Does Housing Mobility Policy Improve Health? · between housing and health and brieﬂy discuss...

Does Housing Mobility Policy Improve Health?

Dolores Acevedo-Garcia, Theresa L. Osypuk, and Rebecca E. WerbelHarvard School of Public Health

Ellen R. Meara Harvard Medical School

David M. CutlerHarvard University

Lisa F. Berkman Harvard School of Public Health

Abstract

This article summarizes the empirical evidence for the effect of housing mobility poli-cies on health outcomes. Our focus derived from our interest in housing policies thatmight help reduce health disparities and our finding that, excluding policies concernedwith the physical characteristics of housing (e.g., exposure to lead), only housing mobil-ity has been evaluated for its effects on health.

We reviewed 13 articles covering five housing mobility studies and ranked them accord-ing to their methodological strength. Although health data have been collected in just a few studies, our review finds that this policy may potentially contribute toimproving the health of both adults and children. Yet the empirical evidence is sparse,and only a handful of studies are methodologically sound. To date, the strongest evi-dence derives from the Moving to Opportunity (MTO) demonstration and from theYonkers evaluation of scattered-site public housing.

Keywords: Low-income housing; Neighborhood; Urban policy

Introduction

The aim of this article is to summarize the research evidence that U.S.housing mobility policies may help improve health outcomes. First, wepresent a conceptual framework to distinguish various pathwaysbetween housing and health and briefly discuss several recent reviewsthat have examined some of these pathways. Second, we explain ourfocus on housing mobility policies by highlighting the role that suchpolicies may play in reducing health disparities. Third, we review andassess the research on the health impact of housing mobility policies.Finally, we discuss the implications of our review for future researchand data collection and call for expanded long-term follow-up of housingpolicy demonstrations to determine their health consequences and themechanisms by which such housing policies could help improve health.

Housing Policy Debate · Volume 15, Issue 1 49© Fannie Mae Foundation 2004. All Rights Reserved. 49

Fannie Mae Foundation

50 D. Acevedo-Garcia, T. Osypuk, R. Werbel, E. Meara, D. Cutler, and L. Berkman

Housing and health

Historically, housing settlement patterns and conditions have beenclosely linked to the health of the population. The urbanization processthat accompanied the Industrial Revolution in both the United Statesand Western Europe gave rise to an increase in mortality in urbanareas. Infectious disease (e.g., tuberculosis) rates soared because ofincreased population density and overcrowding (Preston and Haines1991). Until the 19th century, both housing policy and public healthpolicy in Britain and other European countries were the responsibilityof a single government entity: the Ministry of Health (Whitehead2000). Although these functions were later separated, it is still commonfor public health practitioners to regard housing as an area for publichealth advocacy and action (Freeman 2002; Krieger and Higgins 2002;Srinivasan, O’Fallon, and Dearry 2003; Thiele 2002).

Although housing has long been hypothesized to affect health, docu-menting this relationship has been challenging. Some authors haveargued that although the role of housing is part of the conventionalwisdom on public health and is presented as a “factual reality that (byimplication) precedes and/or transcends the epistemology of social sci-ence” (Allen 2000, 49), there is in fact limited research evidence thatbad housing causes illness. In an editorial summarizing the findings of a special issue of Housing Studies dedicated to housing and health,Whitehead concluded that “[the articles fell] short of proof, either ofthe cause and effect or of the value of particular interventions, but indoing so [reflected] the current state of understanding of the relation-ship between health and housing” (2000, 339).

Pathways between housing and health

Several recent (2000–2002) reviews have examined the empirical evi-dence that housing has an effect on health. The conclusions are mixed.While some argued that the relationship between housing and health isclearly established (Krieger and Higgins 2002), others found a lack ofevidence (Thomson, Petticrew, and Morrison 2001). Summarizing thesereviews is difficult, however, because they focused on different path-ways between housing and health.

The pathways proposed in the literature can be grouped into threecategories1:

1 This is only one possible framework. Alternatively, Dunn (2002) theorizes that hous-ing affects health via three pathways: material (the role of housing affordability and

Does Housing Mobility Policy Improve Health? 51

1. Housing units as an immediate living environment. Housing unitsmay be the source of exposure to dangerous physical (poorlydesigned stairs), chemical (lead), and biological (cockroach aller-gens) conditions. Most of the literature on housing and health hasfocused on the effect of this material aspect of housing on specificillnesses (Allen 2000; Dunn 2000).

2. Housing as an expression of socioeconomic status. Housing is oftenexamined as a socioeconomic factor, categorized as homelessnessand housing tenure (homeownership status). As suggested by Dunn(2000), the research on homelessness generally has a disease focusand emphasizes the gap between the homeless and the rest of soci-ety, thus ignoring the possibility of a monotonic housing gradient inhealth comprising various aspects of housing, such as tenure, hous-ing wealth, mortgage debt burden, and neighborhood quality. Likeother indicators of socioeconomic status, homeownership displaysan association with health: Namely, people with higher socio-economic status (homeowners) may have better health outcomesthan those with lower socioeconomic status (renters). Some studieshave explored the robustness of predicting health with homeowner-ship as a measure of individual socioeconomic advantage at differ-ent stages of life (Kuh et al. 2002; Robert and House 1996;Wadsworth, Montgomery, and Bartley 1999). The magnitude of thehomeownership effect could be modest at the individual level, aboveand beyond other socioeconomic characteristics. In some healthstudies that controlled for the characteristics of the unit, housingtenure was no longer significant (Allen 2000), suggesting thathomeownership may be better for health because of the better phys-ical condition of the unit. Others have hypothesized that the psy-chosocial benefits of homeownership may be mediated by housingand neighborhood conditions (Kearns et al. 2000). Dunn’s (2000)criticism of the literature on homelessness and health can beapplied to research on homeownership and health, since this bodyof work also tends to overlook the complexity of the housing gradi-ent in health.

3. Locational aspect of housing. The location of housing may influencehealth through access to adequate physical and social environments(safe areas for children to play, social support networks) and publicand private services (policing, transport) (Macintyre and Ellaway2000). A growing literature is documenting the health effects ofneighborhood environments (Ellen, Mijanovich, and Dillman 2001;

Housing Policy Debate

accumulation of wealth, as well as the physical conditions of housing), meaningful (howpeople perceive and internalize their housing circumstances), and spatial (the role ofhousehold location and neighborhood).



Ellen and Turner 1997; Kawachi and Berkman 2003b; Leventhaland Brooks-Gunn 2000) and segregated housing patterns (Acevedo-Garcia and Lochner 2003; Acevedo-Garcia et al. 2003; Ellen 2000;Williams and Collins 2001). In addition, some studies have tried todisentangle the effect of the quality of the housing unit from thequality of the neighborhood where the unit is located (Kearns et al.2000).

Next, we summarize the findings of some recent reviews of the empiri-cal evidence on the link between housing and health; we place ourreview in the context of other recent work that underscores a renewedinterest in housing issues in the public health field. However, we do notaim to provide a comprehensive meta-analysis of this literature.

In the British Medical Journal, Thomson, Petticrew, and Morrison(2001) examined housing interventions aimed at rehousing and improv-ing infrastructure, defined as heating, double-glazing, and generalrefurbishment.2 They found that although most studies demonstratedsome health gains, small samples and inadequate control for con-founders limited the ability to generalize the findings. In an issue of theAmerican Journal of Public Health dedicated to the built environmentand health, Saegert et al. (2003) reviewed U.S. interventions to improvehealth by modifying housing units from 1990 to 2001 and concluded(1) that most studies addressed a single condition (typically lead poison-ing, injuries, or asthma) and (2) that while most evaluations reportedstatistically significant improvements, only 14 percent were regarded as “extremely successful” (1471).

In an issue of the American Journal of Public Health dedicated to hous-ing, Krieger and Higgins (2002) concluded that “an increasing body ofevidence” (758) has shown that housing as a determinant of health isassociated with various conditions through material aspects of thehousing unit (such as crowding or dampness), socioeconomic aspects(such as housing deprivation and the availability of subsidies), and loca-tional aspects (such as air quality, noise, and fear of crime).

In a review of the North American and British literature in HousingStudies, Dunn (2000) identified three main areas of housing and healthresearch: housing as a social determinant of health (which has beenoften examined in relation to homelessness), the pathological effects ofinadequate housing, and the psychological distress associated with liv-ing in inadequate housing. There is sufficient evidence that homeless-ness is associated with poor physical and mental health and limited

2 These authors excluded interventions to address lead, urea formaldehyde foam, poorair quality, allergens, and radon. No rationale was provided for these exclusions.


access to health care. The literature on the pathological effects of suchhousing attributes as high-rise housing, crowding, dampness, and coldis vast, although Dunn (2000) did not present an assessment of the evi-dence. Research on housing and stress is relatively recent, and the evi-dence is limited. Regarding housing policy, Dunn (2000) found thatthere is equivocal evidence on the effectiveness of British policies aimedat enhancing the health of individuals with medical problems by givingthem priority for public housing.

In 2002, the Centers for Disease Control and Prevention (CDC) TaskForce on Community Preventive Services published a series of litera-ture reviews on policy interventions that could help improve commu-nity health (Anderson et al. 2002). Mixed-income housing and housingvouchers were examined. The authors did not find studies of mixed-income housing that met their design criteria for inclusion in theirreview, but did find sufficient evidence that housing voucher (mobility)policies reduce crime victimization and thus can be recommended. Ourreview differs from the CDC study in four main respects:

1. We examine housing mobility policies in the context of other hous-ing and health research and other housing policies that might helpreduce health disparities.

2. We examine various U.S. housing mobility policies, not just theMoving to Opportunity (MTO) demonstration.

3. We focus on health outcomes, while the CDC also included broadindicators that could have health implications, such as housingquality and neighborhood conditions.

4. In addition to summarizing the evidence, we discuss the design andfindings of each study.3

Methodological challenges in housing and health research

Research on housing and health presents several methodological chal-lenges. First, although there have been important advances in concep-tualizing the relationship between housing and health—for example,Dunn’s (2002) work on the material, meaningful, and spatial dimen-sions of housing—for the most part, conceptual frameworks have yet


3 Our review is also different from the first cross-site analysis of MTO by Goering,Feins, and Richardson (2002), who focused only on MTO and discussed health as partof a series of MTO outcomes, including crime, education, employment, and reliance onpublic assistance.

to be incorporated into the design of empirical studies. In a review ofthe literature, Allen (2000) concluded that research on housing andhealth has been largely atheoretical. Macintyre and Ellaway (2003)similarly assert that neighborhood and health research lacks theoreti-cal development.

Second, it is difficult to establish whether bad housing leads to poorhealth or vice versa. Most research has focused on cross-sectional sta-tistical associations between housing and health variables. The selec-tion of individuals into housing and neighborhoods presents the largestpotential source of bias in nonexperimental studies (which to date con-stitute the bulk of research in this area), since those with better healthmay be able to settle in better homes and neighborhoods. Researcherstherefore need to construct rigorous research designs to simulate aplausible counterfactual scenario (what would have happened in theabsence of the policy) (Orr 1999) in the absence of randomized designs(e.g., quasi-experimental designs, time-series designs). Longitudinalresearch designs are required to explore the direction of the linkbetween housing and health, as well as to examine the temporal dimen-sions of these associations: that is, variation in housing/neighborhoodenvironment and accumulation of deprivation across the life course,susceptibility to illness during social or physiological critical periods,and latency periods for the development of illness (Bartley, Blane, andMontgomery 1997). However, few longitudinal studies have shown anassociation between housing conditions in childhood and adult health(Marsh et al. 2000).

Third, only a few empirical studies have carefully examined specificpathways between housing and health. It has been difficult to isolate agiven pathway from other possible mechanisms. For example, it may behard to separate poor housing conditions from other facets of socioeco-nomic deprivation such as poverty. Similarly, it may be difficult to dis-entangle the effect of a housing unit’s condition from locational effects,specifically neighborhood conditions. Multilevel research designs arebeing used to disentangle these effects (Acevedo-Garcia et al. 2003;Ellen et al. 2001; O’Campo 2003; Subramanian, Jones, and Duncan2003). Identifying mechanisms is especially important for tying any onespecific housing policy to health, since policy often occurs far upstreamof health problems in the causal chain of events: Specifically, Williamsand Collins (2001) point to such mediators as improved physical orsocial neighborhood environment.

Fourth, plausible biological pathways between housing and health arewell established for some exposure/outcome combinations (exposure tolead and cognitive function in children), but most research has not




addressed the biological links between housing and health. Biomarkers,which are objectively measured indicators of “normal biologic proc-esses, pathogenic processes, or pharmacologic responses” (De Gruttolaet al. 2001, 487), are commonly employed to gauge toxic exposure inoccupational and environmental health (lead and carbon monoxide), todocument smoking status and exposure to secondhand smoke (nicotinelevels), and to evaluate the effectiveness of pharmacologic interventions(to reduce blood pressure or to treat AIDS) (Christiani 1996; De Grut-tola et al. 2001). Yet the use of biomarkers in housing and health inter-ventions, aside from gauging toxic exposure, is incipient (Loucks et al.2002). Use of objective measures of physiologic responses to housingmay provide unbiased measures for outcomes subject to bias fromsocial desirability or recall or for outcomes that are imperceptible forself-report, such as blood pressure and immune response (Loucks et al.2002).

Housing and health disparities

Understanding the source of health disparities and addressing themalong racial/ethnic and socioeconomic lines has become an increasinglyimportant topic in public health research and practice (Geronimus2000; Williams 1997; Williams and Collins 2001; Williams and Harris-Reid 1999). Sociological research suggests that housing factors mayunderlie disparities in socioeconomic outcomes, which have been widelyrecognized as upstream determinants of health (Berkman and Kawachi2000). For example, at the household level, limited access to homeown-ership contributes to substantial differences in wealth and its intergen-erational accumulation across racial/ethnic groups. In turn, wealthdisparities are linked to educational and labor market disparities (Con-ley 1999). There is also concern that housing segregation along raciallines and the spatial concentration of poverty separate racial/ethnicgroups into vastly different neighborhood environments. In turn, thenegative effects imposed on individuals living in highly disadvantagedneighborhoods are the source of disparities in socioeconomic outcomes(Altshuler et al. 1999; Iannotta, Ross, and the National ResearchCouncil 2002). Disparities in both individual-level and neighborhood-level socioeconomic outcomes may underlie health disparities alongracial/ethnic lines (Williams 1996, 1997; Williams and Collins 2001).

Social epidemiology research is beginning to show the effect of neigh-borhood characteristics independent of individual-level socioeconomicfactors on racial/ethnic disparities in health outcomes—for example, low birth weight (Buka et al. 2003). Research has also documented a




positive association between housing segregation and mortality andmorbidity among blacks. Although most of these studies have somemethodological weaknesses, the evidence to date consistently suggeststhat blacks in urban areas characterized by higher levels of housing seg-regation experience higher mortality and morbidity (Acevedo-Garciaand Lochner 2003; Acevedo-Garcia et al. 2003; Williams and Collins2001). However, research on both neighborhoods and segregation inrelation to health outcomes has been atheoretical, and the identificationof causal mechanisms has only recently become an area of exploration(Acevedo-Garcia and Lochner 2003; Acevedo-Garcia et al. 2003;O’Campo 2003). As noted by O’Campo, “We still lack a clear picture ofthe intervention and policy implications of this body of work” (2003, 9).4

In this article, we focus on the evidence that housing mobility policymay serve as a tool to improve health outcomes. Our previous work hasfocused on the locational aspects of housing—segregation and neighbor-hood environment—as a social determinant of health disparities.Therefore, we are particularly interested in whether housing policyinterventions could help address health disparities. Most of the empiri-cal evidence to date addresses the physical condition of units, and hous-ing interventions and policies aimed at improving those conditions arewidely recognized as tools to protect public health (Mood 1986). How-ever, while policies that address the locational aspects of housing arejust beginning to be recognized as public health tools, in the past threedecades housing policy has been increasingly informed by concernsabout racial and income segregation. We set out to examine theresearch evidence that public policies that have influenced or tried tocorrect the spatial and social isolation of low-income and racial/ethnicminority groups have also had an impact on their health outcomes.Recent social policy experiments have begun to document how, byenhancing people’s neighborhood environment, housing mobility policymight benefit health.

