«Moving to opportunity voluMe 14, nuMber 2 • 2012 U.S. Department of Housing and Urban Development | Office of Policy Development and Research ...»
The remaining subsections review the previous empirical research based around the key youth outcome domains of interest and the extent to which this literature sheds any light on the pathways by which neighborhoods have these effects.
Education and Employment Perhaps the most obvious way in which neighborhood context may affect educational outcomes is through the quality of local public schools. Moving to better neighborhoods for better schools arose as one key motivation for MTO study families to subsequently move, although realizing their aspirations for improved educational opportunities was difficult and often influenced by informal networks (Ferryman et al., 2008). Indeed, the composition of neighborhood residents might also matter, because adults convey shared prosocial (or antisocial) values or serve as positive or negative role models; that is, the Jencks-Mayer collective socialization model (Connell et al., 1995; Crane, 1991; Sampson, 1993; Sampson and Groves, 1989; Wilson, 1987). Such neighborhoods also may provide youth with a safe physical environment, which may be conducive to academic success (Connell et al., 1995). Epidemic models raise the possibility of a variety of spillover effects from exposure to higher achieving peers; for example, through opportunities to participate in more productive study groups, exposure to more rigorous instruction, and increased time on task from reductions in student disruptions (Lazear, 2001). On the other hand, the competition for grades may be more intense in more affluent areas. Increased competition could have a detrimental effect on some MTO children, although these effects might dissipate over time if their academic competencies improve with exposure to new schools.
Previous nonexperimental research generally has found positive correlations between affluent neighbors and a variety of academic outcomes, such as IQ, reading and math achievement scores, school completion, and self-reported grades for children and adolescents (for example, BrooksGunn et al., 1993; Chase-Lansdale et al., 1997; Connell and Halpern-Felsher, 1997; Crane, 1991;
Dornbusch, Ritter, and Steinberg, 1991; Duncan, Brooks-Gunn, and Klebanov, 1994; Entwistle, Alexander, and Olson, 1994). A study of the Gautreaux mobility program found that young adults in households that had moved from public housing in the city of Chicago to suburban locations were less than one-fourth as likely to drop out of school and more than twice as likely to attend college compared with the outcomes of young adults initially living in the same public housing units whose families moved to other parts of the city (Rubinowitz and Rosenbaum, 2000). Preliminary analyses using longer run data on a larger group of Gautreaux children suggested smaller and more specialized effects (Keels et al., 2005).
Schools also can be gateways to other types of educational or work programs. School-to-work programs administered by local public schools may help youth secure internships while they are still enrolled in high school and help non-college-bound youth secure employment after high school. Factors such as the stigma surrounding entry-level jobs or local criminal activity, the level of violence associated with the local illegal economy and the quality of local policing, and the level of difficulty in competing for jobs and related positive rewards for behavior that supports schooling or employment can also influence youth decisions about whether to participate in the formal labor market or to pursue underground or informal work. Although having more affluent neighbors appears to correspond with having improved labor market outcomes (see, for example, Corcoran et al., 1992; Page and Solon, 2003; Sharkey, 2008), several more recent studies suggest a mixed pattern of neighborhood environment influence. Child neighborhood environments do not appear related to adult labor market outcomes among children assigned to public housing projects in substantially different neighborhoods of Toronto (Oreopoulos, 2003). Further analyses of Gautreaux Cityscape 141 Gennetian, Sciandra, Sanbonmatsu, Ludwig, Katz, Duncan, Kling, and Kessler (Rosenbaum, 1995) found strong positive gains in educational and economic outcomes for the children of suburban movers relative to those of city movers, but longer term followups found less striking contrasts between suburban and city movers (DeLuca et al., 2009). On the other hand, Gould, Lavy, and Paserman (2009) found positive effects on long-term adult economic outcomes for Yemenite refugees to Israel who, as children, were placed initially in more prosperous neighborhoods with better infrastructure.
