of disorder. Indeed, in many places ‘grit’ is becoming the new glamour (Lloyd 2006). Before getting to the empirical patterns, however, I first lay out the historical and theoretical groundwork that motivates the logic of my investigation.
Theorizing disorder can be seen as an outgrowth of the writings of the influential Chicago School sociologists of the mid-twentieth century. Louis Wirth, one of the most famous theorists of the American city who hailed from the Chicago School, emphasized the general ‘disorganizing’ effects of the disorder and diversity that he argued flowed from increasing urbanization (Wirth 1938). Although not in the Wirthian camp, Richard Sennett argues in The Uses of Disorder that concern with disorder is fundamentally a concern about the loss of control in an increasingly urbanized world. More than that, he argues it is about an attempt to restore the myth of the ‘purified community,’ to keep unknown and disorderly events at bay (Sennett 1970). For Sennett, anxiety over disorder is rooted in psychological needs for control and at a general theoretical level, to efforts aimed at restoring the imagined community of solidarity.2 That the community of solidarity never existed, much less the ‘urban village’ of close personal ties, makes the obsession with controlling disorderly people in urban spaces all the more interesting.3
Tangible manifestations of disorder, or what Hunter (1985) called ‘incivilities,’ were argued by another Chicago School theorist as central to a neighbourhood's public presentation of self. Erving Goffman (1963a:9) cites the obligation in medieval times to keep one’s pigs out of the streets to demonstrate how the norms regulating public order covered not just face-to-face interaction among strangers or acquaintances, but the visual ordering of the physical landscape. He also studied how shared expectations formed about the maintenance of sidewalks and keeping the streets free of refuse. Jane Jacobs’ observation of urban life in the 1950s evoked a broader concern with the impact of disorder on neighbourhood civility (1961: 29–54), especially the negotiation of public encounters in the ‘world of strangers’ (Lofland 1973).
These thinkers did not consider disorder in literal or essentialist terms, nor did they propose disorder as somehow random or chaotic. Disorder can be socially and spatially patterned—highly organized even. What was important to Goffman and Sennett, I believe, were the expectations and perceptions surrounding signs or cues, the rules and functions of urban social order, as it were. These symbolic expectations are as powerful as the signs themselves, motivating theoretical and empirical interrogation. The fundamental importance of public observation in the process of social order was put well by the American urban theorist Lynn Lofland:
The answer to the question of how city life was to be possible, then, is this. City life is 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 (Lofland 1973: 22, emphasis in original).
The key to Lofland’s argument is that social ordering is a visual process that involves classification. People divide the urban world into manageable bits, with one of the most important differentiating characteristics being signs of disorder. But there is more to this process than meets the eye. The human tendency to categorize racial and other groups despite their lack of scientific separateness, our ability to easily observe and code skin colour, and our sensitivity to the opinions of others in the form of reputations or identities that stigmatize areas of ‘the mad, the bad, and the sad,’ makes for a potent combination.
Debates about disorder and diversity in the urban context continue to inspire passion but with new twists. Diversity and the increasing presence of minority and immigrant groups in cities around the world has led to a growing social anxiety, with some scholars proposing a direct link between diversity and declines in public trust (Putnam 2007). Disorder in cities has produced similar anxiety and institutionalized action. According to the world famous ‘broken windows’ theory of urban decline, 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 predatory crime because potential offenders assume from them that residents are indifferent to what goes on in their neighbourhood. At its core, broken windows theory sees visual cues as objective and natural in their meaning – signs of disorder are negative and serve as a signal of the unwillingness of residents to confront strangers, intervene in a crime, or call the police (Skogan 1990:75). Proponents thus assume that physical disorder and social disorder provide important environmental cues that entice potential predators and eventually, crime.
Few ideas are more influential than broken windows in the urban policy world, with police crackdowns in numerous cities on elements of social and physical disorder. New York City is the most well known example of aggressive police tactics to control public incivilities (Kelling and Coles 1996:108–156). The tactics of broken windows policing and a neoliberal approach to public order have been exported around the world, even to liberal Paris and visibly so, to England. The government’s attempt to ‘soothe the savage beast’ and tamp down antisocial behaviour has led to a declared ‘war on incivility’ (Economist 2006a). In the Stockwell neighbourhood, the London police keep an ‘aggressive’ watch, which has apparently proven ‘a comfort to many Londoners’ yet provoked anger within the neighbourhood (Jordan 2005).
The concept of disorder has penetrated social psychology and the study of health as well, following the train of thought of broken windows theory. Here again the notion is that cues of disorder are a negative, but with harmful consequences for individual health and overall well-being. A growing number of recent studies have linked perceived disorder to physical decline, depression, psychological distress, and perceived powerlessness (Geis and Ross 1998; Ross, Reynolds and Geis 2000). Residents are thought to read signs of disorder as evidence of a deeper neighbourhood malaise, undermining personal health and trust (see Sampson and Raudenbush 2004).
Even if we wish it were not so, then, disorder triggers attributions and predictions in the minds of insiders and outsiders alike. It changes the calculus of prospective homebuyers, real estate agents, insurance agents, investors, the police, and politicians, and shapes the perceptions of residents who might be considering moving out or moving in. Evidence of disorder also gives a running account of the effectiveness of residents seeking neighbourhood improvement, and that record may encourage or discourage future activism. Physical and social disorder in public spaces is thus fundamental to a general understanding of how urban neighbourhoods work. Figure X presents a simple heuristic of both received wisdom and the operative stance of many theoretical accounts: the fact of urban disorder and its correspondence to perception is taken for granted, and the consequences are many, none of them good.