Housing policy and health

Housing is an important social policy domain. It represents one of the largest monthly costs for many Americans, and equity from

4 Although public health research has recognized a possible link between segregation,disadvantaged neighborhood environments, and ill health, health research is onlybeginning to address the role of public policy in health inequalities. For example, Hartet al. (1998) showed that metropolitan areas characterized by metropolitan governancehad lower black mortality rates than areas characterized by municipal fragmentationand that housing segregation mediated the effect of metropolitan governance on blackmale mortality.


homeownership is the most common source of household wealth (U.S.Bureau of the Census and Economics and Statistics Administration2001). The government, at all levels, intervenes in the housing marketby providing, subsidizing, or constraining the purchase, location, orrental of housing in a variety of ways, including tax policy, zoning laws,and provision of housing for low-income residents.

In this literature review, we first sought to identify those federal hous-ing policies that appear to have the greatest potential for improving thehealth of low-income individuals and, thus, for reducing health dispari-ties.5 On the basis of empirical evidence for various aspects of housingas a social determinant and for racial/ethnic and socioeconomic dispari-ties in housing outcomes, we identified three such policies: homeowner-ship promotion, enforcement of antidiscrimination laws, and housingmobility. We then focused on housing mobility, since it appeared to bethe only one that has been empirically evaluated for its effects onhealth. Additionally, this focus is warranted because such policies areembedded in rental assistance programs, which constitute the largest(in monetary terms) and the most politically viable vehicle for address-ing the needs of the U.S. low-income population facing the most severehousing problems and concentration of poverty (Sard 2000).

Homeownership policy

There are several pathways through which homeownership may helpimprove health. First, homeownership is the main source of wealth formost American households. Access to home equity provides homeown-ing households with greater financial stability and enables them tobenefit from other investment opportunities. In 2001, the median netwealth for U.S. homeowner households was about $171,800, comparedwith only about $4,810 for renter households (Joint Center for HousingStudies 2003). Potentially, promoting homeownership and thus accu-mulation of wealth could be an effective way to enhance health.6 A vastbody of social epidemiology literature suggests an inverse gradient


5 Our focus does not derive from an assessment that federal policy is more influentialthan state or local housing policy, but rather from an organizing principle: We decidedto start our review work by focusing first on national policy. Future reviews mayaddress state and local housing policy. However, our examination of the literature sug-gests that there may be only limited instances of state and local policies that have beenevaluated for their effect on health with the criteria we established.

6 A related (and controversial) issue is whether redistribution of wealth could helpreduce health inequalities (Deaton 2002).

between socioeconomic status (income, education, and occupation) andphysical and mental health outcomes (Barker et al. 1993; Hattersley1999; Kaplan and Salonen 1990; Lundberg 1993; Marmot et al. 1991;Power et al. 1997; Syme 1987; Syme and Berkman 1976). Althoughstudies of the association between wealth, including homeownership,and health are relatively scarce because of lack of data on accumulationof wealth, the available evidence suggests a positive associationbetween wealth and health (Deaton 2002) and between homeownershipand health (Dunn 2000; Kuh et al. 2002). Some studies found thathomeownership is associated with a lower probability of limiting long-term illness (Wadsworth, Montgomery, and Bartley 1999); others foundthat children living in houses owned by their parents experienced lowerrates of behavioral, emotional, and cognitive problems and recom-mended promoting homeownership as a strategy for reducing suchproblems (Boyle 2002; Haurin, Parcel, and Haurin 2002).

Further, although the evidence is limited, homeownership may conferpsychological benefits (Dunn and Hayes 2000; U.S. Department ofHousing and Urban Development [HUD] 1995) and may contribute tobetter neighborhood physical and social environments (HUD 1995),which in turn may have health-promoting effects. However, the burdenof a home mortgage can also detrimentally influence health. Forinstance, in a nationally representative British panel study, Nettletonand Burrows (1998) documented that when homeowners experiencedproblems paying their mortgage or fell into arrears, their subjectivewell-being declined compared with those who had no mortgage prob-lems, even after adjusting for health problems, change in health status,income, and employment since baseline. Thus, the fear of losing one’shome results in worse health despite the fact that relatively few peopleactually experience repossession (Nettleton and Burrows 1998).

Although it would be important to incorporate health impact assess-ment into policy demonstrations aimed at promoting homeownershipamong racial/ethnic minority and low-income households, to our knowl-edge no such studies exist. Also, current policy goals assume that“homeownership is not an option for everyone”—that some low-incomehouseholds may be better served by rental assistance (HUD 2002, 7).

Fair housing policy

Enforcing antidiscrimination laws is another key component of housingpolicy. Housing discrimination and antidiscrimination policy could alsopotentially affect health outcomes through at least two mechanisms.First, since discrimination limits homeownership among minorities




(Yinger 1995), antidiscrimination policy could enhance minority home-ownership and accumulation of wealth and, as a result, minorityhealth. Further, health research suggests that institutional housing dis-crimination (Gee 2002), as well as perceptions of racism and discrimi-nation, could have detrimental effects on cardiovascular and mentalhealth, for example (Harrell, Hall, and Taliaferro 2003; Jackson et al.1996; Krieger and Sidney 1996; Williams and Neighbors 2001;Williams, Neighbors, and Jackson 2003). However, to our knowledge,housing antidiscrimination policy has not been evaluated for its effectson health.

Housing mobility policy

In the United States, rental assistance is the main form of governmenthousing assistance for low-income households (HUD 2002). Moreover,rental assistance, in the form of housing vouchers, is increasingly beingexamined as a tool for deconcentrating poverty (Sard 2000). Besidesproviding income assistance to low-income households, the Section 8program by design could help reduce the concentration of poverty as itfacilitates housing mobility. Participants are free to choose any housingthat meets program requirements and are not limited to units in subsi-dized housing projects, which are generally located in distressed areas(Sard 2000; Turner, Popkin, and Cunningham 2000). Analyses of neigh-borhood characteristics of participants in different federal housing pro-grams have found that Section 8 certificates and vouchers “reduce theprobability that families live in the most economically and socially dis-tressed areas” (Newman and Schnare 1997, 703). Only about 15 per-cent of Section 8 recipients live in neighborhoods with poverty rates inexcess of 30 percent. By contrast, about 54 percent of public housingresidents live in such neighborhoods (Newman and Schnare 1997;Turner, Popkin, and Cunningham 2000). Therefore, the Section 8 pro-gram may benefit participants both through an effect on householdincome and through an effect on neighborhood quality. Both are likelyto have a beneficial effect on health outcomes.


Empirical evidence on housing mobility policy and health

Literature search methods

We began by casting a wide net for studies of any housing policies thatmight have been evaluated for health,7 but at the same time, we alsoimposed rigorous selection criteria. We searched eight medical, socialscience, policy, and housing databases for articles from 1974 throughApril 2002: MEDLINE, SocioFile, PsychINFO, the Social Sciences Cita-tion Index, the HUD USER Bibliographic Database, ProQuest, Policy-File, and JSTOR (which contains 35 journals for geography, populationstudies, and sociology). We used the following keywords in our search:public policy or policy, housing, and health. We also reviewed referenceswithin relevant articles, Internet sites, and housing journals: Health &Place (1995–2002); Housing Studies (1998–2002); Housing Theory andSociety (October 1999–August 2000); Urban Studies (1995–2002);Housing Policy Debate (1993–2002); Journal of Housing Research(1993–2002); Social Science and Medicine (1995–2002); and the Prince-ton Moving to Opportunity, HUD, and Urban Institute Web sites.

To be included in our review, studies must have fulfilled the followingconditions: (1) empirically evaluated a housing policy in the UnitedStates, (2) included at least one relevant health outcome, and (3) had acomparison group.

Relevant health outcomes were included as (1) direct health outcomes(individually measured mental or physical health, including experienceof violence), (2) hazardous health behaviors (substance abuse), and (3) medical care. Although several studies included indicators thatcould be proxies for health (number of school absences, unsafe housingor neighborhood conditions), we have not included such indirectoutcomes here.

We found three main bodies of research: housing mobility policy, hous-ing policy for the homeless/mentally ill, and lead standards policy. Wechose to focus on housing mobility as a potential tool for reducinghealth disparities. We did not find any studies on the health effects ofhomeownership promotion policy and antidiscrimination policy, the two



7 Some of the earliest work on the effect of housing mobility and health focused on thesevere distress that accompanied forced relocation because of urban renewal projects.For instance, 46 percent of women relocated from Boston’s West End experienced long-term grief reactions; 57 percent of these women reported that they continued to havesymptoms of depression two years after relocation (Fried 1963).


other areas we had identified as possible vehicles for reducing healthdisparities. We excluded housing standards on lead, because, unlikehousing mobility, they are already widely recognized as a tool forprotecting health, and housing policy for the homeless or mentally illpopulation, because this research overlooks the possibility of a morecomplex housing gradient in health (Dunn 2000).8

Housing mobility studies that met our inclusion criteria were docu-mented in an evaluation form developed at the CDC by Briss et al.(2000). Using the guidelines developed by these authors, we reviewedeach of the studies and ranked them by the strength of their studydesigns. Our evaluation was based on a number of criteria, includingstudy population and design, sampling methods, validity and reliabilityof exposure and outcome measures, data analysis, attrition, authors’interpretation of results, and presence of other major threats to validity.

Findings

As shown in figure 1, our search yielded 479 citations, which werereviewed for relevance to our criteria. A total of 94 articles were keptfor closer inspection, and of these, 13 articles qualified for inclusion.These articles discussed five residential mobility policies: the MTOexperiment, the Gautreaux program, the Yonkers scattered-site publichousing program, the Section 8 program, and the Cincinnati SpecialMobility Program (table 1). We will first summarize the effects of eachpolicy on health (table 2) and then summarize the evidence.

Gautreaux program

The Gautreaux Assisted Housing Program, the first and best known ofthe litigation-initiated residential mobility housing programs, resultedfrom a lawsuit that was filed by Dorothy Gautreaux and other publichousing residents against HUD and alleged racial discrimination in theadministration of the Chicago public housing program. In 1976, theSupreme Court ordered the city of Chicago to racially desegregate itspublic housing and offer placements to black families in private unitslocated in other parts of the metropolitan area. Public housing resi-dents and those on the waiting list received Section 8 certificates andhousing counseling to move either within the city or to one of morethan 115 predominantly white suburbs.


8 Additionally, housing policy for the homeless/mentally ill population is a local policyarea, while our focus here is on federal policies.

Figure 1. Search Strategy

Note: A total of 423 citations was generated by the database and 56 more were generated by apurposive search of Web sites, references, and so on.

HUD evaluated the Gautreaux program in 1979, comparing Gautreauxparticipants with eligible nonparticipants and Section 8 participants(Peroff et al. 1979). Rosenbaum later used three surveys to evaluate theeffects of the Gautreaux program: One was a cross-sectional survey ofadults examining employment outcomes, and the other two werepre/post surveys of the same sample of household heads and their chil-dren, analyzing social and educational outcomes over seven years(1994, 1995). Suburban movers were significantly less satisfied withtheir medical care than city movers, and the latter experienced greatersatisfaction at follow-up, compared with baseline (Rosenbaum and Pop-kin 1990). The 1979 HUD evaluation of Gautreaux reported that accessto medical care was more difficult for participants than for eligible non-participants or Section 8 recipients; access was also more difficult forparticipants after the move than it had been before (Peroff et al. 1979).

The Gautreaux evaluation results are subject to selection bias, however.First, in the 1979 survey, those selecting into the program were differ-ent from eligible nonparticipants and Section 8 recipients for reasonsrelating to their participation as well as reasons affecting their health.

479 citations generated

385 disqualified from review: not relevant

(nonhuman, not a U.S. housing mobility policy)

94 articles reviewed for closer inspection; relevant articles

Pool of studies included in the final literature

review: 13

81 studies disqualifiedif there was no health

outcome included or no comparison group




Tab

le 1

.Su

mm

ary

of H

ousi

ng

Mob

ilit

y S

tud

ies

wit

h H

ealt

h-R

elat

ed O

utc

omes

, Cla

ssifi

ed b

y M

eth

odol

ogic

al

Str

engt

h a

nd

Su

itab

ilit

y of

Stu

dy

Des

ign

Hea

lth-

Rel

ated

Out

com

es

Cla

ssifi

cati

on o

fM

etho

dolo

gica

lSu

itab

ility

of

Sour

ceP

olic

yIn

clud

ed in

the

Rev

iew

Out

com

e T

ype

Stre

ngt

h*St

udy

Des

ign

*

Kat

z, K

ling,

an

dL

iebm

an 2

001

MT

O B

osto

nC

rim

e vi

ctim

izat

ion

; in

juri

es, a

ccid

ents

, or

pois

onin

gs;

asth

ma

atta

ck r

equi

rin

g m

edic

al a

tten

tion

; 7-i

tem

par

ent

repo

rt o

f ch

ild b

ehav

ior

prob

lem

s; a

dult

sel

f-re

port

edhe

alth

; men

tal h

ealt

h (r

epor

ted

maj

or d

epre

ssiv

esy

mpt

om);

cal

m/p

eace

ful;

been

to

med

ical

car

e fo

r re

gula

rch

ecku

p or

imm

uniz

atio

n

Hea

lth,

men

tal

heal

th, c

rim

evi

ctim

izat

ion

,he

alth

car

eut

iliza

tion

1 (g

ood)

Gre

ates

t

Lev

enth

al a

nd

Bro

oks-

Gun

n20

03b

MT

O N

ew Y

ork

Men

tal h

ealt

h fo

r ch

ildre

n (

prob

lem

beh

avio

ral o

utco

mes

)an

d m

othe

rs

Men

tal h

ealt

h2

(fai

r)G

reat

est

Faut

h, L

even

thal

, an

d B

rook

s-G

unn

200

2

Yon

kers

A

dole

scen

t ex

posu

re t

o vi

olen

ce a

nd

acce

ss t

o ci

gare

ttes

and/

or a

lcoh

olC

rim

evi

ctim

izat

ion

and

acce

ss t

oill

egal

hea

lth-

harm

ing

subs

tan

ces

3 (f

air)

Gre

ates

t

Ros

enba

um a

nd

Pop

kin

199

0G

autr

eaux

Sati

sfac

tion

wit

h m

edic

al c

are

Hea

lth

care

3

(fai

r)G

reat

est

Lev

enth

al a

nd

Bro

oks-

Gun

n20

03a

MT

O N

ew Y

ork

Ado

lesc

ent

use

of c

igar

ette

s an

d al

coho

l; ch

ild m

enta

lhe

alth

; chi

ld h

ealt

h; e

xpos

ure

to v

iole

nce

Hea

lth,

hea

lth

beha

vior

/su

bsta

nce

abus

e, m

enta

lhe

alth

, cri

me

vict

imiz

atio

n

3 (f

air)

Gre

ates

t


Tab

le 1

.Su

mm

ary

of H

ousi

ng

Mob

ilit

y S

tud

ies

wit

h H

ealt

h-R

elat

ed O

utc

omes

, Cla

ssifi

ed b

y M

eth

odol

ogic

al

Str

engt

h a

nd

Su

itab

ilit

y of

Stu

dy

Des

ign

(co

nti

nu

ed)

Hea

lth-

Rel

ated

Out

com

es

Cla

ssifi

cati

on o

fM

etho

dolo

gica

lSu

itab

ility

of

Sour

ceP

olic

yIn

clud

ed in

the

Rev

iew

Out

com

e T

ype

Stre

ngt

h*St

udy

Des

ign

*

*So

urce

for

rat

ing

syst

em: B

riss

et

al. 2

000

Bri

ggs

1997

an

dth

e Yo

nke

rsFa

mily

an

dC

omm

unit

yP

roje

ct

Yon

kers

Scat

tere

d-Si

teP

ublic

Hou

sin

g

Mot

her’

s re

port

ed m

enta

l hea

lth

sym

ptom

s (d

epre

ssio

n,

anxi

ety)

, sub

stan

ce a

buse

(al

coho

l, dr

ugs)

, exp

erie

nce

of

viol

ence

dur

ing

the

past

12

mon

ths

Men

tal h

ealt

h,he

alth

beh

avio

r/su

bsta

nce

ab

use,

cri

me

vict

imiz

atio

n

4 (f

air)

Gre

ates

t

Mey

ers

et a

l. 19

95Se

ctio

n 8

Gro

wth

of

child

(he

ight

an

d w

eigh

t fo

r ag

e), a

s a

prox

yfo

r n

utri

tion

Hea

lth/

child

deve

lopm

ent

5 (l

imit

ed)

Lea

st

Ros

enba

um a

nd

Har

ris

2001

MT

O C

hica

goE

xper

ien

ce o

f vi

olen

ce o

r cr

ime

Cri

me

vict

imiz

atio

n5

(lim

ited

)G

reat

est

Per

off

et a

l. 19

79G

autr

eaux

Hea

lth

care

acc

ess

easy

or

diffi

cult

Hea

lth

care

5 (l

imit

ed)

Gre

ates

t

Han

ratt

y,M

cLan

ahan

,an

d P

etti

t 20

03

MT

OL

os A

nge

les

Med

ical

car

e ac

cess

, reg

ular

pla

ce f

or c

are

Hea

lth

care

5 (l

imit

ed)

Gre

ates

t

Ros

enba

um 1

994

Gau

trea

uxV

icti

miz

atio

n: W

heth

er y

outh

wer

e hu

rt b

y ot

her

stud

ents

in s

choo

lC

rim

evi

ctim

izat

ion

6 (l

imit

ed)

Gre

ates

t

Ros

enba

um 1

995

Gau

trea

uxV

icti

miz

atio

n: W

heth

er y

outh

wer

e hu

rt b

y ot

her

stud

ents

in s

choo

lC

rim

evi

ctim

izat

ion

6 (l

imit

ed)

Gre

ates

t

Fis

cher

199

1C

inci

nn

ati

Spec

ial M

obili

tyP

rogr

am

Bei

ng

atta

cked

/per

son

al e

xper

ien

ce o

f vi

olen

ce o

r ro

bber

yfo

r se

lf, c

hild

, or

nei

ghbo

r; r

ecei

pt o

f he

alth

car

e be

nefi

ts;

sati

sfac

tion

wit

h he

alth

car

e

Cri

me

vict

imiz

atio

n,

heal

th c

are

8 (l

imit

ed)

Lea

st




Tab

le 2

.Su

mm

ary

of S

ign

ifica

nt

(p<

0.0

5) S

hor

t-T

erm

Hea

lth

, Hea

lth

Car

e, a

nd

Hea

lth

Beh

avio

r O

utc

omes

A

ssoc

iate

d w

ith

Hou

sin

g M

obil

ity

Pol

icie

s Sign

ifica

nt

amon

gw

hich

gro

up?