Delinquency and Risky or Problem Behavior In addition to potentially affecting educational and employment outcomes, MTO may have important effects on problem behaviors. Social stigma associated with criminal behavior may be lower in areas where such behavior is relatively more common. Similarly, if police resources assigned or available to a community are relatively fixed, an increase in criminal activity by one’s peers will reduce the probability that a given criminal offense results in arrest (Cook and Goss, 1996). The literature is more mixed regarding other risky behaviors, such as drug or substance abuse and sexual activity (Brooks-Gunn et al., 1993; Crane, 1991; Esbensen and Huizinga, 1990; Hogan, Astone, and Kitagawa, 1985; Hogan and Kitagawa, 1985).
Health Although Jencks and Mayer (1990) did not consider neighborhood processes to be related to health, we have reason to believe that moves to lower poverty neighborhoods may improve both physical and mental health. Physical health may improve with safer and less stressful environments, greater community resources, or residents who practice healthy behaviors such as exercise. Low-income neighborhoods may also have compromised air quality, which has been linked to coronary heart disease (Kan et al., 2008) and poor health for infants (Currie and Walker, 2011). Poor children living in disadvantaged urban areas may be at higher risk of exposure to lead and secondhand smoke, both of which can impair brain development (Bombard et al., 2010; Filippelli and Laidlaw, 2010).
The prevalence of accidents and injuries—the most common causes of death among children ages 1 to 14 in the United States—may be higher among children living in distressed urban communities, owing to unsafe playgrounds and other features of the environment (Quinlan, 1996; Scharfstein and Sandel, 1998).
Adults and children who live in high-poverty, high-crime urban settings are also at risk for poorer mental health outcomes (for example, Bagley, Jacobson, and Palmer, 1973; Rezaeian et al., 2005;
Whitley et al., 1999). To the extent that MTO reduces exposure to crime and violence, we would expect it to improve overall well-being and reduce psychological distress, depression, and anxiety (Aneshensel and Sucoff, 1996; Ross and Mirowsky, 2001; Silver, Mulvey, and Swanson, 2002).
Moving to lower poverty neighborhoods could influence a variety of externalizing behavior disorders (for example, oppositional defiant disorder), because these disorders are strongly related to contagion processes in peer environments and norms regarding the appropriateness of violence and antisocial behaviors (Deater-Deckard, 2001; Gifford-Smith et al., 2005). Male and female youth may also have different coping styles and capacities as they navigate different neighborhood environments. Adolescent males tend to be subject to less parental supervision than females, and they also tend to be greater risk takers (Block, 1983; Bottcher, 2001; LaGrange and Silverman, 1999).
142 Moving to Opportunity The Long-Term Effects of Moving to Opportunity on Youth Outcomes Psychosocial stress sometimes can have more pronounced effects on males than females, in part, because males are more likely to use confrontational techniques (that is, the Jencks-Mayer collective socialization model) to deal with stress, particularly stress involving interpersonal problems, whereas females are more likely to turn to supportive adults (Coleman and Hendry, 1999; Zaslow and Hayes, 1986).
The Jencks and Mayer (1990) typology and empirical literature, in turn, generally implies that the effect of MTO moves may become more beneficial (or less detrimental) as youth spend more time in lower poverty areas. For example, over time, we may expect MTO youth to become more socially integrated into their new communities and more attuned to local social norms, and thus more responsive to the peer and adult social influences that are central to the epidemic and collective socialization models. Parents may also learn over time how to better navigate the potential opportunities and pitfalls in low-poverty schools. More generally, the effects of exposure to new social environments and institutions may accumulate over time and lead to more pronounced positive effects on youth behavior. A different time path in MTO effects may arise from the effects of neighborhood safety and crime, which, as mentioned previously, may be relevant for outcomes in the schooling, employment, and delinquency domains. On the other hand, some of the theories described in the Jencks and Mayer typology—the competition and relative deprivation models in particular—predict potentially adverse effects as youth spend time in lower poverty neighborhoods. Ultimately, whether youth benefit based on these theories is an empirical question.