At one level disorder theory is on the right track by emphasizing the salience of visual cues. Isn’t it obvious that graffiti or drunken revelers are a problem? But imagine a situation where these same cues are not evaluated negatively. Perhaps the revelers are bankers on a bender, or the graffiti is on a street in Soho. Or perhaps signs of disorder creep into Chelsea as shown above. Does this still cause crime or urban decline? Or might it be perceived instead as ‘edgy’? Walking along the south side of the Seine in Paris one observes a long stretch of graffiti against the backdrop of couples strolling. Why is this ‘disorder’ not seen as problematic and why is Paris thriving? Despite the largely taken-for-granted notion of disorder, there remains a first-order question about what triggers our shared perceptions of it in the first place.
The prevailing view seems to be that seeing disorder is a straightforward matter of cues in the environment visible to our eyes, albeit with correspondence errors in perception. Ross and Mirwosky (1999: 414), for example, conceptualize perceived disorder as ‘a characteristic of the neighbourhood’, an objective place that generates consensus. But it is one thing to perceive, more or less accurately, what is in the objective environment, and another to assign it value, meaning, and to rate its seriousness. Here language and cognition become central, for the dominant method of asking (thinking?) about disorder is to have respondents assess ‘how much of a problem it is’. It follows that we can separate what is in the environment from how (or whether) it is perceived and how much it matters to the observer. Sociologically, we can further ask fundamental questions about context and social order: is the perception of disorder as a problem filtered or altered by the presence of stigmatized groups and disreputable areas? And even further, does seeing disorder as a problem depend on the collective judgments of others?
In a recent contribution to criminological theory, Wikström (2008) argues that the social context of individual-level perception is a missing link in attempts to explain acts of crime. He specifically argues that perceptions are the key to understanding alternative courses of social action. Although I am not attempting to explain crime, I take seriously the idea that the link between social context—in this case the neighbourhood—and perception is variable and necessary to explain. To recognize subjective variations in perception and meaning is not to give up on systematic scientific inquiry. Quite to the contrary, I argue that perceptions, especially when collective (or inter-subjective) in nature, form a causal ingredient that can constrain or enable social behaviour. As Bottoms and Wiles(1992:16) argue, perceptions of order and safety may be seen as rooted in shared understandings of the nature of particular areas and public spaces. Lamont (2000) makes a similar cultural point in her call for studies to assess social meaning in the form of ‘institutionalized cultural repertoires’ and ‘publicly available categorization systems’. What is the ‘the mad, the bad, and the sad’ if not a cultural repertoire and potentially even more consequential, a categorization system?
It turns out, however, that most research on disorder turns on the hegemonic broken-windows theory of whether disorder causes crime or any number of other outcomes (Figure X), and in particular, whether the aggressive policing of said disorder reduces crime. I too have weighed in on this debate (Sampson and Raudenbush 1999) but the results did not bear on the questions posed here regarding the sources and consequences of social perceptions. In a paper in 2004, Stephen Raudenbush and I thus set aside the standard form of inquiry and examined instead what explains a person’s perception that disorder, defined in the manner of broken windows theory, was a problem. Drawing on independent sets of linked data to be described, we examined how the racial, ethnic, and socio-economic structure of neighbourhoods predicted perceptions above and beyond the observable conditions of disorder. We argued that there are multiple mechanisms at work in translating cues in the environment to a rating of disorder.
We first evaluated broken windows theory on its own terms by tracing the logical consequences of the idea that disorder’s visual cues are unambiguous in meaning and that residents’ perceptions map neatly onto objectively observable aspects such as the amount of garbage, broken bottles, litter, graffiti, abandoned cars, and drug paraphernalia. After all, these cues are highly visible. If we imbue human beings with bounded rationality in the classic sense, then it makes sense to hypothesize that, according to the theory, the major factor leading to perceived disorder is externally assessed or observed disorder. Put simply, broken windows theory is a correspondence account of disorder that posits a direct disorder-perception link.
‘A matter of the construction of their inner eyes, those eyes with which they look through their physical eyes upon reality’ (Ellison 1947:3).
‘In the first instance, ‘race’ is a mode of perceptual categorization people use to navigate their way through a murky, uncertain world’ (Loury 2002: 17).
More challenging theoretically is what else happens on the pathway to the formation of social perceptions and their sequelae. Cultural attributions about disorder are prevalent in American society and increasingly in other cities as result of exported American policies, feeding the hunger that humans carry for social information. Stereotypes become especially tempting when, as is almost always the case, residents are not trained as systematic or neutral observers. Relatedly, if cultural stereotypes are pervasive and residents have uncertain information or ambiguous reactions to disorder, then they may, in a Bayesian like way (Rosenkrantz 1977), augment that information with contextual cues about people who can be seen on the streets. It follows that individuals may draw on their prior beliefs in judging whether disorder is a problem – that is, combining uncertain evidence with prior beliefs underwritten by cultural stereotypes.