Age

Com

pari

son

Hea

lth

(In

ten

t to

Tre

atP

olic

ySo

urce

Gro

upM

ade

Indi

cato

rR

esul

ts R

epor

ted)

Inte

rpre

tati

on

Bos

ton

M

TO

Kat

z, K

ling,

and

Lie

bman

2001

Chi

ldre

n6

to 1

5M

TO

gro

up (

also

kn

own

as

the

expe

rim

enta

l gro

up)

and/

or S

ecti

on 8

gro

up v

s.co

ntr

ol g

roup

, at

follo

w-u

p

An

y in

juri

es, a

ccid

ents

, or

pois

onin

gs o

ver

the

past

6

mon

ths

requ

irin

g m

edic

alat

ten

tion

MT

OT

he M

TO

gro

up h

ad f

ewer

inci

den

ts t

han

the

con

trol

grou

p.

Chi

ldre

n6

to 1

5M

TO

gro

up (

also

kn

own

as

the

expe

rim

enta

l gro

up)

and/

or S

ecti

on 8

gro

up v

s.co

ntr

ol g

roup

, at

follo

w-u

p

Fra

ctio

n o

f 7

beha

vior

pro

b-le

ms

amon

g bo

ysB

oth

MT

O a

nd

Sect

ion

8

Bot

h ex

peri

men

tal g

roup

sha

d fe

wer

beh

avio

r pr

oble

ms

than

the

con

trol

gro

up.

Adu

lts

MT

O g

roup

(al

so k

now

n a

sth

e ex

peri

men

tal g

roup

)an

d/or

Sec

tion

8 g

roup

vs.

con

trol

gro

up, a

t fo

llow

-up

Ove

rall

heal

th g

ood

or b

ette

rB

oth

MT

O a

nd

Sect

ion

8

Bot

h ex

peri

men

tal g

roup

sw

ere

heal

thie

r th

an t

heco

ntr

ol g

roup

.

Adu

lts

MT

O g

roup

(al

so k

now

n a

sth

e ex

peri

men

tal g

roup

)an

d/or

Sec

tion

8 g

roup

vs.

con

trol

gro

up, a

t fo

llow

-up

Cal

m a

nd

peac

eful

“a

good

bit

of t

he t

ime”

or

mor

eof

ten

in t

he p

ast

4 w

eeks

Bot

h M

TO

an

dSe

ctio

n 8

Bot

h ex

peri

men

tal g

roup

sw

ere

calm

er t

han

the

con

trol

grou

p.

New

Yor

k M

TO

Lev

enth

al a

nd

Bro

oks-

Gun

n20

03b

Adu

lts

MT

O a

nd/

or S

ecti

on 8

gr

oup

vs. c

ontr

ol g

roup

, at

follo

w-u

p

Dis

tres

s/an

xiet

y sy

mpt

oms

in t

he p

ast

mon

thM

TO

T

he M

TO

gro

up h

ad f

ewer

sym

ptom

s th

an t

he c

ontr

olgr

oup.

Chi

ldre

n8

to 1

8 M

TO

an

d/or

Sec

tion

8

grou

p vs

. con

trol

gro

up, a

tfo

llow

-up

Pro

blem

beh

avio

r:

anxi

ous/

depr

esse

dM

TO

The

MT

O g

roup

had

few

ersy

mpt

oms

than

the

con

trol

grou

p.B

oys

8 to

18

MT

O a

nd/

or S

ecti

on 8

gr

oup

vs. c

ontr

ol g

roup

, at

follo

w-u

p

Pro

blem

beh

avio

r:

anxi

ous/

depr

esse

d M

TO

The

MT

O g

roup

had

few

ersy

mpt

oms

than

the

con

trol

grou

p.


Tab

le 2

.Su

mm

ary

of S

ign

ifica

nt

(p<

0.0

5) S

hor

t-T

erm

Hea

lth

, Hea

lth

Car

e, a

nd

Hea

lth

Beh

avio

r O

utc

omes

A

ssoc

iate

d w

ith

Hou

sin

g M

obil

ity

Pol

icie

s (c

onti

nued

)

Sign

ifica

nt

amon

gw

hich

gro

up?

Age

Com

pari

son

Hea

lth

(In

ten

t to

Tre

atP

olic

ySo

urce

Gro

upM

ade

Indi

cato

rR

esul

ts R

epor

ted)

Inte

rpre

tati

on

New

Yor

k M

TO

Lev

enth

al a

nd

Bro

oks-

Gun

n20

03b

Boy

s 8

to 1

8M

TO

an

d/or

Sec

tion

8

grou

p vs

. con

trol

gro

up, a

tfo

llow

-up

Pro

blem

beh

avio

r:

depe

nde

nt

Bot

h M

TO

an

dSe

ctio

n 8

Bot

h ex

peri

men

tal g

roup

sha

d fe

wer

sym

ptom

s th

an t

heco

ntr

ol g

roup

.

Chi

ldre

n8

to 1

3M

TO

an

d/or

Sec

tion

8

grou

p vs

. con

trol

gro

up, a

tfo

llow

-up

Pro

blem

beh

avio

r:

head

stro

ng

Sect

ion

8T

he S

ecti

on 8

gro

up h

adfe

wer

sym

ptom

s th

an t

heco

ntr

ol g

roup

.

Lev

enth

al a

nd

Bro

oks-

Gun

n,

2003

a

Chi

ldre

n8

to 1

3M

TO

an

d/or

Sec

tion

8

grou

p vs

. con

trol

gro

up, a

tfo

llow

-up

Pro

blem

beh

avio

r: u

nha

ppy,

sad,

or

depr

esse

d in

the

pas

t6

mon

ths

Sect

ion

8T

he S

ecti

on 8

gro

up r

epor

ted

low

er le

vels

of u

nhap

pine

ss/

sadn

ess

than

the

con

trol

gro

up.

Chi

ldre

n8

to 1

3M

TO

an

d/or

Sec

tion

8

grou

p vs

. con

trol

gro

up, a

tfo

llow

-up

Pro

blem

beh

avio

r: n

eed

tobe

nea

r ad

ults

in t

he p

ast

6 m

onth

s

MT

O g

roup

The

MT

O g

roup

rep

orte

dlo

wer

leve

ls o

f pr

oble

m b

e-ha

vior

tha

n th

e co

ntro

l gro

up.

Chi

ldre

n8

to 1

3M

TO

an

d/or

Sec

tion

8

grou

p vs

. con

trol

gro

up, a

tfo

llow

-up

Pro

blem

beh

avio

r: t

oo

fear

ful o

r an

xiou

s in

the

pa

st 6

mon

ths

MT

O g

roup

The

MT

O g

roup

rep

orte

dlo

wer

leve

ls o

f pr

oble

m b

e-ha

vior

tha

n th

e co

ntro

l gro

up.

Boy

s 8

to 1

8M

TO

an

d/or

Sec

tion

8

grou

p vs

. con

trol

gro

up, a

tfo

llow

-up

Pro

blem

beh

avio

r: t

oo

fear

ful o

r an

xiou

s in

the

pa

st 6

mon

ths

MT

O g

roup

The

MT

O g

roup

rep

orte

dlo

wer

leve

ls o

f pr

oble

m b

e-ha

vior

tha

n th

e co

ntro

l gro

up.

Boy

s 8

to 1

8M

TO

an

d/or

Sec

tion

8

grou

p vs

. con

trol

gro

up, a

tfo

llow

-up

Pro

blem

beh

avio

r: n

eed

tobe

nea

r ad

ults

in t

he p

ast

6 m

onth

s

MT

O g

roup

The

MT

O g

roup

rep

orte

dlo

wer

leve

ls o

f pr

oble

m b

e-ha

vior

tha

n th

e co

ntro

l gro

up.

Boy

s 8

to 1

8M

TO

an

d/or

Sec

tion

8

grou

p vs

. con

trol

gro

up, a

tfo

llow

-up

Pro

blem

beh

avio

r: d

epen

dto

o m

uch

on o

ther

s in

the

past

6 m

onth

s

Sect

ion

8gr

oup

The

Sec

tion

8 g

roup

rep

orte

dlo

wer

leve

ls o

f pr

oble

m b

e-ha

vior

tha

n th

e co

ntro

l gro

up.




Tab

le 2

.Su

mm

ary

of S

ign

ifica

nt

(p<

0.0

5) S

hor

t-T

erm

Hea

lth

, Hea

lth

Car

e, a

nd

Hea

lth

Beh

avio

r O

utc

omes

A

ssoc

iate

d w

ith

Hou

sin

g M

obil

ity

Pol

icie

s (c

onti

nued

)

Sign

ifica

nt

amon

gw

hich

gro

up?

Age

Com

pari

son

Hea

lth

(In

ten

t to

Tre

atP

olic

ySo

urce

Gro

upM

ade

Indi

cato

rR

esul

ts R

epor

ted)

Inte

rpre

tati

on

New

Yor

k M

TO

Lev

enth

al a

nd

Bro

oks-

Gun

n,

2003

a

Gir

ls

11 t

o 18

M

TO

an

d/or

Sec

tion

8 g

roup

vs. c

ontr

ol g

roup

, at

follo

w-u

pT

een

gir

l her

self

use

d al

coho

l in

the

pas

t ye

ar

MT

O g

roup

T

he M

TO

gro

up r

epor

ted

high

er u

se o

f al

coho

l tha

n t

heco

ntr

ol g

roup

.

Boy

s 11

to

18M

TO

an

d/or

Sec

tion

8 g

roup

vs. c

ontr

ol g

roup

, at

follo

w-u

pT

een

rep

ort

that

pee

rs u

sed

ciga

rett

es in

the

pas

t ye

arM

TO

gro

up

The

MT

O g

roup

rep

orte

dpe

ers

used

cig

aret

tes

less

than

the

con

trol

gro

up.

Adu

lts

MT

O, S

ecti

on 8

, or

con

trol

grou

p, c

ompa

rin

g ba

selin

ew

ith

the

follo

w-u

p va

lue

Exp

osur

e to

vio

len

ce(m

ugge

d, t

hrea

ten

ed, b

eate

n,

or s

tabb

ed/s

hot)

All

thre

egr

oups

(MT

O,

Sect

ion

8,

and

con

trol

s)

Eac

h gr

oup

expe

rien

ced

ade

clin

e in

exp

osur

e to

vi

olen

ce f

rom

bas

elin

e to

fo

llow

-up.

Los

An

gele

s M

TO

H

anra

tty,

McL

anah

an,

and

Pet

tit

2003

Adu

lts

MT

O, S

ecti

on 8

, or

con

trol

grou

p, c

ompa

rin

g ba

selin

ew

ith

the

follo

w-u

p va

lue

No

sign

ifica

nt

resu

lts

repo

rted

——

Chi

cago

M

TO

R

osen

baum

an

dH

arri

s 20

01A

ny

mem

ber

of t

heho

use-

hold

Com

pari

son

of

base

line

vers

us f

ollo

w-u

p va

lue

for

eith

er t

he M

TO

or

Sect

ion

8gr

oup

(aut

hors

did

not

in

terv

iew

the

con

trol

gro

up

at f

ollo

w-u

p)

Bei

ng

thre

aten

ed w

ith

akn

ife

or g

un

Bot

h M

TO

an

dSe

ctio

n 8

grou

ps

The

re w

as a

sig

nifi

can

tde

clin

e in

the

pro

port

ion

of

the

grou

p th

reat

ened

wit

h a

knif

e or

gun

, wit

hin

eac

hgr

oup,

sin

ce b

asel

ine.


Tab

le 2

.Su

mm

ary

of S

ign

ifica

nt

(p<

0.0

5) S

hor

t-T

erm

Hea

lth

, Hea

lth

Car

e, a

nd

Hea

lth

Beh

avio

r O

utc

omes

A

ssoc

iate

d w

ith

Hou

sin

g M

obil

ity

Pol

icie

s (c

onti

nued

)

Sign

ifica

nt

amon

gw

hich

gro

up?

Age

Com

pari

son

Hea

lth

(In

ten

t to

Tre

atP

olic

ySo

urce

Gro

upM

ade

Indi

cato

rR

esul

ts R

epor

ted)

Inte

rpre

tati

on

Chi

cago

M

TO

R

osen

baum

an

dH

arri

s 20

01

An

ym

embe

rof

the

hous

e-ho

ld

Com

pari

son

of

base

line

vers

us f

ollo

w-u

p va

lue

for

eith

er t

he M

TO

or

Sect

ion

8gr

oup

(aut

hors

did

not

in

terv

iew

the

con

trol

gro

up

at f

ollo

w-u

p)

Bei

ng

beat

en o

r as

saul

ted

Bot

h M

TO

an

dSe

ctio

n 8

grou

ps

The

re w

as a

sig

nifi

can

tde

clin

e in

the

pro

port

ion

of

the

gro

up b

eate

n o

ras

saul

ted,

wit

hin

eac

h gr

oup,

sin

ce b

asel

ine.

An

ym

embe

rof

the

hous

e-ho

ld

Com

pari

son

of

base

line

vers

us f

ollo

w-u

p va

lue

for

eith

er t

he M

TO

or

Sect

ion

8gr

oup

(aut

hors

did

not

in

terv

iew

the

con

trol

gro

up

at f

ollo

w-u

p)

Bei

ng

stab

bed

or s

hot

Sect

ion

8gr

oup

The

re w

as a

sig

nifi

can

tde

clin

e in

the

pro

port

ion

of

the

Sect

ion

8 g

roup

sta

bbed

or

sho

t si

nce

bas

elin

e.

Yon

kers

B

rigg

s an

dYo

nke

rs F

amily

and

Com

mun

ity

Pro

ject

199

7

Adu

ltm

othe

rsM

over

s (t

hose

who

pa

rtic

ipat

ed in

the

pol

icy)

com

pare

d w

ith

stay

ers

(con

trol

gro

up o

f n

onpa

rtic

i-pa

nts

) at

fol

low

-up

Dep

ress

ion

sym

ptom

s du

rin

g th

e pa

st 3

0 da

ysM

over

s M

over

s re

port

ed f

ewer

sy

mpt

oms

than

sta

yers

.