The MTO Study Design, Sample, and Data From 1994 to 1998, HUD launched MTO in five cities: Baltimore, Boston, Chicago, Los Angeles, and New York. HUD limited eligibility to families with children living in public or other government housing in designated high-poverty census tracts. The study then randomly assigned the 4,604 families who signed up to one of three groups. HUD offered families in the experimental group the opportunity to use a rent-subsidy voucher to move into private-market housing but, under the MTO design, families in this group could redeem their vouchers only in census tracts with a 1990 poverty rate of less than 10 percent. Families in the experimental group also received housing search assistance and relocation counseling from local nonprofit organizations. HUD offered the randomly assigned families in the Section 8 group a traditional housing voucher that had no location requirements and did not come with any search assistance beyond what Section 8 voucher recipients normally receive. Families in the control group did not receive a voucher through MTO, but they did not lose access to any housing or other social services to which they would otherwise have been entitled.
The final impacts evaluation youth survey sample frame included up to three youth per original MTO family who were between ages 10 and 20 as of December 2007. Older adolescents (ages 13 to 20 as of December 2007) answered the full-length survey that we developed, whereas younger children (ages 10 to 12 as of December 2007) answered a shorter subset of items. Although MTO participants who were younger than 18 at baseline and older than 20 by December 2007 were not in our survey sample frame, we did try to track their outcomes through proxy reports of parents on the adult MTO surveys and through administrative data on employment, postsecondary schooling, and arrests.
Cityscape 143 Gennetian, Sciandra, Sanbonmatsu, Ludwig, Katz, Duncan, Kling, and Kessler Response rates for the youth survey were very high and were balanced across the control and treatment groups. The overall effective response rate was 89 percent, and the effective response rates by randomization group were as follows: experimental group, 90 percent; Section 8 group, 87 percent;
control group, 89 percent. The analysis sample for the survey-based measures presented in this article includes a total of 5,101 youth ages 10 to 20, comprising 457 younger children and 4,644 older adolescents. All youth ages 10 to 20 (N = 6,645) in the 4,604 families in the program (as opposed to only those youth who were interviewed) were eligible for submission to administrative data agencies.3 Roughly 56 percent (2,969 of 5,345) of the youth who were interviewed as part of the followup survey for the interim impacts evaluation were interviewed again as part of the longterm survey for the final impacts evaluation.
The youth in the long-term survey sample, who on average were age 5 at baseline (ranging from newborn to age 11), were not particularly disadvantaged regarding learning and behavioral problems.
They attended schools, however, characterized by high poverty, high minority composition, and low achievement; their parents had low educational achievement; and they were living in dangerous neighborhoods. The baseline heads of household reported that 12 percent of youth ages 6 to 11 had a learning problem and 6 percent had behavioral or emotional problems. About 13 percent had been enrolled in a program for gifted and talented students or had done advanced coursework.
These numbers are consistent with national averages; about 13 percent of the school-age population receives special education services (Kaufman, Alt, and Chapman, 2001) and 10 percent are enrolled in gifted classes (Fields et al., 2001). On the other hand, about 85 percent of students at the baseline schools of MTO youth were eligible for free or reduced-price lunches, more than 90 percent of students were minorities, and most of the schools were in the bottom 15 percent of the
statewide performance distribution. Furthermore, parents had relatively low educational attainment:
only 35 percent held a high school diploma, and another 18 percent had earned a certificate of General Educational Development (GED). Finally, as Ludwig (2012) described, when families listed their reasons for wanting to move, about three-fourths reported wanting to get away from gangs and drugs (that is, safety) as their first or second most important reason, about one-half listed better schools for the children, and about 45 percent listed a bigger or better apartment.
We estimate both the effects of being offered an MTO low-poverty voucher or a traditional Section 8 voucher, known as the intention-to-treat (ITT) effect in the program evaluation literature, and the effects of actually moving with a low-poverty or traditional voucher, known as the treatment-onthe-treated (TOT) effect. We calculate ITT using an ordinary least squares regression in which the outcome of interest is the dependent variable being predicted on group assignment and a series of baseline covariates. The basic equation is