Evidence from cognitive psychology suggests that categorical distinctions are important for the organization of information in everyday life (Fiske 1998). Categories of relevance are hardly random. Research suggests that Americans hold persistent beliefs linking blacks, disadvantaged minorities, and recent immigrant groups to many social images, including crime, violence, disorder, welfare, and undesirability as neighbors (e.g., Bobo 2001; Quillian and Pager 2001; Rumbaut and Ewing 2007). Beliefs about disorder are reinforced by the historical association of non-voluntary racial segregation with concentrated poverty, which in turn is linked to institutional disinvestments and neighbourhood decline (Massey and Denton 1993; Skogan 1990; Wilson 1987). As Glenn Loury (2002) argues, while race may be denied as a legitimate biological classification, dark skin is an easily observable trait that has become a statistical marker in American society, one imbued with meanings about crime, disorder, and violence that stigmatize not only people but also the places in which they are concentrated.
The use of racial and ethnic context to encode disorder does not mean that people are necessarily prejudiced in the sense of group hostility. The power of cultural stereotypes is that they can operate beneath the radar screen of our conscious reasoning, forming what has been termed implicit bias (Banaji 2002; Bobo 2001; Fiske 1998). Suppose that someone without racial animus has none the less been exposed to the historically and structurally induced inequality that is urban America: on average, for example, rates of violence such as homicide are higher among blacks than whites. Implicit bias arises when this person automatically concludes from such a statistical generalization that a specific black person, without corroborating evidence, is prone to violence. Research in social psychology has shown that automatic racial stereotypes can persist regardless of conscious or personal rejection of prejudice toward blacks (Devine 1989), leading to what Bobo (2001: 292) calls ‘laissez-faire racism’ and others institutionalized racism.
Consider the effect of race in a vignette study where experimental subjects were told to shoot armed targets and not to shoot unarmed targets. Participants made the correct decision to shoot an armed target more quickly if the target was African American than if he was white (Correll, et al. 2002: 1325). The magnitude of this racial bias in shooting decisions varied with perception of cultural stereotypes but not with personal racial prejudice. In fact, the study revealed equivalent levels of shooting bias in African American and in white participants. This finding underscores the potentially far-reaching consequences of statistical discrimination and cultural stereotypes that reside below the level of conscious racial prejudice. As the authors argue, ethnicity can influence the decision to shoot because cultural traits associated with African Americans, namely ‘violent’ or ‘dangerous’, act as a schema to influence perceptions of an ambiguously threatening target. African Americans are unlikely to be racially prejudiced against their own ethnic group, but they are exposed, as is everybody, to dominant cultural stereotypes.
Implicit bias and statistical discrimination theory are limited, however, in their tendency to adopt either a psychologically reductionist or a rational choice model of decision making, both of which neglect the social meaning of perceptions – context matters. Indeed, although Goffman’s (1963b) concept of stigma was originally advanced at the individual level, its contextual or group forms are equally compelling. A contextual stance was taken some time ago by Werthman and Piliavin (1967), who argued that the police divide up the territories they patrol into readily understandable, and racially tinged, categories. The result is a process of what they called ecological contamination, whereby all persons encountered in ‘bad’ neighbourhoods are viewed as possessing the moral liability of the neighbourhood itself. This process has general implications in so far as citizens themselves impute the character of disreputability to neighbourhoods containing stigmatized minorities, immigrants, and the ‘rabble class’ (Irwin 1985). Such stigmatization appears to be an enduring mechanism going back at least to Charles Booth’s lower-class London with its ‘loafers and semi-criminals’.
The social structure of everyday life in public places is tied to race and class, reinforcing the production of disrepute (Hagan 1994:150). As Stinchcombe (1963) argued, access to private space is structured such that disorder by the disadvantaged consists of doing many things in public that would be (and are) legitimate in private (e.g., drinking, hanging out). That is, privileged status enhances private access, reducing everyday exposure to public disorder. The resulting social structure of public spaces reinforces the stereotype that disorder is a problem mainly in poor, African-American communities. This stereotype feeds racial stigma and the creation of a durable spoiled identity for the modern American ghetto (Wacquant 1993).
Recent ethnographic work underscores the symbolic importance attached to the intersection of race and disorder. In a study of a white working-class Chicago neighbourhood, Kefalas (2003) sought to understand the fastidiousness with which residents kept up their property and why they seemed to be obsessed with physical signs of order. She found that homeowners fretted about ‘the last Garden’ and the threats that disorder were thought to bring on the neighbourhood (Kefalas 2003: 11, 14, 62, 74). No act of vandalism was too minor; no unkempt yard was too trivial to escape notice. Kefalas argued that residents did not care so much about disorder per se, but were threatened by the idea of the urban underclass, blacks’ encroachment in particular. In Chicago, many residents of the South-west Side perceive a long westward march of decline preceded by visual cues of disorder. Thus in many ways the residents of Kefalas’s Beltway had a ‘broken windows’ theory in mind, but one with a decidedly black face (2003: 43).
None of this is to assert that average city dwellers are somehow irrational or merely ignorant. Visual cues of disorder can be disturbing even to those who study it for a living.4 But predictions can become self-confirming when stigma and spoiled identity intercede, leading to actions that increase the statistical association between race and the observable behaviour. For example, if affluent residents use a neighbourhood’s racial composition as a gauge for the level or seriousness of disorder, unconsciously or not, they may disinvest in predominately minority areas or move out; such actions would tend to increase physical disorder in those neighbourhoods. In this way implicit bias leads to reinforcing mechanisms that perpetuate the connection of race to disorder, therefore helping to explain the dynamics reinforcing racial segregation (Loury 2002). The general framework of this argument is portrayed heuristically in Figure XI.