Adu

ltm

othe

rsM

over

s (t

hose

who

pa

rtic

ipat

ed in

the

pol

icy)

com

pare

d w

ith

stay

ers

(con

trol

gro

up o

f n

onpa

rtic

i-pa

nts

) at

fol

low

-up

Pro

blem

dri

nki

ng

duri

ng

the

past

30

days

Mov

ers

Mov

ers

repo

rted

less

pro

blem

drin

kin

g th

an s

taye

rs.

Adu

ltm

othe

rsM

over

s (t

hose

who

pa

rtic

ipat

ed in

the

pol

icy)

com

pare

d w

ith

stay

ers

(con

trol

gro

up o

f n

onpa

rtic

i-pa

nts

) at

fol

low

-up

Mar

ijuan

a us

e du

rin

g th

epa

st 1

2 m

onth

sM

over

sM

over

s re

port

ed le

ss d

rug

use

than

sta

yers

.




Tab

le 2

.Su

mm

ary

of S

ign

ifica

nt

(p<

0.0

5) S

hor

t-T

erm

Hea

lth

, Hea

lth

Car

e, a

nd

Hea

lth

Beh

avio

r O

utc

omes

A

ssoc

iate

d w

ith

Hou

sin

g M

obil

ity

Pol

icie

s (c

onti

nued

)

Sign

ifica

nt

amon

gw

hich

gro

up?

Age

Com

pari

son

Hea

lth

(In

ten

t to

Tre

atP

olic

ySo

urce

Gro

upM

ade

Indi

cato

rR

esul

ts R

epor

ted)

Inte

rpre

tati

on

Yon

kers

B

rigg

s an

dYo

nke

rs F

amily

and

Com

mun

ity

Pro

ject

199

7

Adu

ltm

othe

rsM

over

s (t

hose

who

pa

rtic

ipat

ed in

the

pol

icy)

com

pare

d w

ith

stay

ers

(con

trol

gro

up o

f n

onpa

rtic

i-pa

nts

) at

fol

low

-up

Exp

erie

nce

of

a vi

olen

t or

trau

mat

ic e

ven

t du

rin

g th

epa

st 1

2 m

onth

s (v

icti

m o

fph

ysic

al a

ssau

lt/a

buse

, of

sexu

al a

ssau

lt/r

ape,

of

life-

thre

aten

ing

mug

gin

g, o

fla

rge

fire

or s

erio

us a

ccid

ent;

wit

nes

s to

ser

ious

ass

ault

or

viol

ent

deat

h; c

hild

or

spou

sedi

ed)

Mov

ers

Mov

ers

repo

rted

low

er le

vels

of v

iole

nt

or t

raum

atic

eve

nts

than

sta

yers

.

Yon

kers

Faut

h,

Lev

enth

al, a

nd

Bro

oks-

Gun

n20

02

Yout

h8

to 1

8M

over

s (t

hose

who

pa

rtic

ipat

ed in

the

pol

icy)

com

pare

d w

ith

stay

ers

(con

trol

gro

up o

f n

onpa

rtic

i-pa

nts

) at

fol

low

-up

Acc

ess

to c

igar

ette

s an

d al

coho

lM

over

sM

over

s ha

ve le

ss a

cces

s to

ciga

rett

es a

nd

alco

hol t

han

stay

ers

do.

Gau

trea

uxR

osen

baum

1994

Yout

hT

hose

who

mov

ed t

o th

e su

burb

s co

mpa

red

wit

h th

ose

who

mov

ed w

ithi

n t

he C

ity

ofC

hica

go, a

t fo

llow

-up

No

sign

ifica

nt

indi

cato

rsre

port

ed—

—

Ros

enba

um19

95Yo

uth

Tho

se w

ho m

oved

to

the

subu

rbs

com

pare

d w

ith

thos

ew

ho m

oved

wit

hin

the

Cit

y of

Chi

cago

, at

follo

w-u

p

No

sign

ifica

nt

indi

cato

rsre

port

ed—

—


Tab

le 2

.Su

mm

ary

of S

ign

ifica

nt

(p<

0.0

5) S

hor

t-T

erm

Hea

lth

, Hea

lth

Car

e, a

nd

Hea

lth

Beh

avio

r O

utc

omes

A

ssoc

iate

d w

ith

Hou

sin

g M

obil

ity

Pol

icie

s (c

onti

nued

)

Sign

ifica

nt

amon

gw

hich

gro

up?

Age

Com

pari

son

Hea

lth

(In

ten

t to

Tre

atP

olic

ySo

urce

Gro

upM

ade

Indi

cato

rR

esul

ts R

epor

ted)

Inte

rpre

tati

on

Gau

trea

uxR

osen

baum

an

dP

opki

n 1

990

Adu

lts

Tho

se w

ho m

oved

to

the

subu

rbs

com

pare

d w

ith

thos

ew

ho m

oved

wit

hin

the

Cit

y of

Chi

cago

, at

follo

w-u

p

Sati

sfac

tion

wit

h m

edic

alca

reC

ity

mov

ers

Cit

y m

over

s ex

peri

ence

dhi

gher

sat

isfa

ctio

n a

t fo

llow

-up

than

tho

se w

hom

oved

to

the

subu

rbs.

Adu

lts

Com

pari

son

of

follo

w-u

p w

ith

base

line,

for

eac

h gr

oup

(cit

yan

d su

burb

an m

over

s)

Sati

sfac

tion

wit

h m

edic

alca

reC

ity

mov

ers

Cit

y m

over

s ex

peri

ence

dhi

gher

sat

isfa

ctio

n a

t fo

llow

-up,

com

pare

d w

ith

base

line.

Per

off

et a

l.19

79A

dult

sT

hose

who

par

tici

pate

d in

the

Gau

trea

ux p

rogr

am (

both

cit

yan

d su

burb

an m

over

s)

com

pare

d w

ith

seve

ral

con

trol

gro

ups

(in

clud

ing

thos

e us

ing

Sect

ion

8 a

nd

thos

e in

pub

lic h

ousi

ng)

Acc

ess

to h

ealt

h cl

inic

/ho

spit

al is

eas

ier/

sam

e/m

ore

diffi

cult

com

pare

d w

ith

the

old

nei

ghbo

rhoo

d

Gau

trea

uxpa

rtic

ipan

tsA

cces

s is

mor

e di

fficu

lt f

orG

autr

eaux

par

tici

pan

ts t

han

for

con

trol

s in

Sec

tion

8 o

rel

igib

le n

onpa

rtic

ipan

ts.

(Not

e: N

o st

atis

tica

l tes

t w

asre

port

ed, s

o it

was

not

cle

arw

heth

er g

roup

s di

ffer

ed

sign

ifica

ntl

y.)

Adu

lts

Tho

se w

ho p

arti

cipa

ted

in t

heG

autr

eaux

pro

gram

(bo

th c

ity

and

subu

rban

mov

ers)

co

mpa

red

wit

h se

vera

l co

ntr

ol g

roup

s (i

ncl

udin

gth

ose

usin

g Se

ctio

n 8

an

dth

ose

in p

ublic

hou

sin

g)

Hea

lth

clin

ic/h

ospi

tal

serv

ices

are

bet

ter/

sam

e/w

orse

com

pare

d w

ith

the

old

nei

ghbo

rhoo

d

Gau

trea

uxpa

rtic

ipan

tsSe

rvic

es a

re n

ot t

he s

ame

(hig

her

repo

rts

of “

bett

er”

and

“wor

se”)

for

Gau

trea

uxpa

rtic

ipan

ts a

fter

the

mov

e,co

mpa

red

wit

h th

ose

in

Sect

ion

8 o

r el

igib

le

non

part

icip

ants

. (N

ote:

No

stat

isti

cal t

est

was

rep

orte

d,so

it w

as n

ot c

lear

whe

ther

grou

ps d

iffe

red

sign

ifica

ntl

y.)




Tab

le 2

.Su

mm

ary

of S

ign

ifica

nt

(p<

0.0

5) S

hor

t-T

erm

Hea

lth

, Hea

lth

Car

e, a

nd

Hea

lth

Beh

avio

r O

utc

omes

A

ssoc

iate

d w

ith

Hou

sin

g M

obil

ity

Pol

icie

s (c

onti

nued

)

Sign

ifica

nt

amon

gw

hich

gro

up?

Age

Com

pari

son

Hea

lth

(In

ten

t to

Tre

atP

olic

ySo

urce

Gro

upM

ade

Indi

cato

rR

esul

ts R

epor

ted)

Inte

rpre

tati

on

Gau

trea

uxP

erof

f et

al.

1979

Adu

lts

Tho

se w

ho p

arti

cipa

ted

in t

heG

autr

eaux

pro

gram

(bo

th c

ity

and

subu

rban

mov

ers)

co

mpa

red

wit

h se

vera

l co

ntr

ol g

roup

s (i

ncl

udin

gth

ose

usin

g Se

ctio

n 8

an

dth

ose

in p

ublic

hou

sin

g)

Get

tin

g to

hea

lth

clin

ic/

hosp

ital

is e

asy/

fair

lyea

sy/d

iffic

ult

Gau

trea

uxpa

rtic

ipan

tsA

cces

s is

mor

e di

fficu

lt f

orG

autr

eaux

par

tici

pan

ts t

han

for

thos

e in

Sec

tion

8 o

r el

igib

le n

onpa

rtic

ipan

ts.

(Not

e: N

o st

atis

tica

l tes

t w

asre

port

ed, s

o it

was

not

cle

arw

heth

er g

roup

s di

ffer

ed

sign

ifica

ntl

y.)

Sect

ion

8M

eyer

s et

al.

1995

Chi

ldre

nle

ss t

han

3 ye

ars

old

Tho

se r

ecei

vin

g Se

ctio

n 8

vouc

hers

com

pare

d w

ith

thos

e w

ho w

ere

on t

he

Sect

ion

8 w

aiti

ng

list,

at

one

poin

t in

tim

e (c

ross

-sec

tion

al)

Low

chi

ld g

row

th (

mea

n

z-sc

ore

of w

eigh

t/ag

e ba

sed

on N

atio

nal

Cen

ter

for

Hea

lth

Stat

isti

cs g

row

th

refe

ren

ce n

orm

s)

Sect

ion

8

part

icip

ants

Chi

ldre

n o

f ad

ults

rec

eivi

ng

the

Sect

ion

8 s

ubsi

dyre

port

ed t

o ha

ve h

ighe

r m

ean

wei

ght/

age

grow

th r

atio

s (b

ette

r gr

owth

) th

an c

hild

ren

of a

dult

s on

the

wai

tin

g lis

tfo

r th

e su

bsid

y.

Cin

cin

nat

i Sp

ecia

l M

obili

ty

Pro

gram

(S

MP

)

Fis

cher

199

1A

dult

sB

efor

e ve

rsus

aft

er t

he m

ove,

amon

g th

ose

in t

he S

MP

pr

ogra

m

Bei

ng

atta

cked

/per

son

alex

peri

ence

of

viol

ence

or

robb

ery

for

self,

chi

ld, o

rn

eigh

bor

SMP

pa

rtic

ipan

tsM

over

s re

port

a r

educ

tion

invi

olen

ce s

ince

the

mov

e.


Tab

le 2

.Su

mm

ary

of S

ign

ifica

nt

(p<

0.0

5) S

hor

t-T

erm

Hea

lth

, Hea

lth

Car

e, a

nd

Hea

lth

Beh

avio

r O

utc

omes

A

ssoc

iate

d w

ith

Hou

sin

g M

obil

ity

Pol

icie

s (c

onti

nued

)

Sign

ifica

nt

amon

gw

hich

gro

up?

Age

Com

pari

son

Hea

lth

(In

ten

t to

Tre

atP

olic

ySo

urce

Gro

upM

ade

Indi

cato

rR

esul

ts R

epor

ted)

Inte

rpre

tati

on

Not

es:I

n t

his

tabl

e, w

e ap

plie

d th

e co

nve

nti

onal

sig

nifi

can

ce v

alue

p<

0.0

5 fo

r in

clus

ion

of

heal

th-r

elat

ed o

utco

mes

ass

ocia

ted

wit

h ho

usin

g m

obili

typo

licy,

so

that

we

coul

d be

con

sist

ent

acro

ss s

tudi

es b

y ap

plyi

ng

the

sam

e cu

toff

val

ue. M

any

stud

ies

repo

rted

sev

eral

val

ues

mor

e st

rin

gen

t th

an 0

.05,

alth

ough

on

ly a

few

rep

orte

d si

gnifi

can

ce v

alue

s at

p <

0.1

0. H

ousi

ng

mob

ility

was

ass

ocia

ted

wit

h se

vera

l hea

lth-

rela

ted

outc

omes

at

a p

valu

ebe

twee

n 0

.05

and

0.10

“m

argi

nal

ly s

ign

ifica

nt,

” fo

r al

l the

out

com

es t

hat

follo

w. I

n B

osto

n, M

TO

tre

atm

ent

child

ren

(ag

e 6

to 1

5) e

xhib

ited

mar

gin

ally

few

er a

sthm

a at

tack

s re

quir

ing

med

ical

att

enti

on in

the

pas

t 6

mon

ths

and

mar

gin

ally

few

er p

erso

nal

cri

mes

(pe

rson

al t

hrea

ts, r

obbe

ries

, or

atta

cks)

(Kat

z, K

ling,

an

d L

iebm

an 2

001)

. Res

pon

den

ts in

the

Chi

cago

MT

O t

reat

men

t gr

oup

expe

rien

ced

a re

duct

ion

in b

ein

g st

abbe

d or

sho

t in

the

ir n

eigh

-bo

rhoo

d fr

om b

asel

ine

to f

ollo

w-u

p (R

osen

baum

an

d H

arri

s 20

01).

You

th m

over

s in

the

Yon

kers

pro

gram

exp

erie

nce

d lo

wer

odd

s of

bei

ng

assa

ulte

d(F

auth

, Lev

enth

al, a

nd

Bro

oks-

Gun

n 2

002)

. You

ng

child

ren

who

se f

amily

rec

eive

d th

e Se

ctio

n 8

vou

cher

had

hig

her

stan

dard

ized

wei

ght/

heig

htgr

owth

rat

ios

than

chi

ldre

n w

hose

fam

ilies

wer

e on

the

Sec

tion

8 w

aiti

ng

list

(Mey

ers

et a

l. 19

95).

Par

ents

in t

he N

ew Y

ork

MT

O t

reat

men

t gr

oup

expe

rien

ced

low

er le

vels

of

depr

essi

ve s

ympt

oms

than

con

trol

s; M

TO

tre

atm

ent

child

ren

age

d 8

to 1

3 ex

peri

ence

d lo

wer

leve

ls o

f an

xiou

s/de

pres

sed

sym

ptom

s, a

nd

both

tre

atm

ent

grou

ps a

ged

8 to

13

disp

laye

d lo

wer

leve

ls o

f de

pen

den

t pr

oble

ms

than

con

trol

s; S

ecti

on 8

you

th a

ged

8 to

18

repo

rted

low

er le

vels

of

depe

nde

ncy

an

d he

adst

ron

g pr

oble

ms

than

con

trol

s (L

even

thal

an

d B

rook

s-G

unn

200

3b).

The

re w

ere

mar

gin

ally

sig

nifi

can

t lo

wer

lev-

els

of p

robl

em b

ehav

iors

/men

tal h

ealt

h fo

r th

ese

New

Yor

k gr

oups

: MT

O t

reat

men

t bo

ys 8

to

18 “

unha

ppy,

sad

, or

depr

esse

d”; M

TO

tre

atm

ent

boys

“dep

end

too

muc

h on

oth

ers”

; Sec

tion

8 b

oys

“nee

d to

be

nea

r ad

ults

”; S

ecti

on 8

you

nge

r ch

ildre

n (

8 to

13)

“n

eed

to b

e n

ear

adul

ts”;

Sec

tion

8 c

hil-

dren

(8

to 1

8) “

depe

nd

too

muc

h on

oth

ers”

(L

even

thal

an

d B

rook

s-G

unn

200

3a).

