Sie sind auf Seite 1von 20

Neighborhood Disadvantage, Disorder, and Health Author(s): Catherine E.

Ross and John Mirowsky Reviewed work(s): Source: Journal of Health and Social Behavior, Vol. 42, No. 3 (Sep., 2001), pp. 258-276 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/3090214 . Accessed: 27/03/2012 16:55
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of Health and Social Behavior.

http://www.jstor.org

Neighborhood Disadvantage, Disorder, and Health*


CATHERINE E. ROSS JOHN MIROWSKY
Ohio State University

Journalof Health and Social Behavior2001, Vol 42 (September):258-276

We examine the question of whether living in a disadvantaged neighborhood damages health, over and above the impact ofpersonal socioeconomic characteristics. We hypothesize that (1) health correlates negatively with neighborhood disadvantage adjustingfor personal disadvantage, and that (2) neighborhood disorder mediates the association, (3) partly because disorder and thefear associated with it discourage walking and (4) partly because they directly impair health. Data are from the 1995 Community,Crime, and Health survey, a probability sample of 2,482 adults in Illinois, with linked information about the respondent's census tract. Wefind that residents of disadvantaged neighborhoods have worse health (worse self-reported health and physicalfunctioning and more chronic conditions) than residents of more advantaged neighborhoods. The association is mediated entirely by perceived neighborhood disorder and the resulting fear It is not mediated by limitation of outdoor physical activity. The daily stress associated with living in a neighborhood where danger, trouble, crime and incivility are common apparently damages health. We call for a bio-demography of stress that links chronic exposure to threatening conditions faced by disadvantaged individuals in disadvantaged neighborhoods with physiological responses that may impair health. Does neighborhooddisadvantageimpairthe physical health of residents? If so, how? Residentialareascharacterized by high ratesof poverty or single-mother households and by low rates of college educationand home ownership add collective or environmentaldisadvantages to the personal ones of residents (Masseyl996; Wilson 1996). Living in a disadvantagedneighborhoodmay damage health, over and above the impact of personal socioeconomic characteristicsthat limit residential options (Jones and Duncan 1995; LeClere, Rogers, and Peters 1997; Robert 1998, 1999). This study tests the hypothesis that neighborhood disorderaccounts for the association of neighborhood disadvantage with poor health. Signs of neighborhood disorder take two forms: physical disorder such as abandoned buildings, noise, graffiti, vandalism, filth, and disrepair; and social disordersuch as crime, loitering, public drinking or drug use, conflicts, and indifference. Many individuals may find life under such conditions threatening and forbidding. Neighborhood disorder may discourage healthful outdoor activities such as walking. Beyond that, it may stimulate frequentterrorand chronic foreboding,repeatedly flooding the body with adrenalhormones that directlyunderminehealth.

*This research was supportedby a grant from the National Institute of Mental Health, "Community, Crime, and Health" (ROI MH51558) to Catherine NEIGHBORHOOD DISADVANTAGE, E. Ross (p..) and Chester Britt (co p.i.). We thank DISORDER, AND HEALTH Shana Pribesh and the JHSB reviewers and editor for their help. Direct all correspondence to

CatherineRoss, Department Disadvantage andDisorder of Sociology, 300 Neighborhood


Bricker Hall, The Ohio State University, 190 N.

The lack of economic and social resources Oval Mall, Columbus, OH 43210. E-mail: Ross.131@osu.edu. in disadvantaged neighborhoods predisposes
258

DISORDER, AND HEALTH NEIGHBORHOOD DISADVANTAGE, them to physical and social disorder. Disadvantage is a condition or circumstance unfavorableto success. To the extent that disadvantaged individuals are concentrated in geographically defined areas, disadvantage becomes characteristicof their neighborhoods (Massey 1996). Neighborhooddisadvantageis indicated by things such as the prevalence of poverty and of mother-only households. Massey (1996) argues that concentration of poverty creates alienation of many types because it erodes public order.Wilson (1996) also considers neighborhoods to be a fundamental cause of many social problems,but he argues that family structureis a critical aspect of neighborhoods, too. According to him, poverty itself did not demolish social order when poor families typically had two parents, because two-parentfamilies provide orderand stability even in the presence of poverty. On the opposite end of the scale, neighborhood advantagessuch as a high prevalence of college education and home ownership can favor success. College educatedresidents in a neighborhoodprovidecollective humancapital and positive role models (Wilson 1996). They contributeto an environmentin which many adultshave skills, jobs, opportunities,and connections outside the neighborhood.Likewise, in neighborhoods with high levels of home ownership many residents have wealth that inheresin the neighborhood.Because the value of each home depends on the quality of the neighborhood,home-owning residents have a substantialpersonal interest in preservingand improvingit. Informal social control weakens in disadvantaged neighborhoods, where many residents are poor and poorly educated, few own homes, and single-parentfamilies are common (Sampson, Raudenbush,and Earls 1997). The combinationof few economic resources, little human capital, and weak control generates a threateningand disorderedenvironmentcharand crime. acterized by incivility Neighborhood disadvantage might provoke disorder because of limited opportunity (Wilson 1987, 1996), lack of social integration and cohesion (Sampsonand Groves 1989), and normative climates conducive to disorderly behavior (Brewster, Billy, and Grady 1993; Elliot et al. 1996; Jencks and Mayer 1990). Poor and isolated neighborhoodsmay have few jobs, so thatresidentsperceive little opportunity for employment(Wilson 1996). Youngpeo-

259

ple who see little chance to succeed may be less likely to stay in school and more likely to engage in illegitimate activities, thus increasing the level of disorderin the neighborhood. Disadvantaged neighborhoods may also lack the informal social ties that bind neighbors together and help maintain social order (Sampson and Groves 1989). Disadvantaged neighborhoodsalso have fewer resources like good schools, parks, and services, which may indicate to residents that mainstream society has abandoned them; residents, in response, may abandon conventional, orderly behavior (LaGrangeet al. 1992; Taylorand Hale 1986; Wilson 1987). In neighborhoodswith greater social and economic resources, residents have the assets, abilities and self-interestsconducive to orderand safety Evidence RegardingNeighborhood Disadvantage and Health Multilevel analyses suggest that neighborhood disadvantagemay negatively affect residents'health,over and above the effects of personal disadvantages. A multilevel analysis adjusts for individual socioeconomic and demographic status when correlating health with aggregate neighborhood conditions. Multilevel research often finds poorer health associated with indications of neighborhood disadvantage net of personal attributes, althoughthe estimatedeffects of neighborhood characteristicstend to be small and inconsistent comparedto those of individualattributes. Robert(1998) correlatedthreehealthmeasures with the proportionof households in the census tract receiving public assistance, the percent of families with incomes of less than $30,000, and the percent of adults unemployed. An index of the three aspects of neighborhood economic disadvantage correlates significantly with a resident'snumberof diagnosed chronic conditions, and the percentage of families receiving public assistance correlates significantly with self-reported health, even with adjustmentfor personal education and household income and assets. However, the correlationswith physical functioningwere not statistically significant after adjusting for individual-levelsocioeconomic status (Robert 1998). Two multilevel studies examine the associationof heartdisease with neighborhood disadvantage. One finds that neighborhood

260

JOURNALOF HEALTHAND SOCIALBEHAVIOR graffiti,people hangingout on the streets,public drinking, run-down and abandonedbuildings, drug use, danger,troublewith neighbors, and other incivilities associated with a breakdown of social control (Geis and Ross 1998; LaGrange,Ferraro,and Supancic 1992; Lewis and Maxfield 1980; Lewis and Salem 1986; Skogan 1986, 1990). These signs indicate a potential for harm, even to residentswho have not been victimized. The signs of disordersuggest that many neighbors do not respect other people or their property,that agents of social control are unable or unwilling to cope with local problems,and that the neighborhoodhas been abandonedand its residentsmust fend for themselves (Skogan 1990; Taylor and Hale 1986). A threatening and noxious environment characterizedby crime, harassment, danger, and incivility may underminephysical health for several reasons. One reason is that such an environment may discourage the physical activity needed to maintain health. Another reason is that it may stimulate psychophysiological responses that directly undermine health. We discuss these below. Disorder,fear and walking. The stress of living in a neighborhood with high levels of disordermay damage health indirectly if fear decreasesoutdoorphysical activity.Whenpeople live with disorder in their neighborhood, they often feel afraid (LaGrange et al. 1992; Lewis and Salem 1986; Taylorand Shumaker 1990). People who fear being robbed,attacked, or physically injured and who are afraid to leave the house are unlikely to walk for pleasure, exercise, or transportation (Ross 1993). At the other extreme, quiet, safe, clean neighborhoods may invite outdoor activity. Where there is no apparentdanger, people may take walks for pleasure and exercise, and commutersmay walk to the trainstationor to work. The disorderfound in disadvantaged neighborhoods may impair health because it discourages outdoor physical activity. Lack of physical activity is second only to smoking in its negative health consequences. Walking is the most common physical activity, reported by about 20 percent of Americans (U. S. Bureau of the Census 1985). Walking is typically a moderate, non-aerobic exercise. Compared to the inactivity of a sedentary lifestyle, any physical activity, aerobic or not, increases life expectancy (Berkman and Breslow 1983). Walking reduces the risk of