As with Booth’s depiction of drunken Irish women noted earlier, racial and ethnic categories subject to hierarchical classification are historically variable. In many US cities circa World War I, for example, it was not blacks but Irish and Italian immigrants that constituted the dangerous and disorderly class (Laub and Sampson 1995). National context matters too. In present day London it may well be that social distinctions within the white working class are just as pernicious as black-white distinctions in the USA. Watt (2006) reports that social distinctions in the borough of Camden have a powerful spatial component in council estates where concentrations of the white working class/poor are centre stage in narratives of disorder and decline (e.g., homelessness, ‘filth’, public alcoholism) that are implicated in the creation of social boundaries and status differentials by neighbourhood.
Surprisingly little research bears on the general theoretical framing in Figure XI of the social and cultural structures that imbue disorder as a problem. In this paper I therefore exploit an original set of recently collected data to examine the proposition that perceptions of neighbourhood disorder are socially mediated, surprisingly stable, and that they are contextually shaped by much more than actual levels of disorder. I begin with a brief description of the analytic approach I used earlier with Steve Raudenbush, which motivates the current follow-up.5 We first reasoned, according to the logic of the broken windows theory, that if the perception of disorder is governed by actual, observed levels of disorder, we should find that residents in any given neighbourhood are largely in agreement on perceived disorder within that neighbourhood. Their views of disorder in the neighbourhood should not, for example, vary systematically and substantially by social class. Moreover, we should find few if any variations in perceived disorder between neighbourhoods that are linked to population characteristics once standardly defined and systematically observed disorder is accounted for.
Based on the alternative theory elucidated above and sketched in Figure XI, however, we hypothesized that the perception of disorder is socially grounded and that the racial, ethnic, and class composition of the neighbourhood would emerge as independent explanatory factors. We still expected residents’ perceptions to stem in important ways from visual cues such as trash, graffiti, abandoned cars and buildings, or the presence of loitering, drunken, or hostile adults. But our prediction should hold up even after we made adjustments for observed disorder, which we systematically measured using video cameras and trained observers. Because skin colour in particular is easily observed and carries powerful stereotypes, we expected that racial composition would loom large in people’s reporting of disorder; for some respondents, the racialized context of the neighbourhood might trump observed disorder, especially in segregated city like Chicago with a troubled history of racial strife in its recent past.
If race turned out to be a powerful indicator of disorder in people’s minds, we next asked: Does the perception of disorder reflect racial prejudice rather than statistical discrimination or racial stigma in the way that Glenn Loury describes it? If the perception of disorder is based on prejudice against African Americans, it is likely to affect the perceptions of whites, Latinos, and Asians more strongly than the perceptions of blacks. Thus in reporting disorder non-blacks might be expected to be more sensitive than blacks to increases in the concentration of black residents within a neighbourhood. But the notions of stigma and statistical discrimination suggest that if there is an association between racial composition and perceived disorder, it ought to be independent of the race or ethnicity of the observer (consider, for example, the black citizen who crosses the street walking late at night to avoid a group of approaching young black males).
Our data began with a comprehensive neighbourhood survey of Chicago residents living in some 500 block groups (small neighbourhoods) within Chicago, conducted in 1995. Census block groups average about 1,300 residents, compared to about 4,000 for the average census tract, and appear to well reflect the layout of pedestrian streets and patterns of social interaction. We interviewed over 3,500 randomly chosen adult residents within households selected according to a multi-stage probability sample. Perceptions of disorder were measured from six questions that asked about physical disorder (e.g., litter, graffiti, vacant housing) and about social disorder (e.g., public drinking, fighting, drug-dealing). Residents were asked: ‘Are these a big problem? Somewhat of a problem? Not a problem?’ From these questions we constructed scales of disorder at the level of the individual and block group. From the neighbourhood survey we also examined a large set of personal demographic and background characteristics that might anchor one’s perception of disorder, including age, sex, home ownership, and a composite measure of socio-economic status that took into account education, income, and occupational prestige.
From a separate study of community positional leaders and key informants we relied on over 1,000 interviews with representatives of the educational, political, business, religious, community organizational, and law enforcement domains. These leaders were asked detailed questions about their personal characteristics and social network ties in addition to their perceptions of disorder in the communities they represented or worked in (see Sampson and Raudenbush 2004:334). For the block groups in our study we also collected independent information from the US census that was likely to have bearing on perceptions of disorder: the proportion of families in poverty, population density, and the proportion black and Latino. Then from Chicago police records of violent offenses such as robbery, homicide, rape, and aggravated assault, we constructed the violent crime rate in each block group per 100,000 residents.