Cin

cin

nat

i Sp

ecia

l M

obili

ty

Pro

gram

(S

MP

)

Fis

cher

199

1A

dult

sB

efor

e ve

rsus

aft

er t

he m

ove,

amon

g th

ose

in t

he S

MP

pr

ogra

m

Job

prov

ides

med

ical

car

ebe

nefi

ts (

amon

g th

ose

empl

oyed

)

SMP

pa

rtic

ipan

tsM

over

s re

port

tha

t th

ey a

rem

ore

likel

y to

hav

e be

nefi

tsaf

ter

they

mov

ed (

com

pare

dw

ith

befo

re).

(N

ote:

No

stat

isti

cal t

est

was

rep

orte

d,so

it w

as n

ot c

lear

whe

ther

grou

ps d

iffe

red

sign

ifica

ntl

y.)

Adu

lts

Tho

se in

the

SM

P p

rogr

amco

mpa

red

wit

h co

ntr

ols

inpu

blic

hou

sin

g

Job

prov

ides

med

ical

car

ebe

nefi

ts (

amon

g th

ose

empl

oyed

)

SMP

pa

rtic

ipan

tsT

hose

in t

he S

MP

pro

gram

wer

e m

ore

likel

y to

hav

em

edic

al c

are

ben

efits

tha

n t

heco

ntr

ol g

roup

in p

ublic

hou

s-in

g. (

Not

e: N

o st

atis

tica

l tes

tw

as r

epor

ted,

so

it w

as n

otcl

ear

whe

ther

gro

ups

diff

ered

sign

ifica

ntl

y.)




Second, if relocation was not assigned randomly, agents’ perceptions orclients’ preferences introduced selection bias. Although researchersasserted that participants were assigned to their neighborhood regard-less of their locational preference (Rosenbaum 1995), other evidencesuggests that program managers may have screened families theybelieved were better suited for moving to the suburbs (Goering 2003).Third, the pre/post study of children experienced substantial loss tofollow-up (40 percent attrition). In addition to limiting the power todetect meaningful effects (since the sample size was only 114 to beginwith), the widely cited beneficial educational and social outcomes accru-ing to suburban children may have resulted from selection if those who dropped out of the study differed systematically from those whoremained (Briggs 1997). Fourth, the Gautreaux program did not includea control group (both groups did move), so we cannot glean what mighthave happened to participants in the absence of the policy. We concludetherefore that the Gautreaux studies have methodological problems.

Section 8

The Section 8 program gives qualified low-income participants a rentalsubsidy for use in private apartment units. The program has been eval-uated over the past 25 years in many comprehensive studies, includingseveral federally funded policy experiments (Kennedy and Leger 1990;Peroff et al. 1979). Yet most of these studies did not include outcomesrelevant to our literature review. We did, however, find one study thatmeasured health. Meyers et al. (1995) found that children under agethree from families on the Section 8 waiting list had decreased growth(weight for age) compared with children in families receiving vouchers,indicating that the Section 8 policy might prevent undernutrition.However, this study is subject to substantial validity threats because of its convenience sampling method and cross-sectional design.

Yonkers scattered-site public housing

In addition to tenant mobility programs, scattered-site public housingprograms have been implemented as a result of court decrees to raciallydesegregate concentrated areas of poverty in central cities. In 1986, theDistrict Court ordered Yonkers, N.Y., to racially desegregate its publichousing, since 26 of the city’s 27 public housing units were located inthe same high-poverty, racially isolated, southwest quadrant of the city.The court ordered the construction of 200 units of scattered-site publichousing in low-poverty areas, 7 of which were built between 1990 and1993 (Briggs and the Yonkers Family and Community Project 1997;


Peterson and Williams 1995). Tenants were selected by lottery oncefamilies met the qualification criteria.

The Yonkers research team plans a 15-year study, including measure-ment of physical and mental health effects. The team also created acomparison group through a snowball sampling method, beginningwith referrals from movers—they gave the team the names of acquain-tances who stayed behind in the public housing projects.9 Although theteam found few demographic differences between the movers and thestayers (Briggs and the Yonkers Family and Community Project 1997),this does not rule out selection bias resulting from an invalid compari-son, if the controls might not have qualified or might not have wantedto apply for scattered-site public housing for reasons that would makethem not comparable to those who applied.

Briggs and the Yonkers evaluation team documented recent violentvictimization, depression and anxiety symptoms, and substance abuseamong mothers (Briggs, Darden, and Aidala 1999; Briggs and theYonkers Family and Community Project 1997). Compared with stayers,the mover group reported a lower prevalence of depression symptoms,of problem drinking, of marijuana use, and of violent or traumaticevents (Briggs and the Yonkers Family and Community Project 1997).Among youth, movers reported lower access to alcohol and cigarettesthan stayers did (Fauth, Leventhal, and Brooks-Gunn 2002).

The strength of this study resides in its valid and reliable health meas-ures and its prospective design with a concurrent comparison group.Yet the uncertain appropriateness of the control group (and the result-ing selection bias introduced through creation of a snowball controlgroup) remains the largest threat to the study’s validity.

Other litigation studies

Several cities throughout the country are conducting household mobil-ity programs like the Gautreaux program or unit-based mobility pro-grams like the Yonkers scattered-site public housing, most initiated byracial discrimination litigation. Although these studies are rarely evalu-ated,10 we did find that one qualified for inclusion in our review. The



9 Originally, the Yonkers intervention would have exploited an experimental design,since a lottery was used, but there were too few lottery losers to create a large enoughcontrol group.

10 Some housing mobility programs, including those in Cleveland, Dallas, and Durham(N.C.), have been formally evaluated (Peterson and Williams 1995), but they are not


evaluation of the Cincinnati Special Mobility Program measured violentvictimization and health care, although the study is methodologicallyvery weak. Fischer (1991) compared the intervention group before andafter the policy and also compared the intervention group with publichousing controls at follow-up. He found that moving out of concen-trated public housing was associated with fewer attacks or robberiesand with the receipt of employer-supplied health care benefits. How-ever, this was the weakest of all the studies included in our review. Thecontrol group was likely not comparable with the intervention group,and we could not assess the amount of bias, since the author did notreport any multivariate regressions or other analyses to control for con-founding. Moreover, he did not include any statistical tests of differenceor table of numerical results.

MTO

The federal government has implemented a randomized housing mobil-ity policy experiment in five U.S. metropolitan areas with the MTO pro-gram. Sponsored by HUD and begun in 1994, the experiment randomlyassigned selected participants from central-city public housing to one ofthree groups:

1. The treatment group (also referred to as the experimental or MTOgroup) was offered both a Section 8 housing voucher that could beredeemed only in a low-poverty neighborhood and housingcounseling.

2. The Section 8 group was offered a geographically unrestricted Sec-tion 8 housing voucher.

3. The in-place control group did not receive a voucher, but remainedeligible for public housing.

All of the participants consisted of low-income families, and most wereracial minorities (Goering et al. 1999). This review covers results fromthe first follow-up evaluation fielded in each city, which took placeapproximately two to three years after randomization and was releasedin 2002. The results from a more recent interim survey fielded in2001–2002 were released while this article was under review.


considered natural experiments because of their design and implementation. As aresult, conclusions drawn from favorable outcomes cannot be attributed to the policyitself unless a sound comparison group is found. Although we reviewed these studiesfor inclusion in this review, they did not qualify.

Two of the MTO sites have reported mental and physical health out-comes (Boston and New York), three have reported violent crime vic-timization (Boston, New York, and Chicago), two have reported medicalcare outcomes (Los Angeles and Boston), and one has reported health-behavior outcomes (New York) (Hanratty, McLanahan, and Pettit 2003;Katz, Kling, and Liebman 2001; Leventhal and Brooks-Gunn 2003a,2003b; Rosenbaum and Harris 2001).

As a result of the Boston MTO policy, children in treatment group fami-lies have experienced fewer injuries or accidents, and boys in the treat-ment and Section 8 groups both displayed fewer reported behaviorproblems than controls. Investigators reported a marginal reduction in the treatment of children’s asthma attacks and in personal crimevictimization (attacks, threats, or robberies). Boston treatment andSection 8 parents reported better overall health, as well as higher pro-portions of being calm and peaceful (Katz, Kling, and Liebman 2001).

Leventhal and Brooks-Gunn (2003a) provide evidence that moving tolow-poverty neighborhoods in New York resulted in large improve-ments in mental health for boys, especially among the treatment group.As table 2 illustrates, compared with controls, the New York experi-mental and Section 8 children, as well as subgroups (usually boys orthe younger children), reported significantly improved mental health,which manifested as fewer problem behaviors such as anxiety ordepression, dependency, fear, and the need to be near adults. In somecases, Section 8 children experienced lower levels of problem behavior(being headstrong, unhappy, sad, or depressed and depending too muchon others) than controls, when the experimental children did not (Lev-enthal and Brooks-Gunn 2003a, 2003b).

The New York MTO study found different significant effects for teenproblem health behaviors (using cigarettes or alcohol) by gender. Unex-pectedly, adolescent girls in the experimental group were significantlymore likely to have used alcohol in the past year than in-place controls,but teen boys in the experimental group reported fewer peers using cig-arettes. New York treatment group parents were significantly less likelyto report distress symptoms than controls and marginally less likely toreport depressive symptoms (Leventhal and Brooks-Gunn 2003a,2003b).

In terms of medical care, neither of the two sites that measured med-ical care found an effect (Hanratty, McLanahan, and Pettit 2003; Katz,Kling, and Liebman 2001). None of the three sites that measured expe-rience of violent crime found significant results between experimentaland control groups (p < 0.05), although the New York and Chicago




sites both found a reduction in crime from baseline to follow-up (Leven-thal and Brooks-Gunn 2003a; Rosenbaum and Harris 2001). This sug-gests that a secular decrease in crime may have occurred.

The design of the Chicago follow-up survey (investigators re-interviewed movers only, with no control group comparison) suggeststhat inferences from this site may be biased. The study may have expe-rienced secular effects without control group measurement at follow-up; for instance, if the participants were affected by an external factornot derived from the MTO study, such as economic changes, the studywould incorrectly infer an effect caused by MTO (Orr 1999).

While this article was under review, HUD released the MTO InterimImpacts Evaluation (Orr et al. 2003), which included detail of healthoutcomes resulting from relocation four to seven years after randomiza-tion. In that report, investigators did not report site-specific findings.Adults in the experimental and Section 8 groups experienced signifi-cantly lower obesity (measured as body mass index) than the controlgroup, and adults in the experimental group had a lower prevalence ofmental health problems (psychological distress and depression). Noeffects were observed among adults for asthma, self-reported health,high blood pressure, smoking, drinking alcohol, or activities of dailyliving. Among children, girls in both experimental groups reportedimproved mental health. Boys aged 12 to 19 in the Section 8 groupreported more injuries requiring medical attention. Aside from injuries,no physical health effects were observed for children or teens, andthere were no significant group differences in access to health care (Orr et al. 2003).

The MTO interim evaluation also measured risky health behaviorsamong adolescents. Teen girls in the experimental group reportedlower lifetime use of marijuana and of smoking tobacco. Teen boys inboth treatment groups unexpectedly reported significantly higher ratesof smoking tobacco than controls (Orr et al. 2003).

The health findings of the 2003 MTO interim report are somewhat con-sistent with the first follow-up study, indicating mainly mental healthbenefits, in addition to the lower levels of adult obesity, which had notbeen previously measured, among the experimental groups. This policymay differentially affect children by gender, as suggested by Leventhaland Brooks-Gunn (2003a), where girls’ mental health may benefit,whereas boys’ health may be harmed by smoking uptake and injuries inadolescence. The interim findings are also consistent for health careaccess, since neither evaluation found group differences.


Although the randomized design of the MTO study strengthens thecausal inferences drawn from the research, some threats to validityexist. If one analyzes only those who leased up (took advantage of thevoucher), MTO individuals are highly selected. The low lease-up ratescan be conceptualized as low treatment compliance and may reduce thestudy’s power to detect meaningful differences among groups.11 How-ever, researchers can use and have used an intent-to-treat (ITT) analy-sis to keep randomized groups intact in the analysis. If conductedcorrectly, ITT analysis should treat noncompliant subjects, in this casethose who were selected for the MTO experimental groups but did notuse the voucher, as treatment failures, since noncompliance may beassociated with “side effects” of the intervention. For instance, as sug-gested earlier, losing a regular source of medical care could be a signifi-cant side effect of moving to a new neighborhood. But if all originalstudy members are identified and if the effects of the treatment can beassumed as positive, then when analyzed as ITT, the effects should beunbiased by selection for the average effect on the group regardless ofcompliance (Greenland 2000). But the effect estimated by ITT is theeffect of being offered a housing voucher, not receiving one.

Furthermore, MTO researchers have conducted instrumental variableanalysis (IVA) that strengthens the conclusions that the policy mayimprove health. The authors estimated the effect of the treatment-on-treated (TOT) by using this technique, which corrects for bias thatwould usually arise if one applied an ITT analysis or directly comparedtreated with untreated individuals in a randomized trial. The techniqueuses a variable, “the instrument,” which is associated with the predic-tor and has no other influence on the outcome except that mediated bythe predictor, to estimate the causal effect of the treatment on thosewho were randomly assigned to treatment and actually received it. TheTOT analyses found significant results for some health outcomes thatwere not significant in the ITT analyses, suggesting that, as expected,



11 The experimental group actually used the low-poverty neighborhood voucher at a lowrate—48 percent across all five sites; see table 5 in Goering et al. (1999). Notably, thelease-up rate—hence selection into low-poverty neighborhoods—may be related tohealth. After conducting a multiple regression analysis predicting lease-up rates withbaseline information across all five sites, Shroder (2002) reported that families receiv-ing disability payments (supplemental security income/Social Security Disability Insur-ance/Social Security survivor benefits) were significantly less likely to lease up or takeadvantage of the housing voucher to move. Further, interviews with MTO mobilitycounselors revealed that many families that failed to use the voucher to move did sobecause they did not want to move farther away from the medical care that they ortheir children were receiving. These findings stimulated researchers to observe that“health care is perhaps the most important place-based service that these families relyon in their neighborhoods” (Kling, Liebman, and Katz 2001, 22). Thus, it is plausiblethat healthier individuals may have been selected into using the voucher.


the policy had stronger effects on those who actually relocated to low-poverty neighborhoods (Greenland 2000).12

The MTO research findings may also be biased by loss to follow-up.This threat presents itself especially in the New York study, where 31 percent of the participants were lost, and in Chicago, where only 51 percent—none of them controls—were interviewed at follow-up.

Although randomized trials eliminate many serious threats to internalvalidity, several remain. The most relevant for MTO are likely compen-satory rivalry and compensatory equalization, both of which occurwhen the treatment is perceived as desirable (Cook and Campbell1979). Compensatory rivalry, also called the John Henry effect, couldhave surfaced if MTO control group members felt cheated by not hav-ing received the treatment and worked harder than they would have inthe absence of this policy to seek some sort of substitute for the treat-ment, such as neighborhood mobility or homeownership. With compen-satory equalization, an administrative body that deems control statusless desirable than treatment steps in to offer a comparable treatmentas a sort of consolation. This could have happened in MTO, forinstance, if housing authorities provided vouchers to controls in a moreexpeditious manner than they would have without knowledge of thecontrol’s participation in the experiment. In both cases, the bias resultsin underestimating the treatment effect, because the controls are per-forming artificially better than they would have without the policy.

Assessment of the empirical evidence

The strongest studies had randomized designs using ITT analysesand/or valid and reliable measures of health. Our review methods gavethe Boston MTO study (Katz, Kling, and Liebman 2001) the strongestmethodological score because it was randomized and analyzed as ITT,used valid and reliable health measures, experienced very low attrition,


12 In general, IVA effect estimates for a trial will be further from the null than ITTestimates because many of those assigned to the treatment may not actually receive it. These individuals presumably dilute the estimated effect of the treatment in an ITTanalysis. The bias in the ITT effect estimates is proportional to the amount of noncom-pliance, so in experiments with substantial noncompliance, such as MTO, the differencebetween ITT and IVA estimates may be large. There is some criticism of using IVA esti-mates in this case, because people who did not adhere to the treatment/policy may bethose who could somehow have been harmed by it. For example, in MTO, it is plausiblethat people who did not use the vouchers were those who were afraid to lose their regu-lar source of medical care. In this case, the causal effect of the treatment on the treatedis not the same as the average causal effect of the treatment on the whole population.

used instrumental analysis to examine effects on movers, and exhibitedfew remaining threats to validity (table 1). It was the only study toattain “good” methodological rigor according to the Briss method (Brisset al. 2000). The New York MTO mental health study (Leventhal andBrooks-Gunn 2003b) ranked next highest because of its randomizeddesign, ITT and IVA, and valid and reliable health measures. Fourother studies fall in this same fairly strong category: the other NewYork MTO study (Leventhal and Brooks-Gunn 2003a), the two Yonkersstudies (Briggs and the Yonkers Family and Community Project 1997;Fauth, Leventhal, and Brooks-Gunn 2002), and one Gautreaux study(Rosenbaum and Popkin 1990). Despite its nonrandomized design, theYonkers study scored high because of its very low attrition rate; itsnumerous valid and reliable health measures; and its thorough policy,population, and methodological discussions (Briggs and the YonkersFamily and Community Project 1997).