socioeconomic deprivationis associated with heartdisease in the United States, adjustingfor individualsocioeconomic status(Diez-Roux et al. 1997); another finds neighborhoodpublic assistance and poverty rates, low median income, and female headship rates predict women's heart disease mortality (LeClere, Rogers, and Peters 1998). Some multilevel studies look at the association of all-cause mortality with measures of neighborhooddisadvantage.One finds that low median income in American neighborhoods predicts greater mortality among men, but not women, adjusting for individual-level education and household income relative to needs (LeClere, Rogers, and Peters 1997). A U.S. urban study finds neighborhood poverty associated with mortality among younger Americans, but not among those over age 55, with adjustmentfor individual level socioeconomic status (Waitzman and Smith 1998). British studies find ward-level deprivation associated with poor respiratoryfunctioning, heart symptoms, and poor self-reported health (Jones and Duncan 1995), and with chronic illness (Sloggett and Joshi 1998), but not with allcause mortality (Sloggett and Joshi 1994), adjusting for individual-level socioeconomic status. The overallpatternof findings suggests that neighborhooddisadvantagemay correlatewith some collective or environmentalfactor that impairs health net of the impact of personal disadvantagesthat lead individuals to live in such neighborhoods.However,the tenuousand inconsistent correlations suggest that neighborhood disadvantagedoes not directly impair health, but rather predisposes neighborhoods to the harmfulconditions. The mechanismsby which neighborhood disadvantagemay affect health are largely unknown.This study examines the possibility that neighborhooddisorder is one link. NeighborhoodDisorder and Health Daily exposure to a threatening, noxious environmentmay erode health.Neighborhoods with high levels of disorder present residents with observablesigns and cues that social control is weak (Skogan 1986; 1990; Skogan and Maxfield 1981; Taylor and Hale 1986; Taylor and Shumaker1990). In these neighborhoods, residents reportnoise, litter,crime, vandalism,

DISORDER, AND HEALTH NEIGHBORHOOD DISADVANTAGE, cardiovascularproblems, colon cancer, back pain, osteoporosis, obesity, high blood pressure, constipation, varicose veins, adult onset diabetes, and it improves subjective health (Caspersen et al. 1992; Duncan, Gordon, and Scott 1991; Leon et al. 1987; Magnus, Matroos, and Strackee 1979; Paffenbargeret al. 1986; Ross and Wu 1995; Segovia, Bartlett, and Edwards1989; U.S. PreventiveTaskForce 1989). Disorder,fear, and stress. Disorder and the fear it engenders may impair health directly, apart from discouraging healthful outdoor activity such as walking. Biomedical research shows that a threateningenvironmentcan produce physiological responses that may impair health in several ways: by creating symptoms experienced as illness, by increasing susceptibility to pathogens and pathological conditions, and by accelerating the degradationof critical physiological systems (Fremont and Bird 2000; McEwen 2000; Taylor,Repetti, and Seeman 1997). According to current biomedical theory, threats stimulate the fight or flight response, which has two phases (Memler, Cohen, and Wood 1996; Thibodeau and Patton 1997). In the initial alarm stage, sympathetic nerve fibers stimulatethe adrenalmedulla to release the hormone epinephrineand the neurotransmitter norepinephrine. This increases heart rate; blood pressure; and respiration rate; dilates the blood vessels of the heart, lungs, and skeletal muscles; constricts the vessels of the digestive tract; and releases glucose from the liver into the blood. The activation of the sympathetic nerves also stimulates the sweat glands and suppresses the salivary glands. These physiological responses may be experienced as illness, particularlyif the response becomes frequent or generalized and thus seemingly detached from specific stimuli. Individuals exposed to chronic psychosocial strains develop heightened reactivity (Pike et al. 1997; Pruessneret al. 1997). They enterthe stage of alarm more readily, quickly, and intensely,and take longer to recoverfrom it. In the follow-up resistance stage of the "fight or flight" response, an endocrine gland in the braincalled the anteriorpituitaryreleases adrenocorticotropic stimulating hormone (ACTH), which stimulatesthe release of cortisone and cortisol (hydrocortisone) from the adrenalcortex. These hormones suppresspain, They inflammation, allergy, and immunityv.

261

raise blood glucose levels by decreasing glucose metabolism and acceleratingthe conversion of fats and proteins (includingmuscle) to glucose. While cortisone and cortisol relieve some symptoms, they appear to create others-notably fatigue and sleep disturbance (Glaser and Kiecolt-Glaser 1998; Brunner 1997). Excess cortisone and cortisol produces central obesity, hypertension, and hyperglycemia (Thibodeauand Patton 1997). The hormonesreleasedin both phases of the stress response may reduceresistanceto infections and cancers (Glaser et al. 1999; Herbert and Cohen 1993; Irwin et al. 1997). Chronic stress appears to inhibit innate, nonspecific immunity,in which naturalkiller lymphocytes detect and destroycells that show signs of viral infection or other abnormalities. It also appearsto suppressthe productionof antibodies and T-lymphocytes keyed to detect and destroy specific invaders. As a result, social and psychological stress undermines the immune system's ability to suppress an infection before it producesunpleasantor incapacitating symptoms. Psychosocial stress correlates positively with the likelihood of developing symptoms after exposure to cold viruses, and with antibody concentrationsthat suggest widespreadviral proliferation(Cohen, Tyrrell, and Smith 1991). It also may reactivatelatent viral infections such as varicella zoster, herpes simplex, and Epstein Barr (Cohen et al. 1999; Cohen and Herbert 1996; Glaser and KiecoltGlaser 1998; Irwin et al. 1998). Chronicexposure to a threateningenvironmentmay undermine the body's naturaldefenses. Stresshormonesalso can exacerbateor even instigate chronic health conditions (Fremont and Bird 2000; McEwen 2000). Stressors can precipitate heart problems such as irregular beat (arrhythmia)and platelet clotting which can produce inadequatebloodflow (ischemia), perhaps resulting in death of heart tissue (infarction).The alarmphase of the "fight-orflight" response thus may stimulate a heart attack. It also can damage the lining of coronary arteries, instigating the formation of plaque that eventually occludes the arteries. Chronic stress increases "allostatic load," which "refers to the price the body pays for being forced to adaptto adverse psychosocial or physical situations" (McEwen 2000:110). For example, the cortisol released in the resisacceleratesthe progrestance phase apparently sive thickening and hardening of arteries

262

JOURNALOF HEALTHAND SOCIAL BEHAVIOR age, are disadvantagedthemselves; thus, it is possible that geographically-defined places have no effect independentof the demographic characteristics of their residents (Jencksand Mayer 1990; Jones and Duncan 1995; Robert 1998; Slogget and Joshi 1994). Disadvantaged individuals who lack social and economic resources often live in disadvantagedneighborhoods with high levels of disorder. Individualswith low incomes and little education, those who are unemployed or employed in low status jobs, minorities, and unmarried people have worse health, on average, than those with high incomes and education, employedpersons, whites, and marriedpeople (Link and Phelan 1995; Mirowsky,Ross, and Reynolds 2000; Ross, Mirowsky, and Goldsteen 1990). We adjust for the personal disadvantagesthat lead individuals to live in disadvantaged neighborhoods and that also underminehealth. We use multilevel data in which the unit of of analysis is the individualand characteristics the respondent's neighborhood are linked to individualsurvey data. Indicatorsof objective neighborhood disadvantageare derived from an exogenous data source-the Census SummaryTape file 3 from the 1990 Census of Population and Housing-which provides independent assessments of disadvantage in the contextual units (Blalock 1985). We, like others, use the census tractas the best approximation of the neighborhood (South and Crowder1997; Tienda 1991). About two-thirdsof the respondentsin the statewide random sample used here reside in the same tract as at least one other member of the sample. That creates the possibility of a regression residual correlated within tracts.

through the buildup of fatty plaque (atherosclerosis) and perhapsalso the buildup of calcium salts and scar tissue (arteriosclerosis) throughoutthe body, includingarteriessupplying the heart, brain, and other vital organs. Atherosclerosis interacting with high blood pressure (anotherresult of chronic stress) can develop into coronary heart disease (McEwen 2000). In sum, repeated exposure to threatening conditions may impairhealth. People exposed to neighborhooddisorderseem likely to experience more frequentand intense activationof the stress response, with possible consequences for their health. Hypotheses We hypothesize that (1) health correlates negatively with neighborhood disadvantage adjusting for personal disadvantage,and that (2) neighborhooddisordermediatesthe association, (3) partly because disorderand the fear associated with it discourage walking and (4) partly because they directly impair health.' Figure 1 shows the processes by which neighborhood disadvantagemay influence health. METHODS MultilevelData and Model In orderto test the hypothesisthatthe neighborhood in which a person lives affects physical well-being, we distinguish individualfrom neighborhood disadvantage. Disadvantaged neighborhoodscontain persons who, on aver-

AffectsHealth Disadvantage FIGURE1. Theoretical Modelof the Processesby whichNeighborhood


Sociodemographic Characteristics
/

Fear

R~~ ~~+1Disorder

+Y

Health

Neighborhood Disadvantage

Walking

AND HEALTH DISORDER, DISADVANTAGE, NEIGHBORHOOD Thus, when estimatingthe effects of neighborhood context on health, ordinaryleast squares errors techniquesmay producebiased standard (Blalock 1985; Bryk and Raudenbush 1992; DiPrete and Forristal 1994; Goldstein 1995).2 We use the multilevel statisticalmodeling program MLn to addressthe possibility (Rasbash et al. 1995), by distinguishing a tract-level residual from the individual-level one (Goldstein 1995; Bryk and Raudenbush1992). MLn uses iterativegeneralizedleast squaresto estimate the slopes and two components of residual variance((u2): residualvarianceat the individuallevel (u 2) and residualvariancethat is constantacross individualswithin a tractbut randomacross tracts(ou2). The multilevel model can be summarizedas follows:

263

select and contacta respondent,and up to 10 to complete the interviewonce contactwas made. Interviews were completed with 73.1 percent of the eligible persons who were contacted. The final sample has 2,482 respondents, ranging in age from 18 to 92, with an average age of 45. The sample'sdemographicprofile is similar to the population (Illinois residents over age 18), but, like most surveys, is somewhat more educated, well-to-do, white, and female. The median family income in the sample is $40,000, compared to $38,664 in the population; the mean education level in the sample is 13.8, comparedto 12.7 in the population; the percentagewhite is 84.0, compared to 80.5; and the percentagemale is 41.0 percent, compared to 48.5 percent. Regression models with income, education,race, and sex as independentvariables inherentlyadjust for these differences between the sample and the y = a + I3'xf+ V11 population it represents(Winship and Radbill where t indexes tractsand i indexes individuals 1994).3 within tracts,and
Vd = Ut + sta

Conceptsand Measurement Physical health is the dependentvariable.It is measuredas an index of self-reportedhealth, physical functioning,and lack of chronic conditions. Self-reported health is the respondent's subjective assessment of his or her general health as very poor (coded 1), poor (2), satisfactory (3), good (4), or very good (5). Physical functioning is measured by asking respondents how much difficulty they have with (1) going up and down stairs;(2) kneeling or stooping; (3) lifting or carryingobjects less than 10 pounds, like a bag of groceries; (4) preparing meals, cleaning house, or doing other household work; (5) shopping or getting aroundtown; (6) seeing, even with glasses; and (7) hearing, even with a hearing aid (Nagi 1976; McDowell and Newell 1987). The response categories are "a great deal of difficulty" (coded 0), "some difficulty" (coded 1), and "no difficulty"(coded 2). Exploratoryfactor analysis indicatesthatthe seven items form a single factor. Our measure of physical functioning sums these seven items. Low scores indicate physical impairment or disability; physical funchigh scores indicateunimpaired tioning. To assess chronic medical conditions, respondentswere asked abouta series of health problems:"Thenext set of questions ask about conditions that some people have been diag-

The model assumes that:ut are constantacross individuals within tracts but random across tractsand normallydistributedwith a mean of 0 and variance of uF2; 8dt. are random across tracts and individuals within tracts and normany distributedwith a mean of 0 and vanance of rE2; and 8td and ut are not correlated. Sample Community,Crime and Health (CCH) is a 1995 survey of a probabilitysample of Illinois households linked to census tract information. Respondents were interviewed by telephone. They were selected using a pre-screenedrandom-digit dialing method that increases the rate of contacting eligible numbers (i.e., decreases the rate of contacting business and non-workingnumbers)and decreases standard errors compared to the standard MitofskyWaksbergmethod while producing a sample with the same demographicprofile (Lund and Wright 1994). The survey was limited to English-speaking adults. The adult (18 or older) with the most recent birthday was selected as respondent, which is an efficient method to randomlyselect a respondentin the household. Up to 10 call-backs were made to

264

JOURNALOF HEALTHAND SOCIALBEHAVIOR equivalent to an average increase of ten percentage points across the four components. The index ranges from advantagedneighborhoods in which many adults have college degrees and own their homes and few households are poor or female-headed,on the low neighborhoodsin which end, to disadvantaged few adults have college degrees, many rent ratherthan own their homes, and many households are poor and female-headed,on the high end. Both theoreticaland empiricalobservations warrant measuringneighborhooddisadvantage by reference to poor and mother-onlyhouseholds and the absence of home ownershipand adults with college degrees. The prevalenceof poverty is the core measure of economic disadvantagein neighborhoods(Jargowsky1997; Massey 1996; Wilson 1987). Home ownership is an indicator of wealth that inheres in the neighborhood;it indicates neighborhoodeconomic advantage and collective commitment to the neighborhood.The prevalenceof mother-only households captures social disadvantage which is correlatedwith economic disadvantage but potentially makes an independent contributionto disorderbecause single parents may be less able to control their children and single-parent neighbors may be less able to watch each other's children (Wilson 1996). Mother-childfamilies, moreover,are the poorest of any family type (McLanahanand Booth 1989), and poor households tend to be common where mother-headed households are common. Adults with college degrees indicate collective humancapital;they providepositive role models of adultswith skills, jobs, and connections outside the neighborhood,and their presence signifies to teens that opportunities exist if one stays in school and out of jail (Wilson 1996). Thus, a lack of well-educated adults in the neighborhood,too, may make an independent contribution to disorder. Poor households, mother-only households, owner occupied houses, and adults with college degrees load on a single factorat .84, .89, -.55, and -.60, respectively,and the alphareliability of the index is .61. Objective neighborhood disadvantage is measured using information from the Summary Tape File 3 of the 1990 Census (Bureau of the Census 1992). We matched tract-level data to the geographic location of each respondent. Seven hundred and eleven cases were missing tract level data. For the

nosed as having. Have you ever been diagnosed or told by a doctor that you have: (1) heart disease, (2) high blood pressure,(3) lung disease like emphysema or lung cancer, (4) breast cancer, (5) any other type of cancer,(6) diabetes, (7) arthritisor rheumatism,(8) osteoporosis (brittlebones), (9) allergies or asthma, or (10) ulcers, ulcerativecolitis, or otherdigestive problems."These items representthe most common health problems that threaten survival, function, and quality of life (Kochanek and Hudson 1995). The absence of each condition is scored 1 and the presence of each condition is scored 0. The resulting index counts the instances where chronic health problems are absent. Self-reportedhealth,physical functioning, and the absence of chronic conditions load on a single factor at .82, .83, and .79, respectively.The final index is a result of standardizing each measure and taking the mean score of the three; it ranges from poor to good health (alpha = .75). The index of objective neighborhooddisadvantage adds the prevalenceof poverty and of mother-only households and subtracts the prevalence of home ownership and college educated residents in the respondent'sCensus tract.The prevalenceof poverty is the percentage of households with incomes below the federal poverty threshold. The prevalence of mother-only households is the percentage of female-headed households with children. The prevalence of college educated adults is the percentage of adults over the age of 24 with college degrees. The prevalenceof home ownership is the percentage of housing units that are owner occupied. Among the tracts in the sample, the percent of households in poverty ranges from 0 to 83, with a mean of 10. The percent of mother-only households ranges from 0 to 67, with a mean of 6. The percent of adults with college degrees ranges from 0 to 51, with a mean of 14. The percent of homes that are owned ranges from .3 to 97, with a mean of 64. The index of neighborhooddisadvantage divides each of the four percentages by ten, adds poor and mother-onlyhouseholds, subtractshome ownershipand college educated residents, and divides by four. Thus, a unit increase in the scale is equivalent to an increase of ten percentagepoints in each of the components: the prevalence of poor households, of mother-only households, of nonowner occupied units, and of adults without a college degree. Alternatively,a unit increase is

NEIGHBORHOOD DISADVANTAGE, DISORDER, AND HEALTH majority of these cases (511 of 711) we were able to use dataat the zip code level instead of the tract-level. Zip codes are somewhat larger units, but they are the next best approximation of a neighborhood.In order to determine the effect of substituting zip code for tract we added to the preliminaryregressions a dummy variabledistinguishingthe two. Its coefficient was never significant and was droppedin the subsequentanalyses presentedhere. Perceived neighborhood disorder is measured with the Ross-Mirowsky neighborhood disorder scale (1999). Neighborhood disorder refers to conditions and activities, both major and minor,criminaland non-criminal,thatresidents perceive to be signs of the breakdownof social order. The index measures physical signs of disorder such as graffiti, vandalism, noise, and abandoned buildings, and social signs such as crime, people hanging out on the street, and people drinking or using drugs. It also includes reverse-codedsigns of neighborhood order,such as safety, people taking care of theirhouses and apartments or watchingout for each other. Disorder is perceived and reportedby residents of the neighborhood.In orderto describe his or her neighborhood, a person must be aware of it and perceive it, and two people in the same neighborhood might describe it somewhat differently. Nonetheless, both are describing a certain place, and correlations between respondents' reports of disorder in

265

their neighborhood and independent assessments by researchers are moderate to high (Perkins and Taylor 1996). Table 1 lists the items in the index, along with loadings on the first rotated factor from an exploratoryfactor analysis.The perceivedneighborhooddisorder scale ranges from orderon the low end to disorder on the high end of the continuum, and has an alphareliabilityof .916. Walking is measuredas the numberof days walked per week. Respondents were asked, "How often do you take a walk?"Open-ended responses are coded into number of days walked per week. Fear is measured as a mean-score index of the numberof days in the last week that someone (1) fearedbeing robbed,attacked,or physically injured; (2) worried that their home would be broken into; and (3) felt afraid to leave the house (alpha = .69). Individual sociodemographic disadvantage may create apparent contextual effects that actually are compositional, due to the disadvantaged sociodemographiccharacteristicsof the individuals who live in disadvantaged neighborhoods(Jencks and Mayer 1990). Our models include the following individual sociodemographicattributes.Age is scored in numberof years. Sex is a binary scored 1 for males and 0 for females. Race is a binary scored 1 for whites and 0 for non-whites. Education is scored in number of years. Household income is coded in thousands of