The last and I believe most innovate method was systematic social observation (SSO). By ‘systematic’ I mean that observation and recording were conducted according to explicit rules that allow others to replicate the observations. During the time that the community survey was conducted, our research team drove a vehicle very slowly down every street within the sample of almost 500 block groups – approximately 22,000 street segments. A pair of video recorders captured social activities and physical features on both sides, while trained observers simultaneously recorded observations in a log. Blocks were observed randomly and videotaped at any time from 7 a.m. to 7 p.m. A random subsample of these videotapes was then viewed and coded, again by trained observers. The key to the method was that the same trained raters assessed each block using the same instrument and under the same rules. Inter-rater agreement was very high as a result. As with the survey, we looked for signs of both physical and social disorder, but we measured a much more extensive array of observable items than was available in the survey questions. For example, we examined the separate contribution of the density of liquor stores and bars and the physical decay that can arise from institutional disinvestments, signaled by vacant or badly deteriorated housing, burned, boarded up, or abandoned commercial buildings, and deteriorated recreational facilities. Multiple scales accounting for measurement error in observed disorder were created and validated (Raudenbush and Sampson 1999).
Overall the results supported the neighbourhood stigma theory. Racial and ethnic immigrant concentration proved more powerful predictors of perceived disorder than did carefully observed disorder. We argued that this disparity was not necessarily ‘irrational’ or a reflection of simple prejudice. The rational basis of beliefs lies in a social history of urban America that links geographically isolated minority groups and recent immigrants with poverty, economic disinvestment, and visible signs of disorder. Skin colour is not only visual but psychologically salient in a society with a long history of slavery, segregation, and racial conflict. That the findings need not reflect pure racial prejudice was supported by another pattern in the data: blacks were no less likely than whites to be influenced by racial composition in predicting disorder. If racial prejudice were determining the association between percentage black and perceived disorder, this association ought to be much stronger for whites than for blacks: few would contend that blacks are as prone to anti-black racial prejudice as are whites. Although blacks perceive less disorder than whites living in the same block group, this tendency was not linked to the percentage of blacks living there (Sampson and Raudenbush 2004: 332).
Further, we replicated the main finding on the independent set of data collected from community leaders. We selected leaders who lived outside the community they worked in so as to remove as much as possible the role of inside information. Yet racial composition strongly predicted leaders’ evaluations of disorder controlling for observed disorder, just as for residents. Moreover, perceptions of disorder by the residents themselves independently predicted leaders’ perceptions (2004: 333). What mattered for the residents they served was thus what mattered for community leaders. Systematically observed disorder took a back seat in the process.6
The present analysis seeks to accomplish four major objectives. First, I briefly describe results of a replication of the Sampson and Raudenbush (2004) study with a three-pronged follow-up study in 2002 that provides longitudinal panel data on the prediction of disorder. The goal here is to assess whether the key findings on racial and immigrant context stand up.
Second, I examine the relationship between neighbourhood perceptions – the collective or group dynamics of social disorder – and individual perception. The question here moves beyond racial/ethnic composition vs. systematically observed disorder; rather, my question is how observed disorder and inter-subjective evaluations of disorder matter in the development of an individual’s perceptions. If the mechanism is largely statistical discrimination then percent black should remain a predictor and social perceptions should be less important than observed disorder. But if social meaning and cultural attributes are at work, it stands to reason that neighbourhood social perceptions – the inter-subjective variance component – should have a direct association with an individual’s assessment. To my knowledge this is the first time this hypothesis has been considered in a community study of residents, providing a unique look at the social context of both the structural and cultural sources of individual perceptions.
Third, I assess the stability of perceived disorder and concentrated poverty over time at the neighbourhood level. This question was left unaddressed by Sampson and Raudenbush (2004), but the extended theory I have outlined herein logically implies a ‘poverty trap’-like effect for neighbourhood social processes (Sampson and Morenoff 2006), in this case where social perceptions reinforce later disorder and potentially poverty absent an exogenous intervention. A pattern of neighbourhood level stability should thus be evident for both disorder and poverty.
The final and related goal is to probe the association of socially perceived disorder with later neighbourhood-level outcomes, especially the prediction of later poverty from prior collective evaluations of the neighbourhood. My argument is that neighbourhoods with high crime and signs of disorder are prone to developing reputations as ‘bad’ and thus to be avoided. When linked to the historical legacy in US cities of racial segregation and poverty fusing with structural patterns of disinvestment (Massey and Denton 1993), I argue that stigmatization sets in and a form of self-fulfilling prophecy (Loury 2002) or ‘Matthew Effect’ takes over, whereby residents acting on their perceptions of disorder undertake actions that have the effect of increasing that very disorder, ultimately leading me to predict out-migration and the increased concentration of poorer residents. It follows that if socially perceived disorder predicts later poverty better than observed disorder and perhaps even prior poverty, this is evidence in favour of the cultural aspects of the social reproduction of inequality and the stability of poverty traps.
To address these objectives I combine the data from Sampson and Raudenbush (2004) with a panel study of the same neighbourhoods (Sampson et al. 2007). Briefly stated, the first component of the panel is a multi-stage probability-based community survey (CS) based on interviews carried out in 2001–2002 by the Institute of Social Research based at the University of Michigan with a new sample of 3,105 Chicago residents living with the same neighbourhoods as the 1995 study. The core interview schedule from 1995 was repeated and augmented with additional questions. The design is thus a repeated cross-sectional survey that is well equipped to measure stability and change at the neighbourhood level and that is representative of Chicago. The second study is also a repeated cross section, but this time based on systematic social observation (SSO) of all block faces within the neighbourhoods within which the 1995 and 2002 community survey residents lived. Based on successful results from a pre-test and to save on costs we collected observer logs rather than videotapes on over 1,500 block groups or small neighbourhoods (about 700 census tracts). Analyses not shown replicated the high neighbourhood-level reliability of SSO measures of disorder (Sampson, et al. 2007). The third data source is an integration of the 2000 US Census with Chicago police records on crime from 2000 to 2003.