We evaluated the rest of the studies as being of limited methodologicalstrength, because of five or more serious threats to validity. Falling inthis category were the Chicago (Rosenbaum and Harris 2001) and LosAngeles (Hanratty, McLanahan, and Pettit 2003) MTO studies, the Sec-tion 8 study assessing child growth/undernutrition (Meyers et al. 1995),and the original Gautreaux evaluation assessing health care satisfac-tion (Peroff et al. 1979). The Chicago MTO study experienced substan-tial attrition (50 percent) and did not interview controls at follow-up, sothe investigators could not conduct an ITT analysis. The Los Angelesstudy was ranked as limited because it was not clear how reliable orvalid the health care outcome was, and the researchers were notexplicit about their statistical and analytical methods. Two Gautreauxarticles using the same study data (Rosenbaum 1994, 1995) and theCincinnati litigation program (Fischer 1991) finished last in our analy-sis. These Gautreaux studies experienced substantial loss to follow-up(40 percent attrition), the author was not explicit about statisticalmethods used to account for confounding, and the validity and reliabil-ity of the victimization measure were not addressed. The Fischer(1991) study was methodologically weak because the author was notexplicit about the statistical methods used to control for confounding,and he did not present tests of statistical significance, numeric results,or validity/reliability of measures.

Overall, the evidence suggests that both tenant- and unit-based hous-ing mobility policies may contribute to improving child, adolescent, andadult health and health behaviors. However, this preliminary evidencecomes from a small number of studies, many of which have method-ological limitations associated with the lack of an experimental orquasi-experimental design.




Further, although there is some evidence that relocation to betterneighborhoods could lead to improved health among public housing res-idents, none of the studies reviewed explicitly tested the mechanismsthrough which housing mobility policies could impact health. As Briggsnotes, the literature on the neighborhood effects of housing mobilityprograms has not elucidated the mechanisms by which the observedoutcomes came to be, or the “how” of the processes involved (1997).Some mechanisms are being tested in the Yonkers study, but accordingto our review, very few of these have been tested thus far in the MTOexperiment. However, the potential exists to include such hypotheses inthe data collection for the 10-year MTO follow-up.

Discussion

The housing mobility policies described in this article were conceived as racial and economic desegregation policies. Although they were notintended to affect health, it appears that they may have. Indeed, Leven-thal and Brooks-Gunn claim that “the most significant benefits of theMTO program were noneconomic” (2003b, 1580). From a public healthperspective, the most important implication of the evidence on housingmobility is that housing policy has the potential to improve individualhealth. Mobility and other housing policies should continue to be evalu-ated for their impact on health. If the long-term benefits are positive,as suggested thus far by this limited research on housing mobility, suchpolicies may promise to improve individual health and quality of life, aswell as the health of the population as a whole. Because housing mobil-ity policies target very low-income families, these policies may also havethe potential to reduce health disparities by improving the health ofthese disadvantaged groups.

Generalizability

The available evidence suggests that housing mobility programs couldimprove health. However, there is strong empirical evidence from onlytwo MTO sites, Boston and perhaps New York, and from the Yonkersstudies. These results may not be generalizable to other cities with dif-ferent demographic and social characteristics. Moreover, as discussedearlier, MTO results could be generalizable by design only to familiesthat volunteer to leave public housing. At least in the MTO demonstra-tion, those families appear to be more disadvantaged than the generalpublic housing population and may have had strong, specific motives tomove. As summarized by Shroder (2002), “[W]ith this experiment wewill have a chance to evaluate without bias the benefit of moving out of


high poverty to those who want to move and are willing to keep search-ing until they succeed” (337, emphasis added). Accordingly, it is possiblethat MTO results overestimate the possible effects on health that theintervention could have on families that already are participating in theSection 8 program or families that could be transferred from publichousing to Section 8 without volunteering for it.

Effects on health outcomes

Before MTO, much of the evidence on health and housing mobility had little scientific weight because of the lack of experimental designsand the associated threat of selection bias. The evidence from MTO isstronger because of its experimental design, but since this policydemonstration was not designed to measure the effects of housingmobility on health, the range of health outcomes is limited for the mostpart to a few self-reported measures. Since none of the sites at the firstfollow-up study collected any physiological health measures or bloodsamples, there is no evidence of change in biological markers. However,the apparently positive effect on self-reported outcomes suggests that,at a minimum, psychological well-being improved.

Additionally, although the research teams at the various sites havespeculated about the possible mechanisms through which MTO mayimprove health, the demonstration was not designed to explore specificmechanisms. For example, Katz, Kling, and Liebman (2001) have pro-posed that the marginally significant reduction in the number of self-reported asthma attacks among children in experimental householdsmay be linked to the better air quality and the decreased stress, crowd-ing, lack of heat, or exposure to allergens from vermin that are associ-ated with living in better neighborhoods. Other questions forexploration include whether the housing units in the new neighbor-hoods contain less mold caused by aging or leaky pipes or whether win-dows are less likely to be nailed or painted shut, thus preventing airflow that could disperse indoor allergens or toxins that trigger asthma.In other words, the marginal reduction in asthma rates could be due toa neighborhood factor (safety, air quality) or a household factor (mold,allergens), or both. Given the present MTO data, it is not possible toelucidate the relative weight of these mechanisms.

Segregation, neighborhoods, and health

Although the MTO results may provide some indirect support for socialepidemiologic evidence that racial and economic residential segregation




could be positively associated with mortality and morbidity rates, theMTO’s scope is simply too small to alter racial segregation or povertyconcentration patterns at the metropolitan level. For example, the sizeof the black population in the Boston metropolitan area in 1990 wasapproximately 198,000. In the same year, the dissimilarity index forblacks in this metropolitan area was 69.6, indicating that nearly 70 per-cent of the black population—more than 138,000 individuals—wouldhave had to move to different neighborhoods to achieve an even distri-bution with respect to the non-Hispanic white population (Lewis Mum-ford Center for Comparative Urban and Regional Research 2001). Bycontrast, the MTO demonstration in Boston involved only about 135(mostly black) families.

In relation to the epidemiologic evidence on neighborhood effects onhealth, the randomization of study participants to the control group,the regular Section 8 group, or the MTO group means that it is safe toassume that any observed effects on health are not due to individual-level characteristics. That is, there is no threat of selection bias. How-ever, it is not clear that effects can be attributed to neighborhoodenvironment.

Future research and data

Conceptual frameworks. As discussed earlier, epidemiologic research onhousing and health has largely lacked a conceptual foundation. Hous-ing policy research constitutes a unique opportunity to explore variousneighborhood effects on health through experimental designs. Concep-tual frameworks should outline plausible mechanisms through whichneighborhood conditions could affect specific health outcomes. Thedevelopment of such conceptual frameworks will require multidiscipli-nary research teams that include housing policy and public healthexperts (Srinivasan, O’Fallon, and Dearry 2003). Public health researchcan help inform the selection of health outcomes that are more sensi-tive to certain neighborhood conditions at certain times in the lifecourse. For instance, Macintyre and Ellaway (2003) discuss the ideathat fear of crime or violence may influence women and elders morethan it does men or younger people, affecting mental but not physicalhealth. Conversely, children may be more sensitive than adults to mate-rial conditions like damp, which affects respiratory disease (Macintyreand Ellaway 2003). In the latest MTO report (Orr et al. 2003), theinvestigators found it difficult to interpret the lower rates of obesityamong experimental adults. However, public health research suggeststhat disadvantaged neighborhoods are associated with reduced access tohealthful foods because fewer grocery stores are available (Morland


et al. 2002) and with unsafe or “low-walkability” conditions that maynot be conducive to physical activity (Saelens et al. 2003). In turn,those factors may be linked to higher rates of obesity. Alternatively, the MTO interim finding that experimental families eat more mealstogether as a family may explain part of the obesity effect; for example,parents may be preparing healthier meals more frequently.

Since housing mobility programs have been introduced in part to helpremedy racial segregation, which originates in both institutional andinterpersonal racial discrimination, it is important to address discrimi-nation in the conceptual framework and variables measured in futurehousing mobility studies. The MTO team has taken a first step towardthis end by including an item in the interim report dealing withwhether participants faced housing discrimination during their lasthousing search (Orr et al. 2003). Yet the authors have not presentedthis information within a racial context. The MTO sites varied substan-tially in the racial makeup of their population, and different metropoli-tan areas display markedly different rates of housing discrimination(Acevedo-Garcia, Osypuk, and Krimgold 2003; Gabriel 1996), as doneighborhoods with different racial composition (Gabriel 1996). More-over, racial discrimination is experienced in many forms, across variouslife domains in addition to the housing domain, and many of these havebeen shown to affect health (Gee 2002; Krieger and Sidney 1996;Williams and Neighbors 2001; Williams, Neighbors, and Jackson 2003).Racial disparities in health are well-established (Williams 2001;Williams et al. 1997), but rarely do health investigators measure thesocial process of racism—that is, the mechanism by which race is mostlikely associated with health. Including a more comprehensive measureof racial discrimination, presenting racism experience by MTO site andby racial group, and analyzing data on discrimination by the racialcomposition of the census tract may be invaluable to better under-standing both housing-specific and non-housing-specific discriminationand its effects on health. Also, the Section 8 program faces importantchallenges, such as the negative reactions that some communities haveexpressed against the influx of families on housing assistance (Briggs1997, 1998; Rosenbaum 1995). Therefore, it would be important toassess the potentially negative effects—for example, stress and depres-sion—on the health of families moving into hostile communities.

Additionally, future research should explore whether improvements inhealth outcomes might eventually lead to long-term improvements ineconomic outcomes. For example, parents who receive public assistanceare more likely to have children with such chronic health conditions asasthma, which in turn limits their ability to work (Heymann and Earle1999). If housing mobility policies can effectively improve children’shealth, parents may be more likely to succeed in the labor force.




Multilevel data and analytic methods. Future research on housingmobility should include a larger sample of individuals, neighborhoods of origin, receiving neighborhoods, and metropolitan areas, as well asmore comprehensive data on unit characteristics. Many of the researchquestions that future studies should address are inherently multileveland thus will require multilevel analytic methods (Subramanian, Jones,and Duncan 2003). For instance, given that improvements in neighbor-hood quality that result from mobility policies are conceivably accompa-nied by improvements in the quality of housing units, it is important to determine whether mobility policies result in improved health out-comes because of housing unit effects, neighborhood effects, or both.Similarly, improvements in housing and neighborhood quality couldhave differential effects on individuals with various demographic andsocioeconomic characteristics. The health effects of improving theneighborhood environment may be greater for individuals/householdswith fewer coping resources, such as those with low socioeconomic sta-tus or single-headed households.

Multilevel methods may also allow us to assess whether there is signifi-cant variation in health outcomes across neighborhoods, includingacross experimental neighborhoods, and, if so, which factors couldexplain this variation. For instance, health effects may be greater if thereceiving neighborhoods have a low poverty rate (below 10 percent), asopposed to being simply less disadvantaged than the neighborhoods oforigin.

Further, to the extent that housing mobility demonstrations are con-ducted in a large cross-section of metropolitan areas, multilevel meth-ods may allow us to determine whether health outcomes are affected bymetropolitan area characteristics after we control for neighborhood orhousehold characteristics. For example, are health effects greater ifmobility policies are implemented in metropolitan areas with higherlevels of racial or economic segregation, since movers may have more togain than their counterparts in areas with lower levels of segregation?

Qualitative research methods. Our review focused on studies with anexperimental design, which social epidemiologists regard as a valuableyet often not feasible method of evaluating neighborhood effects (Oakes2004). The 2001 qualitative study of Boston’s MTO (Kling, Liebman,and Katz 2001), which brought the importance of health to the atten-tion of investigators, and the 2002 qualitative study of MTO (Popkin,Harris, and Cunningham 2002), as well as the qualitative neighborhoodand health studies of Boston by Gans and of Chicago by Raudenbushand Sampson (Kawachi and Berkman 2003a) suggest the potential ofsuch research methods for understanding the pathways through which


neighborhood conditions might affect health. Qualitative investigationprovides a historical context and also communicates the reciprocal anddynamic process of how people shape their neighborhoods (Macintyreand Ellaway 2003). Qualitative research methods could also be used toinform the design of the quantitative data collection and help explainthe results from quantitative analyses (Popkin, Harris, and Cunning-ham 2002).

Better health, housing, and neighborhood data. Future studies shouldalso include better measurement of health outcomes, including baselineand follow-up measurements and triangulation of methods—biologicalmeasures, self-reported health measures, and validated scales (forexample, depression). Given proper ethical and confidentiality protec-tions, it would also be important to link participant data to healthinsurance claims or medical records, especially for diagnostic-specificinformation. Such linking may be feasible for policies such as MTO ifsubstantial proportions of participants were enrolled in Medicaid orstate children’s health insurance programs. Linking to administrativehealth data seems feasible, since the MTO follow-up study plans to link to other administrative data systems, such as welfare, arrestrecords, and schools (Orr et al. 2003). In addition to being uniformacross all the study sites (the original survey instruments were not),the 2001–2002 follow-up MTO household survey prepared by AbtAssociates, Inc., partially addressed some of these issues. Namely, itincluded detailed sections on injuries, asthma symptoms, body massindex, and blood pressure measurement. Yet additional measures mayinclude other health outcomes that could be associated with neighbor-hood conditions, for example, intimate partner violence and infectiousdiseases such as tuberculosis, HIV, and sexually transmitted diseases(Acevedo-Garcia 2000; Fullilove 2003; Kawachi and Berkman 2003b;O’Campo et al. 1995).

Biomarkers can be employed to test specific physiologic mechanisms ofhousing or neighborhood effects on health. For instance, the physiologi-cal stress response can be tested by measuring the stress hormone cor-tisol in saliva samples. Several biomarkers are available for asthma(Christiani 1996; Loucks 2003) (i.e., inflammatory markers), as well asfor immune function (Kiecolt-Glaser et al. 2002), both of which can beaffected by stress. If we hypothesize that housing mobility affects nutri-tional intake in the form of increased access to higher-quality super-markets in low-poverty neighborhoods or reduced access to fast food inhigh-poverty neighborhoods, we can measure serum fasting glucose orserum cholesterol. Biomarkers are useful for valid measurement of out-comes that may be influenced by cognitive status or by social desirabil-ity (such as smoking, intake of fruits and vegetables, or weight) or for




intermediate outcomes for chronic diseases that take a long time todevelop (such as blood pressure or cholesterol for heart disease) (DeGruttola et al. 2001; Loucks 2003).

To date, few social epidemiologic studies have progressed beyonddemonstrating that certain structural characteristics of neighborhoods,such as poverty or disadvantage, appear to exert a contextual influenceon health after taking into account compositional characteristics suchas the socioeconomic status of residents. Few attempts have been madeto characterize and measure the social, physical, and service aspects ofneighborhood environments and to test the causal mechanisms under-lying the influences of these aspects of neighborhood environments onhealth outcomes. Social epidemiologists concur that future research onneighborhood effects should incorporate the development of valid, reli-able measures and scales of various aspects of neighborhood environ-ments (Morland et al. 2002; O’Campo 2003; Saelens et al. 2003). Forexample, assessing whether a neighborhood is conducive to physicalactivity could involve gathering data on sidewalks, bicycle lanes,shared-used paths, work sites, green ways, or recreational facilities(Librett, Yore, and Schmid 2003). Recent work by the Urban Instituteunder the National Neighborhood Indicator Partnership project (Pettitet al. 2003) constitutes an example of the types of specific neighborhoodindicators that could be useful in future health research on neighbor-hood effects.