TABLE 1. Items in Ross-Mirowsky Perceived Neighborhood Disorder Scale (1999), Theoretical Distinction between Social and Physical Disorder and Order and Empirical Associations
FactorLoadingsa Physical Disorder and Orderb There is a lot of graffiti in my neighborhood .804 .747 My neighborhoodis noisy .832 Vandalismis common in my neighborhood .723 There are lot of abandonedbuildings in my neighborhood .603 My neighborhoodis clean .558 People in my neighborhoodtake good care of their houses and apartments Social Disorder and Orderb .754 There are too many people hanging aroundon the streetsnear my home .847 There is a lot of crime in my neighborhood .817 There is too much druguse in my neighborhood .754 There is too much alcohol use in my neighborhood .438 I'm always having troublewith my neighbors .442 In my neighborhood,people watch out for each other .608 My neighborhoodis safe .916 Alpha reliability 1.811 Mean aFactor loadings from structurematrix,factor 1, oblimin rotation. bAll items are scored so that a high score indicates disorder.Disorderitems are scored stronglydisagree (1), disagree (2), agree (3), and stronglyagree (4). Orderitems are scored stronglyagree (1), agree (2), disagree (3), and strongly disagree (4).

266

JOURNALOF HEALTHAND SOCIALBEHAVIOR less likely to have high levels of physical functioning, less likely to feel healthy, and more likely to have chronichealth problems.4 Individual socioeconomic disadvantage is also associated with worse health. The well educated report better physical functioning, better self-reported health, and fewer health problems than the poorly educated. People with higher incomes have fewer physical limitations, fewer chronic conditions, and feel healthiermore than those with lower incomes. The health returns to income diminish as income rises, as indicated by the fact that logged income was more significant thanother specifications. Each dollarhas a largerpositive influence on health at lower levels of income than at higher.Employedpersons score higher on the health index than those who are not employed,but occupationalstatusis not significant. Men report better physical functioning, better perceived health, and fewer chronic healthproblemsthanwomen, people with children score higher on the health index than those without children,divorcedpersons score lower on the health index than do marriedpersons, and older persons score lower on the health index than youngerpersons. Standardizedcoefficients (not shown) indicate that an individual's own socioeconomic status has a larger effect on health than does the neighborhoodin which one lives. The beta associated with neighborhooddisadvantageis -.05 comparedwith .11 for householdincome, .12 for education,and .14 for employment.The health effects of one's own education,employment, and household income are more than double that of neighborhooddisadvantage. What are the mechanisms by which neighborhood disadvantage influences health? When neighborhood disorder is added in model 3, the coefficient associatedwith neighborhooddisadvantageis reducedby 57 percent from model 2 and becomes insignificant at conventional levels (-.056 - (-.024) /-.056 = .57). Neighborhood disorderhas a significant negative association with health. People who reportthat there is a lot of crime, graffiti, vandalism, trouble, drug use, dirt, and danger in their neighborhoodhave more chronic health problems, worse self-reported health, and worse physical functioning than people in neighborhoodstypified by orderand safety. Some readers may wonder whether the effect of neighborhood disorder on health results from threator from squalor.The physi-

dollars and logged in the regression analyses. Employment status contrastspersons employed full- or part-time for pay (coded 1) with persons who are not employed (0). Occupational status is measured by the National Opinion Research Center/General Social Survey Socioeconomic Index score (Nakao, Hodge, and Treas 1990). The non-employed are assigned the mean and employment status is included in the regressions. Marital status is coded as a series of binaries:divorced,single, widowed,and married,with marriedpersons as the comparison category in the regression analyses. Number of children is the number underthe age of 18 in the home. Because urban areas concentratedisadvantaged neighborhoods, apparentneighborhood effects could be due to city residence.Thus, we also include urban residence, measured as a dummy variable which contrasts living in the city of Chicago (coded 1) with residence in suburbs, small cities, small towns, and rural areas (coded 0). RESULTS Table 2 shows the prediction of physical health in five steps. Model 1 shows the total effect of neighborhooddisadvantage.Model 2 adds adjustment for individual sociodemographiccharacteristics. Models 3, 4, and 5 add the series of mediators:neighborhooddisorder, fear, and walking. Residents of disadvantagedneighborhoods report significantly worse health than those in more advantagedplaces, as shown in model 1 of Table 2. However, this significant negative coefficient may be biased by the omission of individual disadvantage that correlates with worse health and with living in a disadvantaged neighborhood.With adjustmentfor individual sociodemographicattributesin model 2, the coefficient associated with neighborhood disadvantageis reduced by almost 57 percent (-.129 - (-.056)/-.129 = .566). More than half of the apparent neighborhoodeffect was due to the sociodemographic characteristicsof residents. Nonetheless, a significant contextual effect remains.With adjustmentfor individual characteristicsin model 2, living in a disadvantaged neighborhood is still significantly associated with worse health. Comparedwith residents of more advantagedneighborhoods, residents of disadvantagedneighborhoodsare

DISORDER, ANDHEALTH NEIGHBORHOOD DISADVANTAGE,

267

TABLE 2. Health (Physical functioning, self-reported health, and absence of chronic conditions) Regressed on Neighborhood Disadvantage (Model 1), Sociodemographic Characteristics (Model 2), Perceived Neighborhood Disorder (Model 3), Fear (Model 4), and Walking (Model 5) (CCH, Illinois, 1995; N = 2,252; metric coefficients with standard errors in parentheses are shown)
Model 1 Contextual Neighborhood disadvantage Sociodemographic Education Household incomea Employmentstatus (employed= 1) Occupational Status Divorcedb Singleb Widowedb Numberof children Age Race
(white=

Model 2 -.056* (.027) .043*** (.008) .104*** (.020) .307*** (.045) .001 (.002) -.128* (.057) -.067 (.055) .033 (.074) .031+ (.017) -.020*** (.002) .064 (.056) .076* (.038) -.052 (.059)

Model 3 -.024 (.027) .038*** (.008) .093*** (.020) .312*** (.045) .001 (.002) -.119* (.056) -.056 (.055) .040 (.074) .033+ (.017) -.020*** (.002) .037 (.056) .079* (.038) .023 (.061)

Model 4 -.021 (.027) .038*** (.008) .091*** (.020) .311*** (.045) .001 (.002) -.118* (.056) -.053 (.055) .041 (.073) .036* (.017) -.020*** (.002) .035 (.056) .076* (.038) .030 (.060)

Model 5 -.025 (.027) .036*** (.008) .089*** (.020) .321*** (.045) .001 (.002) -.127* (.056) -.053 (.055) .027 (.073) .037* (.017) -.020*** (.002) .032 (.056) .070* (.037) .007 (.061)

-.129*** (.027)

1)

Sex
(male= 1) Urban residence

Mediators Disorder Fear Walking Constant


R2 Yd

-.226*** (.043)

-.203*** (.046) -.054* (.023)

.242 .204 .458 -.175 .284 .281 .273 .011 .707*** .708*** .956*** .715*** Level Indivkdual (.023) (.023) (.023) ErrorVariance(cre2) (.031) .004 .005 .006 .027+ NeighborhoodLevel (.010) (.010) ErrorVariance(ru2) (.010) (.015) +p <.10; *p <.05; **p <.01; ***p <.001 (2-tailed tests) Notes: Neighborhooddisadvantage= (% households below poverty line + % households female-headedwith children - % adults with college degree - % households own home) / 40. a logged b comparedto married

-.203*** (.046) -.050* (.023) .021*** (.005) .148 .288 .700*** (.023) .006 (.010)

cal signs of disordermay indicate exposure to toxins, carcinogens, pathogens, and physical stressorssuch as noise or filth. In a subsidiary analysis, we separatedthe disorder index into physical and social components,following the distinction in Table 1. We calculated a new index representing the difference between scores on the two subscales,physical (P) minus

social (S), and added it to the regression of health on the sum of the two subscales. (Scores on each subscale are divided by the numberof items, to make the two comparable.)Thus, the regression has the following form: H = bo + bI(P + S) + b2(P- S) + b3X+ etc. The coefficient b1representsthe averageeffect of the two subscales. It is estimatedas -.215 (t = -4.760,