My analysis revealed a nearly complete substantive replication using independent data on all of the core measures that were repeated in the 2002 community survey and SSO . Once again, whites perceived more disorder than blacks, Latinos, and Asian/others even when they lived in the same neighbourhood. First generation immigrants also perceived more disorder than the second or third generation within neighbourhoods. Yet all groups perceive more disorder as per cent black increases across neighbourhoods, adjusting for observed levels of disorder. The effect of racial composition on perceived disorder was much greater than observed disorder. Adjusting for poverty and a host of both individual level and neighbourhood-level control variables, the magnitude of difference was over 3:1 for racial composition (percent black) versus SSO (observed) disorder in predicting an individual’s perception of disorder. It is implausible that measurement error in SSO could account for this very large disparity. Note too that over a dozen individual-level covariates were adjusted and thus cannot explain the disparity either. Although I included controls for the race and ethnicity of the observer there was no difference between blacks and whites in the prediction of perceived disorder from percent black, further supporting the notion of implicit bias associated with neighbourhood racial context. That is, there is no significant interaction of neighbourhood racial composition with respondent’s race in predicting individually perceived disorder.7
Another theoretical objective is to investigate the power of social perceptions and the prevailing cultural climate. The question I pose is deceptively simple: Do shared perceptions of disorder in one’s environment predict an individual’s perception of disorder many years later, adjusting for current levels of observed disorder, poverty, and an individual’s social position? In answering this question I again adjusted for dozens of individual-level characteristics, including fear of crime, age, race, sex, home ownership, social class, friendship ties, and even perceptions of cohesion in the neighbourhood (which are correlated with perception of disorder). Despite these stringent controls, there was a large effect of shared perceptions of disorder in 1995 – and not present levels of observed disorder – on an individual’s perceptions up to seven years later. I conducted a complete replication of this analysis controlling for systematically observed disorder in 1995 as well, obtaining identical results. That social perceptions have such persistent and strong predictive power adjusting for current and lagged observed levels of disorder is rather remarkable and suggests in a different way the sensitivity of humans to the evaluations of others.
Consider next the neighbourhood stability hypothesis. Figure XII shows that neighbourhood variations in perceived disorder are large and that neighbourhoods largely maintain their relative positions over time. We see ‘pockets’ of high disorder that are quite durable and apparently hard to shake off. This finding suggests that cultural and social aspects of neighbourhood disorder are coherent, durable, and of potential causal relevance. Concentrated poverty is also surprisingly stable through time. Over a period of 40 years, during a time of rapid social change, riots, crime rate swings, racial change, economic recessions, and gentrification at the end of the twentieth century, we see in Figure XIII a 0.78 correlation between poverty in 1960 and 2000. This finding extends that of Sampson and Morenoff (2006) two decades in each direction. In short, if I know the poverty level of a neighbourhood at one point in time it is possible to quite accurately forecast its relative outcome decades later. It is not obvious that this should be possible given that people move in and out – neighbourhoods are constantly in flux. How do we account for stability amidst change?
The last part of my argument set out expectations for how collectively perceived disorder might be implicated in the social reproduction of neighbourhood inequality. In the present data, the answer appears to be that shared perceptions of disorder are firmly implicated in how the character of a neighbourhood evolves over time. Consider the basic prediction of later poverty from current socially perceived disorder (Figure XIV). The correlation is high and positive (0.91), unusual in social science even for community level data. To be sure, we know that a number of factors predict perceived disorder that are also predictive of poverty, such as race, immigration, violence, and prior poverty but I explored confounding factors and the results held up. Setting details aside and focusing on the big picture, I found that systematically observed disorder had no independent association with later poverty. Second, shared perceptions were as strong if not stronger in predicting later poverty than population composition by race and even prior poverty itself. Indeed, social perceptions of disorder had a larger effect on later poverty levels than the inertial path dependence for which the indicator of prior poverty serves as a direct proxy.8
We can and should worry about proof of causality. Some might demand further controls, more complex statistical methods, or perhaps even a community-level experiment. Recognizing this concern I make no strong causal claims but would still argue that the data are consistent in pointing to a fairly strong and durable pattern. Namely, individual perceptions of disorder are strongly predicted by social context and perceptions of disorder among others. In turn the latter predicts a community’s later rate of poverty, suggesting that disorder – its meaning and inter-subjective assessments rather than observed disorder – matter for spatially linked inequality.
Another piece of evidence works to support this line of reasoning. Figure XV divides the sample of communities into those that are predominantly black and those that are predominantly white (75 per cent or more). There is a clear pattern showing that black neighbourhoods bear the brunt of the underlying dynamics – perceptions of disorder strongly foretell where a neighbourhood will end up in the stratification hierarchy. The prediction line is flatter and the correlation smaller for white areas. Is this just due to past poverty? Apparently not – the prediction slope for 2000 poverty adjusts for 1990 poverty but is still steeper in black than white communities and the correlation is approximately double in magnitude. Therefore the data suggest that shared perceptions are more consequential for a downward trajectory when they intersect with communities of colour.