Moreover, subjective and objective neighborhood data may tap differenthealth mechanisms. For instance, subjective rating of crime may bemore inhibiting for exercise than objective crime rates; alternatively,subjective reports may introduce reporting bias. For instance, someonewith respiratory problems reports that his or her home is dampbecause this person believes it contributes to his or her illness. Further,subjective reports may empirically demonstrate less variation thanobjective assessments (Kawachi and Berkman 2003a; Macintyre andEllaway 2003).

Evaluation of other housing policies. Given that they target low-incomeand very low-income families, housing vouchers and mobility initiativesshould continue to be evaluated for their impact on the health of disad-vantaged groups and their contribution to reducing health disparities.Housing vouchers are becoming the dominant strategy for providinghousing assistance, with the potential for reaching approximately 5.4 million households—the estimated worst-case housing needs popu-lation (HUD, Office of Policy Development and Research 2000). As apolicy demonstration, MTO has been very limited in scope (about 1,800experimental families), and its political viability as a large-scale pro-gram is not certain. Therefore, it will be important to further evaluate


the health impact of Section 8 and some of its politically viable varia-tions—such as regional, as opposed to local, administration of the pro-gram, which has been proposed as a way to enhance its housingmobility potential (Katz and Turner 2001; Turner 1998).

Although the evidence from the housing mobility studies is not defini-tive, it will be important to put the magnitude of these results into per-spective. How big are the health effects of housing policy comparedwith the effects measured for explicit health interventions? Forinstance, are the effects on asthma from housing mobility large com-pared with the effects from health education programs specificallytargeting asthma? A closely related issue is the approximate cost ofhousing mobility programs. We should take the relative effects of hous-ing interventions and explicit health interventions on health outcomesin the context of their relative costs. A crude comparison shows thatthe total annual HUD budget for rental assistance is approximately$17.5 billion, compared with more than $140 billion annually for Med-icaid. Incorporating more specific cost-benefit considerations into ouranalyses could inform considerations of the specific health outcomes ofhousing mobility policies.

During the 1990s, under the HOPE (Home Ownership for PeopleEverywhere) VI program, the federal government changed its housingpolicy toward low-income households and moved away from project-based assistance (public housing projects) toward an increased use ofhousing vouchers and mixed-income housing developments (Buron etal. 2002; Sard 2000). The nature and scale of such policy changes war-rant an examination of their possible health effects, especially on fami-lies that have been displaced and unable to find affordable housing. Ina tracking study of HOPE VI, the Urban Institute has identified healthproblems as a major issue for former residents of distressed publichousing (Buron et al. 2002).

Earlier, we identified promotion of homeownership among low-incomehouseholds and antidiscrimination policy as two strategies that, likehousing mobility policy, could potentially mitigate health disparities. Inthe future, health research may identify quasi-experimental or experi-mental policy demonstrations in those two areas, where health out-comes could be incorporated into the evaluation. Further, otherimportant housing policies, such as federal tax deductions to promotehomeownership, policies to promote homeownership among low-incomehouseholds, and fair housing policies, should be evaluated for theireffect on health outcomes. For instance, in terms of health impactamong low-income groups, does homeownership have a larger positivehealth influence than housing mobility?




The evidence from Yonkers and MTO suggests that housing mobilitypolicy could have positive and, ironically, largely unintended effects onhealth. HUD has a strong, long-standing tradition aimed at conductinghousing policy demonstrations with an experimental design (Shroder2000). Therefore, there is already a good infrastructure for carrying outhousing policy demonstrations with attention to health outcomes. Intheir vision for housing policy in the new millennium, policy makersbelieve that “housing policy must be linked to other social policies”(Wachter 2000, 1). Integrating better health as an explicitly desirableoutcome of housing policy could help fulfill this vision.

Authors

Dolores Acevedo-Garcia is an Assistant Professor in the Department of Society, HumanDevelopment, and Health at the Harvard School of Public Health. Theresa L. Osypukis a doctoral student in the Department of Society, Human Development, and Health atthe Harvard School of Public Health. At the time of this writing, Rebecca E. Werbelwas a graduate student in the Department of Society, Human Development, and Healthat the Harvard School of Public Health. Ellen R. Meara is an Assistant Professor in theDepartment of Health Care Policy at Harvard Medical School. David M. Cutler is a Pro-fessor in the Department of Economics at Harvard University. Lisa F. Berkman is theThomas D. Cabot Professor of Public Policy in the Department of Society, HumanDevelopment, and Health at the Harvard School of Public Health.

References

Acevedo-Garcia, Dolores. 2000. Residential Segregation and the Epidemiology of Infec-tious Diseases. Social Science and Medicine 51(8):1143–61.

Acevedo-Garcia, Dolores, and Kimberly A. Lochner. 2003. Residential Segregation andHealth. In Neighborhoods and Health, ed. Ichiro Kawachi and Lisa F. Berkman,265–87. New York: Oxford University Press.

Acevedo-Garcia, Dolores, Kimberly A. Lochner, Theresa L. Osypuk, and S.V. Subraman-ian. 2003. Future Directions in Residential Segregation and Health Research: A Multi-level Approach. American Journal of Public Health 93(2):215–21.

Acevedo-Garcia, Dolores, Theresa L. Osypuk, and Barbara Krimgold. 2003. Metropoli-tan Report Cards: How Well Are Different Urban Areas Providing Quality of Life forAmericans of Color? Presented at the Color Lines Conference, sponsored by the CivilRights Project at Harvard University, Cambridge, MA. August 29–September 1.

Allen, Chris. 2000. On the “Physiological Dope” Problematic in Housing and IllnessResearch: Towards a Critical Realism of Home and Health. Housing, Theory, and Soci-ety 17:49–67.


Altshuler, Alan, William Morrill, Harold Wolman, Faith Mitchell, and the Committee onImproving the Future of U.S. Cities through Improved Metropolitan Area Governance,eds. 1999. Governance and Opportunity in Metropolitan America. Washington, DC:National Academy Press.

Anderson, Laurie M., Carolynne Shinn, Joseph St. Charles, Mindy T. Fullilove, SusanC. Scrimshaw, Jonathan E. Fielding, Jacques Normand, Ruth Sanchez-Way, ToddRichardson, and the Centers for Disease Control and Prevention. 2002. CommunityInterventions to Promote Healthy Social Environments, Early Childhood Development,and Family Housing: A Report on Recommendations of the Task Force on CommunityPreventive Services. Morbidity and Mortality Weekly Report 51(RR01):1–8.

Barker, David J.P., Peter D. Gluckman, Keith M. Godfrey, Jane E. Harding, Julie A.Owens, and Jeffrey S. Robinson. 1993. Fetal Nutrition and Cardiovascular Disease inAdult Life. Lancet 341:938–41.

Bartley, Mel, David Blane, and Scott Montgomery. 1997. Health and the Lifecourse:Why Safety Nets Matter. BMJ 314:1194–96.

Berkman, Lisa F., and Ichiro Kawachi, eds. 2000. Social Epidemiology. New York:Oxford University Press.

Boyle, Michael H. 2002. Home Ownership and the Emotional and Behavioral Problemsof Children and Youth. Child Development 73(3):883–92.

Briggs, Xavier de Souza. 1997. Moving Up Versus Moving Out: Neighborhood Effects inHousing Mobility Programs. Housing Policy Debate 8(1):195–234.

Briggs, Xavier de Souza. 1998. Brown Kids in White Suburbs: Housing Mobility andthe Many Faces of Social Capital. Housing Policy Debate 9(1):177–221.

Briggs, Xavier de Souza, Joe T. Darden, and Angela A. Aidala. 1999. In the Wake ofDesegregation: Early Impacts of Scattered-Site Public Housing on Neighborhoods inYonkers, New York. Journal of the American Planning Association 65:27–49.

Briggs, Xavier de Souza, and the Yonkers Family and Community Project. 1997.Yonkers Revisited: The Early Impacts of Scattered-Site Public Housing on Families andNeighborhoods. A Report to the Ford Foundation. New York: Columbia UniversityTeachers College, Harvard University, and Michigan State University.

Briss, Peter A., Stephanie Zaza, Marguerite Pappaioanou, Jonathan Fielding, LindaWright-De Agüero, Benedict I. Truman, David P. Hopkins, Patricia Dolan Mullen,Robert S. Thompson, Steven H. Woolf, Vilma G. Carande-Kulis, Laurie Anderson, AlanR. Hinman, David V. McQueen, Steven M. Teutsch, Jeffrey R. Harris, and the TaskForce on Community Preventive Services. 2000. Developing an Evidence-Based Guideto Community Preventative Services—Methods. American Journal of Preventive Medi-cine 18(1S):35–43.

Buka, Stephen L., Robert T. Brennan, Janet W. Rich-Edwards, Stephen W. Raudenbush,and Felton Earls. 2003. Neighborhood Support and the Birth Weight of Urban Infants.American Journal of Epidemiology 157(1):1–8.




Buron, Larry, Susan J. Popkin, Diane Levy, Laura E. Harris, Jill Khadduri, Abt Associ-ates, Inc., the Urban Institute, and the Office of Public Housing Investments. 2002.The HOPE VI Resident Tracking Study. Prepared for the U.S. Department of Housingand Urban Development. Washington, DC: Urban Institute.

Christiani, David C. 1996. Utilization of Biomarker Data for Clinical and Environmen-tal Intervention. Environmental Health Perspectives 104(suppl 5):921–25.

Conley, Dalton. 1999. Being Black, Living in the Red: Race, Wealth, and Social Policy inAmerica. Berkeley, CA: University of California Press.

Cook, Thomas D., and Donald T. Campbell. 1979. Validity. In Quasi-Experimentation:Design and Analysis Issues for Field Settings, 37–94. Boston: Houghton Mifflin.

Deaton, Angus. 2002. Policy Implications of the Gradient of Health and Wealth. HealthAffairs March-April, pp. 13–30.

De Gruttola, Victor G., Pamela Clax, David L. DeMets, Gregory J. Downing, Susan S.Ellenberg, Lawrence Friedman, Mitchell H. Gail, Ross Prentice, Janet Wittes, andScott L. Zeger. 2001. Considerations in the Evaluation of Surrogate Endpoints in Clini-cal Trials: Summary of a National Institutes of Health Workshop. Controlled ClinicalTrials 22:485–502.

Dunn, James R. 2000. Housing and Health Inequalities: Review and Prospects forResearch. Housing Studies 15(3):341–66.

Dunn, James R. 2002. A Population Health Approach to Housing: A Framework forResearch. Ottawa, Ontario, Canada: National Housing Research Committee andCanada Mortgage and Housing Corporation.

Dunn, James R., and Michael V. Hayes. 2000. Social Inequality, Population Health, andHousing: A Study of Two Vancouver Neighborhoods. Social Science and Medicine51:563–87.

Ellen, Ingrid Gould. 2000. Is Segregation Bad for Your Health? The Case of Low BirthWeight. In Papers on Urban Affairs, ed. William G. Gale and Janet Rothenberg Pack,203–38. Washington, DC: Brookings Institution Press.

Ellen, Ingrid Gould, Tod Mijanovich, and Keri-Nicole Dillman. 2001. NeighborhoodEffects on Health: Exploring the Links and Assessing the Evidence. Journal of UrbanAffairs 23(3–4):391–408.

Ellen, Ingrid Gould, and Margery Austin Turner. 1997. Does Neighborhood Matter?Assessing Recent Evidence. Housing Policy Debate 8(4):833–66.

Fauth, Rebecca C., Tama Leventhal, and Jeanne Brooks-Gunn. 2002. They’re MovingOut; Are They Moving Up? Early Impacts of Moving to Low-Poverty Neighborhoods onLow-Income Youth. New York: Columbia University, Teachers College, National Centerfor Children and Families.

Fischer, Paul B. 1991. Is Housing Mobility an Effective Anti-Poverty Strategy? AnExamination of the Cincinnati Experience. Lake Forest, IL: Lake Forest College,Stephen H. Wilder Foundation.


Freeman, Lance. 2002. America’s Affordable Housing Crisis: A Contract Unfulfilled.American Journal of Public Health 92(5):709–12.

Fried, Marc. 1963. Grieving for a Lost Home. In The Urban Condition: People and Pol-icy in the Metropolis, ed. Leonard J. Duhl, 151–71. New York: Simon & Schuster.

Fullilove, Mindy Thompson. 2003. Neighborhoods and Infectious Diseases. In Neigh-borhoods and Health, ed. Ichiro Kawachi and Lisa F. Berkman, 211–22. New York:Oxford University Press.

Gabriel, Stuart A. 1996. The Role of FHA in the Provision of Credit to Minorities. InMortgage Lending, Racial Discrimination, and Federal Policy, ed. John Goering andRon Wienk, 183–250. Washington, DC: Urban Institute Press.

Gee, Gilbert C. 2002. A Multilevel Analysis of the Relationship between Institutionaland Individual Racial Discrimination and Health Status. American Journal of PublicHealth 92(4):615–23.

Geronimus, Arline T. 2000. To Mitigate, Resist, or Undo: Addressing Structural Influ-ences on the Health of Urban Populations. American Journal of Public Health90(6):867–72.

Goering, John. 2003. The Impacts of New Neighborhoods on Poor Families: Evaluatingthe Policy Implications of the Moving to Opportunity Demonstration. FRBNY Eco-nomic Policy Review, June, pp. 113–40.

Goering, John, Judith D. Feins, and Todd M. Richardson. 2002. A Cross-Site Analysis ofInitial Moving to Opportunity Demonstration Results. Journal of Housing Research13(1):1–30.

Goering, John, Joan Kraft, Judith Feins, Debra McInnis, Mary Joel Holin, and HudaElhassan. 1999. Moving to Opportunity for Fair Housing Demonstration Program: Cur-rent Status and Initial Findings. Washington, DC: U.S. Department of Housing andUrban Development.

Greenland, Sander. 2000. An Introduction to Instrumental Variables for Epidemiolo-gists. International Journal of Epidemiology 29(4):722–29.

Hanratty, Maria H., Sara A. McLanahan, and Becky Pettit. 2003. Los Angeles SiteFindings. In Choosing a Better Life? Evaluating the Moving to Opportunity SocialExperiment, ed. John Goering and Judith D. Feins, 245–74. Washington, DC: UrbanInstitute Press.

Harrell, Jules P., Sadiki Hall, and James Taliaferro. 2003. Physiological Responses toRacism and Discrimination: An Assessment of the Evidence. American Journal of Pub-lic Health 93(2):243–48.

Hart, Kevin D., Stephen J. Kunitz, Ralph R. Sell, and Dana B. Mukamel. 1998. Metro-politan Governance, Residential Segregation, and Mortality among African Americans.American Journal of Public Health 88(3):434–38.

Hattersley, Lin. 1999. Trends in Life Expectancy by Social Class—An Update. HealthStatistics Quarterly 2:16–24.




Haurin, Donald R., Toby L. Parcel, and R. Jean Haurin. 2002. Impact of Homeowner-ship on Child Outcomes. In Low-Income Homeownership: Examining the UnexaminedGoal, ed. Nicolas P. Retsinas and Eric S. Belsky, 427–46. Washington, DC: Joint Centerfor Housing Studies, Harvard University, and Brookings Institution Press.

Heymann, S. Jody, and Alison Earle. 1999. The Impact of Welfare Reform on Parents’Ability to Care for Their Children’s Health. American Journal of Public Health89(4):502–05.

Iannotta, Joah G., Jane L. Ross, and the National Research Council. 2002. Equality ofOpportunity and the Importance of Place: Summary of a Workshop. Washington, DC:National Academy Press.

Jackson, James S., Tony N. Brown, David R. Williams, Myriam Torres, Sherrill L. Sell-ers, and Kendrick Brown. 1996. Racism and the Physical and Mental Health Status ofAfrican Americans: A Thirteen-Year National Panel Study. Ethnicity & Disease6(1–2):132–47.

Joint Center for Housing Studies. 2003. The State of the Nation’s Housing: 2003. Cam-bridge, MA: Harvard University.

Kaplan, George A., and Jukka T. Salonen. 1990. Socioeconomic Conditions in Childhoodand Ischaemic Heart Disease during Middle Age. British Medical Journal 301:1121–23.