268

JOURNALOF HEALTHAND SOCIALBEHAVIOR hoods do not walk less. Instead,there is some evidence that they walk more. Residents of neighborhoods with a lot of disorder do not walk less than those in neighborhoods with higher levels of order, although people who feel afraiddo walk less. Health is damagedby residence in a disadvantagedneighborhoodbecause disadvantaged neighborhoodshave high levels of disorder.In these neighborhoodsresidents face a noxious, which threatening,and dangerousenvironment negativelyimpactshealth directlyand indirectly because it is associated with fear.Thus, the first part of our theoreticalexplanationis supported. It appears as if the small additional reductionof the disadvantagecoefficient with the introductionof fear is simply due to the attenuationinherent in two levels of indirect effects. On the other hand,the hypothesis that residents of disadvantaged and disordered neighborhoodswalk less is not supported.In orderto furtherunderstandwhy, we disaggregate the neighborhooddisadvantageindex into its componentsand predictwalking in Table4. Overall,neighborhooddisadvantagehas little effect on walking because economic disadvantage and educational disadvantage have opposite effects. The percent of college educatedresidentsin the neighborhoodis positively associatedwith walking, which supportsour hypothesis. Stated the other way, the absence of college educated residents in a neighborhood, which indicates disadvantage,is associated with a decreasedlikelihood of walking on the part of residents.However,both aspects of economic disadvantage, poverty and the absence of home ownership, are positively associatedwith walking, which contradictsour theory.When both are included in the Table4, neither is significant, and home ownership is more significantthanpoverty(with adjustment for neighborhood education), so we show its effect in the table. Residents of neighborhoods where most people own their homes walk significantly less than those in neighborhoods with a high percentageof renters.When neighborhoodpoverty is substitutedfor home ownership in the neighborhood,poverty has a significant positive effect on walking (b = .024, Seb = .0 12, p = .05). Residents of poor neighborhoodswhere few residentsown theirhomes walk more, not less, which contradictsour theory.5 Walking is typically done on neighborhood streets, so we emphasized it in our theory of the mechanisms by which neighborhoods

p < .00 1). The coefficient b2representsthe difference in the effects of the two subscales. It is estimatedas .165 (t = 3.016,p < .010). The significant positive b2 coefficient implies that the effect of physical disorder on health is less negative (-.215 + .165 = -.050) and that the effect of social disorder is more negative (-.215 - .165 = -.380). Social disorderapparently has a largernegative effect on healththan does physical disorder.This suggests that the operative factor is exposure to threat, rather than to squalor. Adding adjustmentfor frequency of fear in the past week in model 4 of Table 2 further reduces the coefficient associated with neighborhood disadvantage.Comparisonof models 2 and 4 shows that over sixty percent of the impact of neighborhood disadvantage on health is accounted for by disorder and fear (-.056 - (-.021)/-.060 = .625). People who are afraidof being robbed,attackedor injuredand are afraidof leaving their house reportsignificantly worse health than people who are not afraid.The adjustmentfor frequencyof fear in the past week also reducesthe coefficient associated with neighborhooddisorderby about 9 percent,althoughdisorderis still significant at the .001 level. (-.226 - (-.203)/-.226 = .899). Walking is significantly associated with good health, but its introductionin model 5 does not explain any of the effect neighborhood disadvantageon health. In fact, the coefficient associated with neighborhood disadvantageactuallyincreases somewhat.Nor does walking explain any of the associationbetween neighborhooddisorderand health, which also increases a little with the introductionof walking. Why do disorder and fear explain much of the effect of neighborhood disadvantage on poor health, but walking does not? In orderto furtherunderstandthese effects we show the predictions of the three potential mediatorsin Table 3. As hypothesized, in disadvantaged neighborhoods, residents report higher levels of disorder than they do in more advantaged neighborhoods.Furthermore, residents of disadvantagedneighborhoodsreportsignificantly more fear, and this is due in part to the higher levels of disorder in disadvantagedneighborhoods. Both disadvantageand disorder influence fear, and the introduction of disorder explains about 48 percent of the association between disadvantage and fear (.124 .065/.124 .476). Contrary to expectations, though, residents of disadvantagedneighbor-

NEIGHBORHOOD DISADVANTAGE, DISORDER, AND HEALTH

269

TABLE 3. Neighborhood Disorder, Fear, and Walking Regressed on Neighborhood Disadvantage, Adjusting for Sociodemographic Characteristics; Disorder Added in Model 2 Predicting Fear, and Disorder and Fear Added in Model 2 Predicting Walking (CCH, Illinois, 1995; N = 2,252; metric coefficients with standard errors in parentheses are shown)
Disorder 1 Neighborhood disadvantage Disorder Fear Education .140*** (.013) .124*** (.026)
-

Fear 2 .065** (.026) .422*** (.042) 1 .172 (.100)

Walking 2 .193+ (.112) .019 (.189) -.188* (.093) .101** (.034) .072 (.082) -.471** (.184) -.005 (.007) .402+ (.230) -.021 (.224) .634* (.300) -.050 (.070) -.022** (.006) .135 (.230) .273+ (.154) 1.068*** (.252) 2.734 .035 11.759*** (.383) .154 (.167)

-.024*** -.012 -.002 .103** (.004) (.008) (.008) (.034) Household -.049*** -.063*** -.043* .084 incomea (.009) (.019) (.019) (.082) Employmentstatus .024 -.003 -.013 -.471** (employed = 1) (.021) (.042) (.041) (.184) -.001 Occupational .001 .002 -.005 Status (.001) (.002) (.002) (.007) Divorcedb .037 .040 .023 .397+ (.026) (.053) (.052) (.230) Singleb .044+ .088+ .068 -.035 (.026) (.051) (.050) (.224) Widowedb .028 .036 .024 .626* (.034) (.069) (.068) (.301) Numberof .011 .038** .034** -.058 children (.008) (.016) (.016) (.070) Age .000 .001 .001 -.022*** (.001) (.001) (.001) (.006) Race -.122*** -.068 -.026 .149 (white= 1) (.026) (.054) (.053) (.229) Sex .014 -.046 -.053 .283+ (male = 1) (.018) (.035) (.035) (.154) Urban .337*** .309*** 1.023*** .160* residence (.030) (.065) (.064) (.245) Constant 1.836 -.068 -.698 2.745 .201 .079 .122 .034 YP e R2bs IndividualLevel .151*** .595*** .574*** 11.779*** ErrorVariance(Ure2) (.005) (.020) (.019) (.384) NeighborhoodLevel .005* .048*** .039*** .157 ErrorVariance(ru 2) (.002) (.012) (.011) (.166) +p<.10; *p <.05; **p <.01; ***p <.001 (2-tailed tests) Notes: Neighborhooddisadvantage= (% households below poverty line + % households female-headedwith children - % adults with college degree - % households that own home) / 40. a logged b comparedto married

might affect health, but since walking does not link neighborhood disadvantage to poor health, we next look at other health behaviors that might. Neighborhood normative climates could influence smoking, heavy drinking, or exercising (Duncan and Moon 1999; Ross 2000). Smoking is a binary variablefor which .... . currentsmokersare coded 1; heavy drnking is a binary variablefor which persons who drink 4 or more drinksa day are coded 1; and exercise is coded in numberof days per week that the respondentreportsstrenuousexercise such as running,basketball,aerobics, tennis, swimming, biking, and so on. Smoking, drinking,

and exercising do not mediate the effects of neighborhood disadvantage or disorder on health. The coefficients associated with disadvantage and disordereach change by less than 1 percent with the introductionof smoking, drinking and exercising. For instance the drinkieng and ecia sing. eins teu in e, coefficient associated with Fisord disorder equa3; intrincr ion magtion5o -.20 nitude to -.206 with the introductionof exercising; decreases to -.201 with the introduction of smoking;and remainsat -.203 with the introductionof drinking.Smoking, exercising, and heavy drinkingexplain almost none of the

270

JOURNALOF HEALTHAND SOCIALBEHAVIOR Despite the fact that our data reveal an association of neighborhooddisadvantage with poorer health, we also find that about half of the apparentcontextualcorrelationis due to individual disadvantage.Adjustment for individual-level sex, age, race, employment status, income, education, occupation,maritalstatus, and childrenin the home explainedmore than half of the association between health and
neighborhood disadvantage. Persons who are

TABLE4. WalkingRegressedon Neighborhood Home Ownershipand College Education, Adjusting for Sociodemographic Characteristics (CCH, Illinois,1995;N = 2,252;metriccoefficients with standard errors in are shown) parentheses
Contextual PercentCollege Educated .019*