This study underscores the relevance of social psychological mechanisms interacting with cultural and structural processes for understanding urban inequality, an area dominated in recent years by structuralist research. These need not be separate research enterprises. Neighbourhoods with high concentrations of minority and poor residents are stigmatized by historically correlated and structurally induced problems of crime and disorder. These historically resilient, psychologically prominent correlations have deep roots in American social stratification not likely to be overcome easily through short-term interventions. Because people act on their perceptions of disorder, the contributions of racial composition and concentrated poverty are tied reciprocally to the actions of observers. Social perceptions in turn form a meaningful aspect of the neighbourhood environment that influences individual perceptions and actions. By this account, then, the perceptual basis of action alternatives is highly contingent on social context.
If socially interpreted disorder is one the main signposts of population loss and the later deepening of poverty, the present schema might profitably enter the toolkit of stratification theorists and demographers of the city – urban planners even. Many US cities and neighbourhoods, especially in the North and Midwest, have not only lost population they have become poorer and more racially isolated in recent decades (Wilson 1987; Massey and Denton 1993). An important part of this racially selective decline in population and economic status appears to stem from stigmatizing perceptions of disorder that create a self-fulfilling structural prophecy whereby residents are likely to disinvest in or move away from black areas viewed as high risk for disorder, but in which whites are more sensitive in the first place and consequently more likely to move. In this way, shared perceptions of disorder may be one of the underappreciated causes of continued racial segregation in the USA and perhaps elsewhere. At the least, perceptions of disorder (defining deviancy up?) appear to matter for reasons that extend far beyond the agreed-upon presence of broken windows – even in small town England (Girling, Loader and Sparks 1999). Ironically, then, Charles Booth and the later disorder theorists may have been right all along, but for the wrong reasons.
The implications of my theoretical and empirical interrogation of disorder appear so far, I am afraid, to be adding up to a rather grim scenario. Rather than ending on a gloomy note, however, I wish to briefly explore what I believe is an optimistic forecast on the future of cities, one grounded in the logic of my argument coupled with current demographic trends. Namely, while the Wirthian narrative linking disorder with cities and the desire for control will probably always be present, the diversification of society through immigration, differential fertility by race/ethnicity, and increasing inter-group unions of all sorts (e.g., friendship, workplace, marriage) is emerging as a countervailing process breaking down distinctions that have for so long served as the flashpoint for classifications of disorder. As Gilroy (2000) has argued, the increasing mixing and heterogeneity within racial groups works ‘against race’ as it were, which can only help to further elide the bases for social distinctions that are inscribed in space. I wholly agree with Gilroy’s argument (2000: 11) that racial categories are becoming ever more uncertain and that ‘raciology’ is in a form of crisis. It is hard to imagine a more emphatic example of this than the recent election as President of the USA of a biracial man born in Hawaii with a black father from Kenya and a white mother from the heartland of the country. Change is surely afoot.
Moreover, many cities in the USA and internationally are becoming increasingly diverse and attracting an increasing share of the so called ‘creative class’ population (Florida 2002) that is drawn to, not repelled, by diversity, racial heterogeneity, and social difference. As Lloyd (2006) shows in his study of a former decaying but still disorderly neighbourhood (by conventional standards) in Chicago, ‘grit’ is the new glamour. The Wicker Park/Bucktown neighbourhood is hot, a ‘neo-bohemia’ teeming with artists and professionals seeking an ‘edge’, where ‘figurative representations of disorder are translated in the beacons of a new symbolic order’ (2006: 75). Might this describe areas of Chelsea as well and perhaps many others? More generally, cities are back, and I believe that part of the reason is the value that diversity holds and the effect of diversity on attracting those ‘against race’ and, I would add, against the homogenization now widely perceived in suburban sprawl (see also Sennett 1970).
I wish to add to this thesis the consequences of one of the biggest drivers of structural and cultural change – immigration. The relationship between immigration and crime is largely a negative one despite many expectations to the contrary and despite the effect of immigrant concentration on perceived disorder (Sampson 2008). And it turns out that the relationship between violence and indicators of diversity and immigration is if anything greater in high disorder environments. Although details are beyond the scope of this presentation, I submit as final evidence a data point that I believe is provocative in its implications. Figure XVI displays the association between language diversity (based on a Herfindahl index of 125 different languages) and rates of violence in neighbourhoods classified into high and low collectively perceived disorder (cut at the median) based on the data I have been using throughout. In both cases the relationship is negative but the slope is steeper in high disorder neighbourhoods. The data thus suggest that increasing diversity serves as a protective factor in high disorder and high poverty neighbourhoods (see also Sampson 2008: 32). In further analyses I controlled for poverty, density, and residential stability, and allowed the effect of diversity to vary geographically across the city. I found that increasing diversity and immigration have their greatest influence in what were formerly racially segregated areas and historically the areas of greatest social exclusion by the State (Wacquant, 1993), further evidence of the breaking down of racial hierarchies. It is mainly in the predominantly white areas that the diversity-crime association is attenuated, which by now should not be surprising. These are precisely the areas that historically have been most resistant to racial and ethnic change. Even here, however, immigration’s march is unlikely to remain at the border.
In sum, if diversity and immigration are implicated in re-energized cities and the decline in crime, this may help to work against the pernicious association that has been demonstrated between diversity and lower perceived trust (Putnam 2007) and, as we have seen here, the association of racial and immigrant composition with durable patterns of perceived disorder. The logic of my argument also implies that because the link between cues of disorder and perception is socially mediated, it is malleable and thus subject to potential change. We may be in a situation where cities, which not too far in the past were thought to be hotspots of disorder and decay, are thriving and increasingly valued despite or perhaps even because of, the presence of immigration and certain elements of disorder (Lloyd 2006; Sampson 2008).