Katz, Bruce J., and Margery Austin Turner. 2001. Who Should Run the HousingVoucher Program? A Reform Proposal. Housing Policy Debate 12(2):239–62.

Katz, Lawrence F., Jeffrey R. Kling, and Jeffrey B. Liebman. 2001. Moving to Opportu-nity in Boston: Early Results of a Randomized Mobility Experiment. Quarterly Journalof Economics 116(2):607–54.

Kawachi, Ichiro, and Lisa F. Berkman. 2003a. Introduction. In Neighborhoods andHealth, ed. Ichiro Kawachi and Lisa F. Berkman, 1–19. New York: Oxford UniversityPress.

Kawachi, Ichiro, and Lisa F. Berkman, eds. 2003b. Neighborhoods and Health. NewYork: Oxford University Press.

Kearns, Ade, Rosemary Hiscock, Anne Ellaway, and Sally Macintyre. 2000. “BeyondFour Walls”: The Psychosocial Benefits of Home. Evidence from West Central Scotland.Housing Studies 15(3):387–410.

Kennedy, Stephen D., and Mireille L. Leger. 1990. Final Comprehensive Report of theFreestanding Housing Voucher Demonstration. Washington, DC: U.S. Department ofHousing and Urban Development.

Kiecolt-Glaser, Janice K., Lynanne McGuire, Theodore F. Robles, and Ronald Glaser.2002. Emotions, Morbidity, and Mortality: New Perspectives from Psychoneuroim-munology. Annual Review of Psychology 53:83–107.

Kling, Jeffrey R., Jeffrey B. Liebman, and Lawrence F. Katz. 2001. Bullets Don’t GotNo Name: Consequences of Fear in the Ghetto. Working Paper 225. Joint Center forPoverty Research.


Krieger, James, and Donna L. Higgins. 2002. Housing and Health: Time Again for Pub-lic Health Action. American Journal of Public Health 92(5):758–68.

Krieger, Nancy, and Steve Sidney. 1996. Racial Discrimination and Blood Pressure: TheCARDIA Study of Young Black and White Adults. American Journal of Public Health86(10):1370–78.

Kuh, Diana, Rebecca Hardy, Claudia Langenberg, Marcus Richards, and Michael E. J.Wadsworth. 2002. Mortality in Adults Aged 26–54 Years Related to Socioeconomic Con-ditions in Childhood and Adulthood: Post War Birth Cohort Study. British MedicalJournal 325(7372):1076–80.

Leventhal, Tama, and Jeanne Brooks-Gunn. 2000. The Neighborhoods They Live In:The Effects of Neighborhood Residence on Child and Adolescent Outcomes. Psychologi-cal Bulletin 126(2):309–37.

Leventhal, Tama, and Jeanne Brooks-Gunn. 2003a. The Early Impacts of Moving toOpportunity on Children and Youth in New York City. In Choosing a Better Life: Evalu-ating the Moving to Opportunity Social Experiment, ed. John Goering and Judith Feins,213–44. Washington, DC: Urban Institute Press.

Leventhal, Tama, and Jeanne Brooks-Gunn. 2003b. Moving to Opportunity: An Experi-mental Study of Neighborhood Effects on Mental Health. American Journal of PublicHealth 93(9):1576–82.

Lewis Mumford Center for Comparative Urban and Regional Research. 2001. 2000Census Data: Whole Population Public Use Data. State University of New York–Albany.World Wide Web page <http://mumford1.dyndns.org/cen2000/WholePop/WPSegdata/1120msa.htm> (accessed February 19, 2003).

Librett, John J., Michelle M. Yore, and Thomas L. Schmid. 2003. Local OrdinancesThat Promote Physical Activity: A Survey of Municipal Policies. American Journal ofPublic Health 93(9):1399–1403.

Loucks, Eric. 2003. Biological Markers in Housing–Health Research. Presented at theHarvard School of Public Health Symposium on Housing, Neighborhoods, and Health:State of the Evidence, Boston. June 26.

Loucks, Eric B., Jerry M. Spiegel, Carlos Barcelo, Annalee Yassi, Alfredo Aldama, andJames R. Dunn. 2002. Using Biomarkers for Measurement of the Health Effects ofHousing Conditions: Practical Guidelines. Presented at the 5th Pan American HealthOrganization Inter-American Health and Housing Conference, Havana, Cuba. May.

Lundberg, Olle. 1993. The Impact of Childhood Living Conditions on Illness and Mor-tality in Adulthood. Social Science and Medicine 36:1047–52.

Macintyre, Sally, and Anne Ellaway. 2000. Ecological Approaches: Rediscovering theRole of the Physical and Social Environment. In Social Epidemiology, ed. Lisa F. Berk-man and Ichiro Kawachi, 332–48. New York: Oxford University Press.

Macintyre, Sally, and Anne Ellaway. 2003. Neighborhoods and Health: An Overview. InNeighborhoods and Health, ed. Ichiro Kawachi and Lisa F. Berkman, 20–42. New York:Oxford University Press.




Marmot, Michael G., George Davey Smith, Stephen Stansfeld, Chandra Patel, FionaNorth, Jenny Head, Ian White, Eric Brunner, and Amanda Feeney. 1991. HealthInequalities among British Civil Servants—The Whitehall-II Study. Lancet337(8754):1387–93.

Marsh, Alex, David Gordon, Pauline Heslop, and Christina Pantazis. 2000. HousingDeprivation and Health: A Longitudinal Analysis. Housing Studies 15(3):411–42.

Meyers, Alan, Deborah Frank, Nicole Roos, Karen E. Peterson, Virginia A. Casey, L.Adrienne Cupples, and Suzette M. Levenson. 1995. Housing Subsidies and PediatricUndernutrition. Archives of Pediatrics and Adolescent Medicine 149(10):1079–84.

Mood, Eric W. 1986. Housing and Health: APHA–CDC Recommended Minimum Hous-ing Standards. Washington, DC: American Public Health Association.

Morland, Kimberly, Steve Wing, Ana Diez-Roux, and Charles Poole. 2002. Neighbor-hood Characteristics Associated with the Location of Food Stores and Food ServicePlaces. American Journal of Preventive Medicine 22(1):23–29.

Nettleton, Sarah, and Roger Burrows. 1998. Mortgage Debt, Insecure Homeownership,and Health: An Exploratory Analysis. Sociology of Health and Illness 20(5):731–53.

Newman, Sandra J., and Ann B. Schnare. 1997. “...And a Suitable Living Environ-ment”: The Failure of Housing Programs to Deliver on Neighborhood Quality. HousingPolicy Debate 8(4):703–41.

Oakes, J. Michael. 2004. The (Mis)Estimation of Neighborhood Effects: Causal Infer-ence for a Practicable Social Epidemiology. Social Science and Medicine58(10):1929–52.

O’Campo, Patricia. 2003. Advancing Theory and Methods for Multilevel Models of Resi-dential Neighborhoods and Health. American Journal of Epidemiology 157(1):9–13.

O’Campo, Patricia, Andrea C. Gielen, Ruth R. Faden, Xiaonan Xue, Nancy Kass, andMei-Cheng Wang. 1995. Violence by Male Partners against Women during the Child-bearing Years: A Contextual Analysis. American Journal of Public Health 85(8, part1):1092–97.

Orr, Larry L. 1999. Social Experiments: Evaluating Public Programs with ExperimentalMethods. Thousand Oaks, CA: Sage.

Orr, Larry, Judith D. Feins, Robin Jacob, Erik Beecroft, Lisa Sanbonmatsu, Lawrence F.Katz, Jeffrey B. Liebman, and Jeffrey R. Kling. 2003. Moving to Opportunity for FairHousing Demonstration Program. Interim Impacts Evaluation. Washington, DC: U.S.Department of Housing and Urban Development.

Peroff, Kathleen A., Cloteal L. Davis, Ronald Jones, and U.S. Department of Housingand Urban Development, Office of Policy Development and Research. 1979. GautreauxHousing Demonstration: An Evaluation of Its Impact on Participating Households.Washington, DC: U.S. General Accounting Office.

Peterson, George E., and Kale Williams. 1995. Housing Mobility: What Has It Accom-plished and What Is Its Promise? In Housing Mobility: Promise or Illusion? ed. Alexan-der Polikoff, 7–101. Washington, DC: Urban Institute.


Pettit, Kathryn L. S., G. Thomas Kingsley, Claudia J. Coulton, and Jessica Cigna. 2003.Neighborhoods and Health: Building Evidence for Local Policy. Washington, DC: UrbanInstitute.

Popkin, Susan J., Laura E. Harris, and Mary K. Cunningham. 2002. Families in Tran-sition: A Qualitative Analysis of the MTO Experience. Prepared for the U.S. Depart-ment of Housing and Urban Development, Office of Policy Development and Research,under a subcontract with Abt Associates, Inc. Washington, DC: Urban Institute.

Power, Chris, Clyde Hertzman, Sharon Matthews, and Orly Manor. 1997. Social Differ-ences in Health: Life Cycle Effects between Ages 23 and 33 in the 1958 British BirthCohort. American Journal of Public Health 87:1499–503.

Preston, Samuel H., and Michael R. Haines. 1991. Fatal Years: Child Mortality in LateNineteenth-Century America. Princeton, NJ: Princeton University Press.

Robert, Stephanie, and James S. House. 1996. SES Differentials in Health by Age andAlternative Indicators of SES. Journal of Aging and Health 8(3):359–88.

Rosenbaum, Emily, and Laura E. Harris. 2001. Low-Income Families in Their NewNeighborhoods: The Short-Term Effects of Moving from Chicago’s Public Housing.Journal of Family Issues 22(3):183–210.

Rosenbaum, James E. 1994. Housing Mobility Strategies for Changing the Geographyof Opportunity. Working paper. Northwestern University, Institute for Policy Research.

Rosenbaum, James E. 1995. Changing the Geography of Opportunity by ExpandingResidential Choice: Lessons from the Gautreaux Program. Housing Policy Debate6(1):231–69.

Rosenbaum, James E., and Susan J. Popkin. 1990. Economic and Social Impacts ofHousing Integration. Evanston, IL: Northwestern University, Center for Urban Affairsand Policy Research.

Saegert, Susan C., Susan Klitzman, Nicholas Freudenberg, Jana Cooperman-Mroczek,and Salwa Nassar. 2003. Healthy Housing: A Structured Review of Published Evalua-tions of U.S. Interventions to Improve Health by Modifying Housing in the UnitedStates, 1990–2001. American Journal of Public Health 93(9):1471–77.

Saelens, Brian E., James F. Sallis, Jennifer B. Black, and Diana Chen. 2003. Neighbor-hood-Based Differences in Physical Activity: An Environment Scale Evaluation. Ameri-can Journal of Public Health 93(9):1552–58.

Sard, Barbara. 2000. Housing Vouchers Should Be a Major Component of Future Hous-ing Policy for the Lowest-Income Families. Housing Policy in the New Millennium: Pro-ceedings, ed. Susan M. Wachter and R. Leo Penne, 291–318. Washington, DC: U.S.Department of Housing and Urban Development. World Wide Web page <http://www.huduser.org/publications/polleg/hpcproceedings.html> (accessed February 17, 2003).

Shroder, Mark. 2000. Social Experiments in Housing. Cityscape: A Journal of PolicyDevelopment and Research 5(1):237–59.

Shroder, Mark. 2002. Locational Constraint, Housing Counseling, and Successful Lease-Up in a Randomized Housing Voucher Experiment. Journal of Urban Economics51(2):315–38.




Srinivasan, Shobha, Liam R. O’Fallon, and Allen Dearry. 2003. Creating Healthy Com-munities, Healthy Homes, Healthy People: Initiating a Research Agenda on the BuiltEnvironment and Public Health. American Journal of Public Health 93(9):1446–50.

Subramanian, S. V., Kelvyn Jones, and Craig Duncan. 2003. Multilevel Methods forPublic Health Research. In Neighborhoods and Health, ed. Ichiro Kawachi and Lisa F.Berkman, 65–111. New York: Oxford University Press.

Syme, S. Leonard. 1987. Social Determinants of Disease. Annals of Clinical Research19(2):44–52.

Syme, S. Leonard, and Lisa F. Berkman. 1976. Social Class, Susceptibility, and Sickness.American Journal of Epidemiology 104(1):1–8.

Thiele, Bret. 2002. The Human Right to Adequate Housing: A Tool for Promoting andProtecting Individual and Community Health. American Journal of Public Health92(5):712–15.

Thomson, Hilary, Mark Petticrew, and David Morrison. 2001. Health Effects of HousingImprovement: Systematic Review of Intervention Studies. British Medical Journal323(7306):187–90.

Turner, Margery Austin. 1998. Moving Out of Poverty: Expanding Mobility and Choicethrough Tenant-Based Housing Assistance. Housing Policy Debate 9(2):373–94.

Turner, Margery Austin, Susan Popkin, and Mary Cunningham. 2000. Section 8 Mobil-ity and Neighborhood Health: Emerging Issues and Policy Challenges. Washington, DC:Urban Institute.

U.S. Bureau of the Census and Economics and Statistics Administration. 2001. NetWorth of Nation’s Households Unchanged, Census Bureau Reports. Press release. WorldWide Web page <http://www.census.gov/Press-Release/www/2001/cb01-33.html>(accessed May 2002).

U.S. Department of Housing and Urban Development. 1995. Homeownership and ItsBenefits. Urban Policy Brief. Washington, DC.

U.S. Department of Housing and Urban Development. 2002. Fiscal Year 2003 BudgetSummary. Washington, DC.

U.S. Department of Housing and Urban Development, Office of Policy Developmentand Research. 2000. Rental Housing Assistance—The Worsening Crisis. A Report toCongress on Worst-Case Housing Needs. Washington, DC.

Wachter, Susan M. 2000. Introduction. Housing Policy in the New Millennium: Proceed-ings, ed. Susan M. Wachter and R. Leo Penne, 1–12. Washington, DC: U.S. Departmentof Housing and Urban Development. World Wide Web page <http://www.huduser.org/publications/polleg/hpcproceedings.html> (accessed February 17, 2003).

Wadsworth, Michael E. J., Scott M. Montgomery, and Mel J. Bartley. 1999. The Persist-ing Effect of Unemployment on Health and Social Well-Being in Men Early in WorkingLife. Social Science and Medicine 48(10):1491–99.


Whitehead, Christine. 2000. Editorial. Housing Studies 15(3):339–40.

Williams, David R. 1996. Racism and Health: A Research Agenda. Ethnicity & Disease6(1–2):1–8.

Williams, David R. 1997. Race and Health: Basic Questions, Emerging Directions.Annals of Epidemiology 7:322–33.

Williams, David R. 2001. Racial Variations in Adult Health Status: Patterns, Paradoxes,and Prospects. In America Becoming: Racial Trends and Their Consequences. Vol. II,ed. Neil J. Smelser, William Julius Wilson, Faith Mitchell, and the National ResearchCouncil. Washington, DC: National Academy Press.

Williams, David R., and Chiquita Collins. 2001. Racial Residential Segregation: A Fun-damental Cause of Racial Disparities in Health. Public Health Reports 116(5):404–16.

Williams, David R., and Michelle Harris-Reid. 1999. Race and Mental Health: EmergingPatterns and Promising Approaches. In A Handbook for the Study of Mental Health:Social Context, Theories, and Systems, ed. Allan V. Horwitz and Teresa L. Scheid,295–314. Cambridge, England: Cambridge University Press.

Williams, David R., and Harold Neighbors. 2001. Racism, Discrimination, and Hyper-tension: Evidence and Needed Research. Ethnicity & Disease 11(4):800–16.

Williams, David R., Harold W. Neighbors, and James S. Jackson. 2003. Racial/EthnicDiscrimination and Health: Findings from Community Studies. American Journal ofPublic Health 93(2):200–08.

Williams, David R., Yan Yu, James S. Jackson, and Norman B. Anderson. 1997. RacialDifferences in Physical and Mental Health: Socioeconomic Status, Stress, and Discrimi-nation. Journal of Health Psychology 2(3):335–51.

Yinger, John. 1995. Closed Doors, Opportunities Lost: The Continuing Costs of HousingDiscrimination. New York: Russell Sage Foundation.



Does Housing Mobility Policy Improve Health? · between housing and health and brieﬂy discuss...

Documents

Transcript of Does Housing Mobility Policy Improve Health? · between housing and health and brieﬂy discuss...