(.009)** (.016)
Sociodemographic Education Household Household incomea Employmentstatus (employed - 1) Occupational Divorcedb
Singleb

~~~~~~.067
(.082) -.468** (.184) -.007 .363 (.230) -.103 .(.6272* (.300) -.033 (.070) ( 022** .132 (.224) .284+ (.153) (.246) 4.660
.040

.074* (035)

not employed, who have low household incomes and low levels of education report worse health than people who are employed and who have high household incomes and education. Older persons report worse health than younger, men report better health than women, and the presence of children in the home is associated with somewhat better health.An individual'sown education,income, and employmentstatus have a largerrelationship to health than does residence in a disadvantaged neighborhood.

Widowedb
Number of children Age Race
(white=

Furthermore, living near poor, disadvanA neightaged people is not in itself unhealthy. borhood is a distinctivedistrictor area and the people who live near one anotherin it. To the extent that disadvantagedindividualsare concentrated geographically, disadvantage becomes characteristic of theirneighborhoods. Health is not undermined directly by living near poor people, single mothers, or people who have not been to college. Rather, high rates of poverty and mother-onlyfamilies and low rates of college educationand home ownership compromise the ability of residents to create and maintain public order.The breakdown of social control and order in disadvantaged neighborhoods appears to form the major link to individualhealth.

1)

Sex (malen1) residence Constant Indiviual Level ErrorVariance(ae2) NeighborhoodLevel ErrorVariance(ru2) +p<.10; tests)
a b R2'bYse

11.841*** (.383) .016 (.151) *p <.05; **p < .01; ***p < .001 (2-tailed

logged to married compared

association between neighborhoodsand health NeighborhoodDisorder and Health (analyses availableupon request). Living in a poor neighborhoodwhere many families with children are headed by women, few residentsown their homes, and few adults DISCUSSION have college degrees appears to erode health because of neighborhooddisorderand the fear NeighborhoodDisadvantage, Individual associated with it. The stress associated with Disadvantage, and Health disorder,trouble, crime, danger, and the perception that social orderhas broken down are Individuals who live in disadvantaged associatedwith worse health.Looked at anothneighborhoodsappearto sufferworse healthas er way, in neighborhoods with few poor or a result of the environmentin which they live. mother-only households and many college

HEALTH DISORDER, AND DISADVANTAGE, NEIGHBORHOOD educated adults and home owners, residents generally reportthat the neighborhoodis safe, clean, and quiet, that there is little vandalism, graffiti or crime, and thatpeople maintaintheir houses and apartmentsand watch out for each other. These perceptions of social order are associated with better health for all residents, whatevertheir personal status. Causationand Selection

271

Could the association between neighborhood disadvantageand poor health result from unhealthy individuals moving into disadvantaged neighborhoods? While possible, it seems unlikely. The models adjust for stable traits such as race, age, and educationthat influence health and constrain residential options and choices. They also adjust for possible consequences of poor health that might conceivably limit housing options to disadvantagedneighborhoods: income, employment, and marital status. There seems little reason to think that unhealthypersons would move into disordered neighborhoodsmore frequentlythan healthier individuals with similar demographic and socieconmicprofles.Howverit m be neighborhoodsdo have higher levels of fear: they are more afraid of being assaulted and eteen poo tha some ofites. so that and fearefctsvnetfelt by injured,more afraid to go out on the streets, . Alith we and more afraid of having their home broken oulerasic thoeah ar peared im may have overestimatedfear'sdamagingeffect into. In partthis fear reflects high levels of dison health somewhat,it is precisely this associ- order in disadvantaged neighborhoods. ation for which the best biological evidence Residents of economically disadvantaged exists. Fear stimulates the release of epineph- neighborhoods walk more than residents of rine and norepinephrine, followed by release of more affluent neighborhoods, despite their cortisone and cortisol, which, when chronic, fear. Residents of neighborhoodsin which a high increases blood pressure, serum cholesterol, serum glucose, atherosclerosis, and conse- proportionof adults have college degrees also quently the risk of diabetes, stroke, heart dis- walk more. Thus, two aspects of neighborhood socioeconomic status-economic status and ease, and so on (McEwen 2000). education-have opposite effects. Possibly neighborhoodswherethe college-educatedlive have a cultureof walking in which people walk is Not the Link Walking for exercise, pleasure, and transportation. Neighborhood economic disadvantage is People may see others walking and adopt the contagion effect associated with the likelihood of walking, but lifestyle themselves-a not in the expected way. People who live in (Crane 1991)-and residents of advantaged poor neighborhoods where few people own neighborhoods are generally not afraid of their homes are more likely to walk than those being victimized on the streets. Because neighborhood disadvantage does in more economically advantagedplaces. This could be due to the structureof neighborhoods not decrease walking, one part of our explanawhere most people rent, where higher density tion for the association between neighborhood encourageswalking. The fact that people who disadvantageand health is not supported.

live in the city of Chicago walk more than residents of suburbs, small cities, small towns, and ruralareas gives credence to the idea that density facilitates walking for transportation. Since we have adjustedfor household income, it doesn't seem likely that this is simply due to the fact that poor people cannot afford cars, althoughit is possible (we have no information on car ownership).The propensityto walk in poor neighborhoods where few people own their homes could also be due to a normative climate where people hang out on the street and walk to visit others, go to the cornerstore, or just go down the street. People walk more in poor neighborhoods where most people rent ratherthan own their homes, despite the fact that living in a disadvantagedneighborhoodis associated with fear of being attackedand injuredand being afraid to leave the house. This effect of dangerous streetsdoes not overcomeothereffects of poor, rental neighborhoods. We had expected that residents of poor neighborhoods would walk less than the residents of affluent neighborhoods because they would be more afraid of being victimized. Residents of disadvantaged

272 Stress Physiology and Bio-Demography

JOURNALOF HEALTHAND SOCIALBEHAVIOR neighborhooddisorderand poor health underscores the need to move psychoendocrinology from laboratoriesand clinics into the world in which some individualslive, where threatand places. dangercharacterize Conclusion Living in a disadvantagedneighborhoodis associated with worse health, net of the health consequences of individual disadvantage. Residents of disadvantaged neighborhoods tend to feel less healthy and have more physical impairmentsand chronic health problems such as high blood pressure, asthma, and arthritis.The impact of living in a disadvantaged neighborhoodon physical well-being is mediated entirely by disorderin the neighborhood, which influences health both directly and indirectly,by way of fear.These neighborhoods present residents with observable signs that social controlhas brokendown:the streets are dirty and dangerous; buildings are rundown and abandoned;graffiti and vandalism are common; and people hang out on the streets, drinking, using drugs, and creating a sense of danger.Residents in these neighborhoods face a threateningand noxious environment characterizedby crime, incivility, and harassment, all of which are stressful. The chronic stress of exposure to disorderappears to impairhealth. NOTES 1. Our theory focuses on neighborhooddisorder as a chronicstressorthatdirectlyimpairs health, and its consequences for fear and outdoor physical activity that indirectly affect health,but there may be othermechanisms by which neighborhooddisadvantage affects health, some of which we addressin the discussion. 2. Since approximately two-thirdsof the tracts (766 of 1,169) contain only one respondent, few residentsare nested in tracts.Thus, clustering is not a problem,and in fact the OLS results are substantivelythe same. 3. Our sample is limited to residents of one state, Illinois. However,Illinois is fairlyrepresentativeof the nation as a whole because it is one of the states that has rural areas, small towns, small cities, and a major met-