In the UK the picture seems potentially bright as well. Despite an immigration panic in Europe that in many ways is now outstripping the USA (especially in continental Europe), here the crime rate continues to go down and London is, and has been for some time, one of the most diverse cities on earth. The concentration of poverty by racial and immigrant groups is also not as severe as in many American cities. Thus while the narrative of community decline and disorder will never go away, there is reason to suspect that the long term picture will improve.
The late sociologist Peter Blau (1977) once remarked that there is too much inequality but that there can never be too much heterogeneity. I agree. I want to further argue that if heterogeneity ultimately serves to reduce disparities of the city through the blurring of boundaries and the slow dissolving of categorical distinctions that to date have been so pervasive, perhaps theorists of urban disorder can help lead the way through efforts such as the present to elucidate what is in fact the social order of the increasingly diverse city, along with the irreducibly social bases for shared perceptions of disorder in the first place.
1Revised version of paper presented at the Annual British Journal of Sociology (BJS) Lecture, London School of Economics, 21 October 2008. This paper is drawn in part from collaborative work with Stephen Raudenbush and a book in progress, The Social Order of the City, being prepared for the University of Chicago Press. I thank John Laub for comments on an early draft.
2Benedict Anderson later wrote of the ‘imagined community’ and identity formation at a ‘nation’ level, but in broad theoretical terms the mechanisms are similar (Anderson 1991 ).
3Small towns in middle England with nary a crime in sight appear no less subject to the concerns of disorder and control than big city counterparts (Girling, Loader and Sparks 1999).
4I am no exception. One morning about three years ago I stepped outside my home only to notice a fresh swath of painted graffiti on the wall of a nearby apartment building. My first reaction was anger and, I admit, an almost instantaneous fear that my wife and I had bought in the ‘wrong’ neighbourhood, one about to decline. But realizing that I lived in a stable, well-off neighbourhood despite its dense urban character and proximity to a park and public transit line, I talked with authorities about the defacement. Others did too and the graffiti was cleaned up. Soon after, the same thing happened again and the process was repeated. After a cycle of 4–5 episodes, the problem went away. In this instance, ‘broken windows’ led to collective action not crime or decline (Sampson and Raudenbush, 1999: 638), but the experience none the less taught me a lesson in the subjective emotions that disorder can inspire.
5For the purposes of this paper I forego a detailed description of data and method. See Sampson and Raudenbush (2004) for details. For other studies that inform my investigation see Taylor (2001), Quillian and Pager (2001), Ross and Mirowsky (1999), Harcourt (2001), and Skogan (1990).
6We undertook a series of other tests to assess the robustness of results (see Sampson and Raudenbush 2004:333–334; 339–340). Some might be concerned about temporal variations in disorder over the course of the day, for example, but we measured and accounted for time of day in all analyses. Not surprisingly, physical disorder turned out to be highly stable over time. Moreover, even if some social disorder emerged at night (e.g., a bar fight), our results would be overturned only if such disorder occurred in a large number of areas where other social disorder was not present during the day. From all we know on the basis of prior research and our knowledge of Chicago, such a reversal of pattern is highly unlikely. Spatial mismatch is another concern. Suppose that a resident, when responding to questions about disorder, recalled an area different from the block group where he or she lived, or maybe an area just outside the boundaries of the block group. Our measures, however, reflected the block group as a whole rather than geographic differences within or outside block groups in the degree of observable disorder. The research design also produced a representative survey sample of individuals within block groups and idiosyncratic definitions were by definition averaged across multiple residents, a procedure unlikely to produce a systematic influence of racial composition on between-area (block group) variations. Furthermore, we validated empirically the robustness of key results to alternative ecological definitions that allowed for the possibility that residents really think about disorder at the level of much larger communities. Results were none the less largely insensitive to variations in the size of neighbourhood unit, yielding similar patterns all the way up to community areas which in Chicago average almost 40,000 residents. The strong similarity of findings across units of analysis suggests that spatial or ecological mismatch does not account for the contributions of race/ethnic and social composition. Finally, by analysing and replicating the key results with elite leaders who lived outside the community, we demonstrated that the results cannot simply be attributed to ‘local knowledge’ among local residents. In short, the large magnitude of the contributions of racial and ethnic structure, especially in models where physical disorder is measured virtually without error, undermines the credibility of counter claims that our results are artifacts of the unreliable or invalid measurement of observed disorder.
7The original study showed an ethnic/immigrant interaction, in that Latinos were significantly more likely to perceive disorder as percent black increased compared to blacks and whites (Sampson and Raudenbush 2004:335). The magnitude of this pattern diminished over the period of study, however, which makes sense if we consider the argument of Loury (2002) that it takes time for immigrant attitudes to adjust.
8For example, in a key model allowing direct comparison, the standardized coefficient for 1990 poverty in predicting 2000 poverty was 0.26 (p < 0.01) whereas the coefficients for shared perceptions of disorder, percent black, and percent immigrants were 0.33, 0.42, and 0.33 (all significant at p < 0.01). Observed disorder was not significant (beta = 0.03).