The endocrinologists' concept of a physiological "fight or flight" response has been around for over half a century (Selye 1956). During that same period demographers,social epidemiologists, and sociologists have repeatedly documented the concentrationof health problems in disadvantaged populations. The idea of stress, and the word, has diffused and generalized throughoutpopular and scientific culture.While much was gained in the process, perhapssomething was also lost: the centrality of imminent danger. Generalization of the stress response to life's undesirable changes, highly supervisedjobs, maritalconflict, and so on should not blind us to its unconditioned operationon dangerousstreets. The correlation of neighborhood disorder with poor health points to a new direction for population research on health. It provides a previously unobserved clue to the explanation of persistentand widening socioeconomic disparities in health (Elo and Preston 1996). Poor and poorly educated individuals often live in disadvantagedneighborhoodswith high levels of disorder. It also underscores the need to move psychoendocrinologybeyond laboratory animals and clinical populations into large scale social and demographic surveys (Umberson, Williams, and Sharp 2000). Psychologists have done much to advance knowledge of the physiological responseslinking stress to poor health.As a practicalmatter, their samples typically consist of subjectssuch as students taking important examinations, married couples in conflict or under strain, family members caring for impaired elders, and the like. Their researchprovides valuable information on likely physiological mechanisms linking stress to poor health. We need a complimentary bio-demography that maps population distributions of bioassays indicating stress, that correlates the bioassays with social ones such as neighborhooddisadvantage and disorder,and that tests the hypothesis that chronic release of endogenous catecholamines and corticosteroids links threateningenvironments to poor health. Despite practical difficulties, advances in biomedical technology increasingly provide assays that use saliva, hair, urine, or blood samples feasible for use in representative household surveys (see Booth, Johnson, and Granger 1999). The association between

ANDHEALTH DISORDER, NEIGHBORHOOD DISADVANTAGE,

273

Journal of Health and Social Behavior 40:130-40. Brewster,KarinL., John O.G. Billy, and William R. Grady. 1993. "Social Context and Adolescent Behavior: The Impact of Community on the Transition to Sexual Activity." Social Forces 71:713-40. Brunner,Eric. 1997. "Stressand the Biology of Inequality." British Medical Journal 314:1472-75. Bryk, Anthony S. and Stephen W. Raudenbush. 1992. HierarchicalLinear Models: Applications and Data AnalysisMethods.Newbury Park,CA: Sage. Caspersen, Carl J., Bennie P.M. Bloemberg, Wim HM. Saris, Robert K. Merritt, and Daan Kromhout. 1992. "The Prevalence of Selected Physical Activities and their Relation with CoronaryHeart Disease Risk Factorsin Elderly Men: The Zutphen Study, 1985." American Journal of Epidemiology 133:1078-92. Cohen, Francis, MargaretE. Kemeny, Kathleen A. Kearney,LeonardS. Zegans, John M. Neuhaus, and MarcusA. Conant. 1999. "PersistentStress as a Predictor of Genital Herpes Recurrence." Archivesof InternalMedicine 159:2430-6. Cohen, S. and T.B. Herbert. 1996. "Health Psychology: Psychological Factorsand Physical Disease from the Perspective of Human Psychoneuroimmunology."Annual Review of Psychology 47:113-42. Cohen, S., D.A. Tyrrell and A.P. Smith. 1991. "Psychological Stress and Susceptibility to the Common Cold." New England Journal of Medicine 325:606-12. Crane, Jonathan.1991. "Effects of Neighborhoods on Dropping out of School and Teenage Childbearing." Pp. 299-320 in The Urban Underclass, edited by ChristopherJencks and Paul E. Peterson. Washington, DC: The Brookings Institute. REFERENCES Diez-Roux, A.V, EJ. Nieto, C. Muntaner, H.A. Tyroler,G.W, Comstock. 1997. "Neighborhood Berkman,Lisa F. and Lester Breslow. 1983. Health Environmentsand Coronary Heart Disease: a and Waysof Living: TheAlameda CountyStudy. Multilevel Analysis." American Journal of New York:Oxford University Press. Epidemiology 146:48-63. Berlin, Jesse A. and GrahamA. Colditz. 1990. " A Meta-Analysis of Physical Activity in the DiPrete, Thomas A. and Jerry D. Forristal. 1994. "Multilevel Models: Methods and Substance." Prevention of Coronary Heart Disease." Annual Review of Sociology 20:331-57. AmericanJournal of Epidemiology132:612-28. Blair, Steven N., Nancy N. Goodyear, Larry W Duncan,JohnJ., Neil F. Gordon,and ChrisB. Scott. 1991. "WomenWalkingfor Health and Fitness." Gibbons, Kenneth H. Cooper. 1984. "Physical Journal of the American Medical Association Fitness and Incidence of Hypertension in 266:3295-99. Healthy Normotensive Men and Women." Journal of the American Medical Association Duncan, Craig, Kelvyn Jones, and GrahamMoon. 1999. "Smoking and Deprivation: Are there 252(4):487-90. Neighbourhood Effects?" Social Science and Blalock, Hubert M. 1985. "Contextual-Effects Medicine 48:497-505. Models: Theoretical and Methodological Issues."AnnualReviewof Sociology 10:353-72. Elliott, Delbert S., William Julius Wilson, David Huizinga, Robert J. Sampson, Amanda Elliott, Booth, Alan, David R. Johnson, and Douglas A. and Bruce Rankin. 1996. "The Effects of Granger. 1999. "Testosterone and Men's Neighborhood Disadvantage on Adolescent Depression: The Role of Social Behavior."

ropolitan area: Chicago. Large cities everywhere in the United States have more disadvantaged neighborhoodswith higher levels of disorder than do non-urban areas, but, while large cities have many things in common, they have many differences. Cities in the Midwest and Northeast may have more disadvantaged neighborhoods than those studied by Fischer (1982) in Northern California;for example, half of the total rise in concentrated poverty duringthe seventies was accounted for by just two cities: New Yorkand Chicago (Wilson 1991). 4. Dichotomizingneighborhoodsinto very disadvantaged (one standard deviation above the mean and higher) versus othersindicates there is not a cut-off above which neighborhood disadvantage affects health. The dummy variable'scoefficient is smaller and less significant than the continuousform. 5. The percent of female-headed households has no significant effect on walking even when entered alone, so we pruned it from the model. The case of walking is the only one in which differentcomponentsof disadvantage have opposite effects on an outcome; all four aspects of disadvantagecorrelate negatively with health and positively with disorderand fear. Because of this, and because the four componentsare highly correlated with each other, when they are entered separatelyin the models predicting health, disorder, and fear, multicolinearity results.

274

JOURNALOF HEALTHAND SOCIALBEHAVIOR


Adrenal Cortical Function, and Sympathetic Tone." Annals of BehavioralMedicine 19:83-90. Jargowsky,Paul A. 1997. Poverty and Place. New York:Russell Sage. Jencks, Christopher and Susan E. Mayer. 1990. "The social consequences of growing up in a poor neighborhood."P. 111-84 in Inner-City Povertyin the UnitedStates, edited by Laurence E. Lynn and Michael G.H. McGeary. Washington,DC: NationalAcademy Press. Jones, Kelvyn and Craig Duncan. 1995. "Individualsand their Ecologies: Analysing the Geography of Chronic Illness within a Health and Multilevel Modeling Framework." Place 1:27-30. Ku, Leighton, Freya L. Sonenstein, and Joseph H. Pleck. 1993. "Neighborhood, Family,and Work: Influences on the Premarital Behaviors of Adolescent Males."Social Forces 72: 479-503. LaGrange, Randy L., Kenneth F. Ferraro, and Michael Supancic. 1992. "Perceived Risk and Fear of Crime: Role of Social and Physical Incivilities."Journal of Research in Crime and Delinquency29:311-334. LeClere, Felicia B., Richard G. Rogers, and Kimberley D. Peters. 1997. "Ethnicity and Mortality in the United States: Individual and Community Correlates" Social Forces 76:169-98. LeClere, Felicia B., Richard G. Rogers, and Kimberley D. Peters. 1998. "Neighborhood Social Context and Racial Differences in Women's Heart Disease Mortality"Journal of Health and Social Behavior 39:91-107. Leon, ArthurS., JohnConnett,David R. Jacobs,and RainerRauramaa.1987. "Leisure-Time Physical Activity Levels and Risk of Coronary Heart Disease and Death: The Multiple Risk Factor Intervention Trial." Journal of the American MedicalAssociation 258:2388 -95. Lewis, Dan A. and Michael G. Maxfield. 1980. "Fearin the Neighborhoods:An Investigationof the Impact of Crime."Journal of Research in Crimeand Delinquency 17:160-189. Lewis, Dan A. and Greta Salem. 1986. Fear of Crime:Incivility and the Productionof a Social Problem.New Brunswick,NJ: Transaction. Link, Bruce G. and Jo C. Phelan. 1995. "Social Conditions as FundamentalCauses of Disease." Journal of Health and Social Behavior. Extra Issue: 80-94. Lund, Laura and William E. Wright. 1994. "Mitofsky-Waksberg vs. Screened Random Digit Dial: Report on a Comparison of the Sample Characteristics of Two RDD Survey Designs." Presented at the Center for Disease Control's 1th Annual BRFSS Conference, Atlanta,GA, June. Magnus, K., A. Matroos, and J. Strackee. 1979. "Walking,Cycling, or Gardening,with or without Seasonal Interruptions, in Relation to Acute

Development" Journal of Researchin Crimeand Delinquency 33:389-426. Elo, Ima T. and Samuel H. Preston. 1996. "EducationalDifferentials in Mortality:United States, 1979-85." Social Science and Medicine 42:47-57. Fiscella, K. and P.Franks.1997. "Povertyor Income Inequality as Predictor of Mortality: Longitudinal Cohort Study." British Medical Journal: 314:1724-27. Fischer, Claude S. 1982. ToDwell among Friends. Chicago: University of Chicago Press. Fremont,Allen M. and Chloe E. Bird. 2000. "Social and Psychological Factors, Physiological Processes, and Physical Health."Pp. 334-352 in The Handbook of Medical Sociology, 5th ed, edited by Chloe E. Bird,PeterConrad,andAllen M. Fremont. Upper Saddle River, NJ: Prentice Hall. Geis, Karlyn J. and Catherine E. Ross. 1998. "A New Look at Urban Alienation: The Effect of Disorder on Perceived Neighborhood Powerlessness." Social Psychology Quarterly61: 232-46. Glaser,Ronald and Janice K. Kielcolt-Glaser.1998. "Stress-Associated Immune Modulation: Relevance to Viral Infections and Chronic Fatigue Syndrome." American Journal of Medicine 105:35s -42s. Glaser, Ronald, Bruce Rabin, Margaret Chesney, Sheldon Cohen, and Benjamin Natelson. 1999. "Stress-Induced Immunomodulation: Implications for Infectious Diseases?" Journal of the AmericanMedical Association 281:2268-71. Goldstein, Harvey. 1995. Multilevel Statistical Models. New York:HalsteadPress. Greene, William H. 1993. Econometric Analysis. New York: Macmillan. Guralnik, Jack M. and George A. Kaplan. 1989. "Predictors of Healthy Aging: Prospective Evidence from the Alameda County Study." AmericanJournal of Public Health 79:703 -08. Herbert, T.B. and S. Cohen. 1993. "Stress and Immunityin Humans:A Meta-analyticReview." PsychosomaticMedicine 55:364-79. House, James S., James M. Lepkowski, Ann M. Kinney,RichardP. Mero, Ronald C. Kessler,and A. Regula Herzog. 1994. "The Social Stratificationof Aging and Health."Journal of Health and Social Behavior 35:213-34. Irwin,Michael, CarolynCostlow,HeatherWilliams, Kamal HaydariArtin, ChristinaY. Chan, Diane L. Stinson, Myron J. Levin, Anthony R. Hayward, and Michael N. Oxman. 1998. "CellularImmunityto Varicella-Zoster Virus in Patientswith MajorDepression."TheJournal of InfectiousDiseases 178:S104-08. Irwin, Michael, Richard Hauger, Thomas L. Patterson,Shirley Semple, Michael Ziegler, and Igor Grant. 1997. "AlzheimerCaregiverStress: Basal Natural Killer Cell Activity, Pituitary-

DISORDER, ANDHEALTH NEIGHBORHOOD DISADVANTAGE, Coronary Events." American Journal of Epidemiology 110:724-33. Massey, Douglas S. 1996. "The Age of Extremes: Concentrated Affluence and Poverty in the Twenty-First Century" Demography 33: 395-412. McDowell, Ian, and Claire Newell. 1987. MeasuringHealth:A Guideto RatingScales and Questionnaires.Oxford,UK: Oxford University Press. McEwen, Bruce S. 2000. "Allostasisand Allostatic Load: Implications for Neuropsychopharmacology." Neuropsychopharmacology 22(2): 108-24. McLanahan,Saraand KarenBooth. 1989. "Motheronly families: Problems, Prospects, and Politics." Journal of Marriage and the Family 51:557-80. Memmler, Ruth L., Barbara Janson Cohen, and Dena Lin Wood. 1996. The Human Body in Health and Disease, 8th ed. New York: LippincottWilliams and Wilkins. Mirowsky, John, Catherine E. Ross, and John Reynolds. 2000. "Links between Social Status and Health Status."Pp. 47-67 in Handbook of Medical Sociology, edited by Chloe E. Bird, Peter Conrad, and Allen M. Fremont. Upper Saddle River,NJ: Prentice-Hall. Nagi, SaadZ. 1976. "AnEpidemiologyof Disability among Adults in the United States." Milbank MemorialFund Quarterly54:439-68. Nakao, Keikeo, RobertW Hodge, and JudithTreas. 1990. "On Revising Prestige Scores for all Survey Occupations." General Social Methodologicalreportno. 69, National Opinion ResearchCenter,Chicago, IL. R. S., Jr.,T. R. Hyde, A. L. Wing, and Paffenbarger, C.C. Hsieh. 1986. "PhysicalActivity,All-Cause Mortality, and Longevity of College Alumni." New EnglandJournal of Medicine 314:605-613. Perkins, Douglas D. and Ralph B. Taylor 1996. "Ecological Assessments of Community Disorder: Their Relationship to Fear of Crime and TheoreticalImplications" AmericanJournal of Community Psychology 24: 63-107. Pike, JenniferL., Tom L. Smith, RichardL. Hauger, Perry M. Nicassio, Thomas L. Patterson,John McLintick, Carolyn Costlow, and Michael R. Irwin. 1997. "Chronic Life Stress Alters Sympathetic, Neuroendocrine, and Immune Responsivityto an Acute Psychological Stressor in Humans." Psychosomatic Medicine 59:447-57. Pruessner,Jens C., Jens Gaab,Dirk H. Hellhammer, Doris Lintz, Nicole Schommer, and Clements Kirschbaum. 1997. "Increasing Correlations between Personality Traits and Cortisol Stress Responses Obtained by Data Aggregation." 22:615-25. Psychoneuroendicrinology Rasbash, Jon, M. Yang, Geoff Woodhouse, and Harvey Goldstein. 1995. MLn: Command

275

Reference Guide. London: Institute of Education. Robert, Stephanie A. 1998. "Community-level Socioeconomic Effects on Adult Health" Journal of Health and Social Behavior 39:18-37. Robert, Stephanie A. 1999. "Socioeconomic Position and Health: The Independent Contribution of Community Socioeconomic Context." Annual Review of Sociology 25: 489-516. Ross, CatherineE. 1993. "Fearof Victimizationand Health." Journal of Quantitative Criminology 9: 159-175. Ross, CatherineE. 2000. "Walking,Exercising, and Smoking: Does Neighborhood Matter?"Social Science and Medicine 51:265-74. Ross, Catherine E. and Diane Hayes. 1988. "Exercise and Psychologic Well-being in the Community." American Journal of Epidemiology 127:762-71. Ross, Catherine E., John Mirowsky, and Karen Goldsteen. 1990. "The Impact of the Family on Health: The Decade in Review." Journal of Marriage and the Family 52:1059-78. Ross, Catherine E. and John Mirowsky. 1999. "Disorder and Decay: The Concept and Measurement of Perceived Neighborhood Disorder."UrbanAffairsReview 34:412-432. Ross, Catherine E. and John Mirowsky. 1999. "Refining the Association between Education and Health:Effects of Quantity,Credential,and Selectivity."Demography36:445-60. Ross, CatherineE. and Chia-ling Wu. 1995. "The American Links between Educationand Health." Sociological Review 60: 719-45. Selye, Hans. 1956. The Stress of Life. New York: McGraw-Hill. Sampson, Robert J. and W Byron Groves. 1989. "Community Structure and Crime: Testing Social-Disorganization Theory." American Journal of Sociology 94:774-802. Sampson, Robert J., Stephen W Raudenbush,and Felton Earls. 1997. "Neighborhoodsand Violent Crime: A Multilevel Study of Collective Efficacy."Science 277: 918-24. Segovia, Jorge, Roy F Bartlett, and Alison C. Edwards. 1989. "TheAssociation between SelfAssessed Health Status and Individual Health Practices."Canadian Journal of Public Health 80:32-37. Skogan,Wesley G. and Michael G. Maxfield. 1981. Coping with Crime.Beverly Hills, CA: Sage. Skogan, Wesley G. 1986. "Fear of Crime and Neighborhood Change." Pp. 39-78 in Communities and Crime, edited by Albert J. Reiss and M. Tonrey. Chicago: University of Chicago Press. Skogan, Wesley G. 1990. Disorder and Decline: Crime and the Spiral of Decay in American

276

JOURNALOF HEALTHAND SOCIALBEHAVIOR


Umberson, Debra, Kristi Williams, and Susan Sharp. 2000. "Medical Sociology and Health Psychology." Pp. 353-64 in The Handbook of Medical Sociology, Fifth Edition, Edited by Chloe E. Bird, Peter Conrad, and Allen M. Fremont.Upper Saddle River,NJ: PrenticeHall. U.S. Bureau of the Census. 1985. Statistical Abstract of the U.S. Washington, DC: U.S. GovernmentPrintingOffice. U.S. Bureau of the Census. 1992. Census of Population and Housing, 1990: SummaryTape File 3 on CD-ROM [MRDF]. Preparedand distributed by the Bureau of the Census. Washington,DC: The Census Bureau. U.S. PreventiveTaskForce. 1989. Guide to Clinical PreventiveServices. Baltimore,MD: Williams& Wilkins. Waitzman,N.J. and K.R. Smith. 1998. "Phantomof the Area: Poverty-Area Residence and Mortality in the United States." American Journal of Public Health 88:973-76. Wilson, William Julius. 1987. The Truly Disadvantaged: The Inner City, the Underclass and Public Policy. Chicago: University of Chicago Press. Wilson, William Julius. 1991. "StudyingInner-City Social Dislocations: The Challenge of Public Agenda Research." American Sociological Review 56:1-14. Wilson, William Julius. 1996. When Work Disappears. The World of the New Urban Poor. New York:Alfred A. Knopf. Winship, Christopher and Larry Radbill. 1994. "Sampling Weights and Regression Analysis." Sociological Methodsand Research 22:230-57.

Neighborhoods. Berkeley and Los Angeles: University of CaliforniaPress. Sloggett, Andrew and HeatherJoshi. 1994. "Higher Mortality in Deprived Areas: Community or Personal Disadvantage?" British Medical Journal 309: 470-74. Sloggett, Andrew and Heather Joshi. 1998. "Deprivation Indicators as Predictors of Life Events 1981-1992 Based on the UK ONS Longitudinal Study."Journal of Epidemiology Health 52: 228-33. and Community South, Scott J. and Kyle D. Crowder. 1997. Distressed "Escaping Neighborhoods: Individual,Community and Metropolitaninfluences." American Journal of Sociology 102:1040-84. Taylor,Ralph B. and MargaretHale. 1986. "Testing AlternativeModels of Fearof Crime." Journal of CriminalLaw and Criminology77:151-89. Taylor, Ralph B. and Sally A. Shumaker. 1990. "Local Crime as a NaturalHazard:Implications for Understanding the Relationship Between Disorder and Fear of Crime."AmericanJournal of Community Psychology 18:619-641. Taylor, Shelley, Rena L. Repetti, and Teresa Seeman. 1997. "HealthPsychology: What is an Unhealthy Environmentand How Does it Get Under the Skin?"Annual Review of Psychology 48:411-47. Thibodeau,GaryA., and Kevin T. Patton. 1997. The Human Body in Health and Disease, 2nd ed. New York:Mosby. Tienda,Marta. 1991. "PoorPeople and Poor places: Deciphering Neighborhood Effects on Poverty Outcomes." Pp. 244-62 in Macro-Micro Linkages in Sociology, edited by Joan Huber. Newbury Park,CA: Sage.

Catherine E. Ross is a Professor in the Departmentof Sociology at the Ohio State University.She studies the effects of socioeconomic status,work, family and communityon men's and women'sphysical and mental health, and their sense of control versus powerlessness. Recent publications include "Does Medical Insurance Contribute to Socioeconomic Differentials in Health?"Milbank Quarterly,2000 (with John Mirowsky); and "The ContingentMeaning of Neighborhood Stability for Residents' Psychological Wellbeing" AmericanSociological Review,2000 (with John Reynolds and KarlynGeis). She is writing a book, Education and Health, with John Mirowsky. John Mirowsky is Professorof Sociology at Ohio State. He is principalinvestigatoron a projectfiuded by the National Instituteon Aging to study "Aging, Status and the Sense of Control"and a project fundedby the National Institute of Mental Health to study "Children,Child-Care and Psychological Well-Being" (Catherine E. Ross is co-principal investigator on both). His recent publications include "Economic Hardship Across the Life Course,"(with CatherineE. Ross, AmericanSociological Review, 1999), and"Age, Depression and Attrition in the National Survey of Families and Households" (with John R. Reynolds, Sociological Methods and Research, 2000). With CatherineRoss, he is working on the second edition of Social Causes of Psychological Distress, Aldine de Gruyter.