«Moving to opportunity voluMe 14, nuMber 2 • 2012 U.S. Department of Housing and Urban Development | Office of Policy Development and Research ...»
Summary of Hypotheses We hypothesized that, compared with adults in the control group, adults offered the opportunity to move to lower poverty neighborhoods would have better mental health, physical health, and economic outcomes through improved neighborhood safety, less stressful environments, greater community resources, positive peer influences, and fewer environmental hazards. We expected to see lower incidences of mental illness such as depression, anxiety, and stress-related disorders as compared with the control group. We also expected to see physical health improvements in terms of lower rates of asthma, obesity, diabetes, and hypertension. We also hypothesized that families who moved to neighborhoods with improved conditions would have higher levels of employment and earnings and would receive less public assistance.
Previous Findings From MTO and Other Studies MTO was inspired by findings from the Gautreaux residential mobility program in Chicago, which was part of the legal settlement of a racial discrimination case. Gautreaux gave African-American families living in the inner city an opportunity to move to new neighborhoods. The initial Gautreaux findings suggested better employment outcomes for low-income African-American families living in public housing who moved to predominantly White suburbs compared with those of their counterparts who moved to predominantly African-American and urban neighborhoods (Rosenbaum, 1995). Gautreaux, however, was not a randomized study and the neighborhood preferences of Gautreaux families may, at least in part, have influenced where they ended up living (Votruba and Kling, 2009). Furthermore, over the long term, the program shows little in the way of gains in self-sufficiency for Gautreaux mothers, suggesting that one might not expect to see economic gains for MTO women (DeLuca et al., 2009).
HUD launched MTO to more systematically evaluate the types of neighborhood effects found in the Gautreaux study. An interim evaluation of MTO was conducted in 2002 by Abt Associates Inc.
in partnership with researchers at the National Bureau of Economic Research (Orr et al., 2003).
The evaluation examined effects an average of 4 to 7 years after the families joined MTO. These interim findings suggested that adults in the experimental group had lower levels of psychological 116 Moving to Opportunity The Long-Term Effects of Moving to Opportunity on Adult Health and Economic Self-Sufficiency distress and felt calmer than adults in the control group. The results also hinted at lower levels of depression, although these results were sensitive to the construction of the depression measure.
Based on self-reported height and weight, the treatment appeared to reduce obesity levels (Body Mass Index [BMI] of 30 or greater) for those in the experimental group (42.0 percent) in comparison with those in the control group (47.1 percent) and perhaps to increase rates of exercise and improve diet (Orr et al., 2003). The results, however, did not show significant effects on other health measures, such as self-rated health, hypertension, physical limitations, asthma, and a summary health index (Kling, Liebman, and Katz, 2007; Orr et al., 2003). Nor did the results show any statistically significant effects on economic outcomes, such as employment or earnings of the adults or youth.
Other studies of residential mobility programs have also yielded mixed findings on economic outcomes. Oreopoulos (2003) took advantage of the fact that public housing units in the city of Toronto were located in different types of neighborhoods to compare the outcomes of families living in different types of neighborhoods.2 He found no evidence of improved economic outcomes for youth who grew up in higher SES neighborhoods. Studies of HUD’s HOPE VI public housing demolition programs also yielded mixed results. Levy’s (2010) study of families who relocated through HOPE VI suggested that families who relocated because of the demolition of their public housing projects experienced improved neighborhood conditions, but their economic self-sufficiency changed little. In contrast, Anil, Sjoquist, and Wallace (2010), studying the HOPE VI demolitions in the Atlanta area, found evidence of employment gains.
The 10- to 15-Year Evaluation Our 10- to 15-year followup with MTO families (Sanbonmatsu et al., 2011) extends beyond previous MTO research by studying the long-term effects of MTO; expanding the outcomes examined;
and using objective measures of health in conjunction with self-reports on health. We expanded the MTO long-term data collection to include new outcomes, such as diabetes and PTSD. In addition, we replaced the brief questions on problems such as anxiety with more detailed and widely used structured diagnostic interview instruments, and we replaced self-reported height and weight measures with anthropometric measurements taken by the interviewers. For the first time, we gathered finger-stick dried blood spot samples from MTO respondents, enabling us to measure biological risk factors and undiagnosed disease.
The long-term survey for the final impacts evaluation enables us to examine how effects have changed over time. MTO’s effects might have followed three very different trajectories. Program effects might have faded over time as the average neighborhood environments of the two treatment groups and the control group converged, which could occur if families in the two treatment groups moved back to their old neighborhoods, if families in the control group moved out of public housing on their own (and into similar low-poverty areas as the treatment groups), or if The Toronto housing authority offered a housing unit with the necessary number of bedrooms to high-need families who reached the top of the waiting list on a first-available basis; families could not specify the housing project or type of housing project that they wanted to live in (Oreopoulos, 2003).
the neighborhoods themselves changed over time. Alternatively, MTO effects on families might have strengthened over time as the initial disruptions of moving diminished and families became increasingly connected to their new communities. Greater connections to neighbors in low-poverty areas might have produced continued changes in diet and exercise patterns or new job referrals.
Over time, we might have seen even larger MTO effects on health and the emergence of MTO effects on economic self-sufficiency. Or, effects might have followed a more intermediate path, whereby families maintained the initial benefits from moving that reached a plateau in the early years of the program. This latter scenario might have occurred if the moves led to some initial change that persisted over time, even as families made additional moves, but that was a one-time change that did not continue to increase. For example, moving from public housing to privatemarket rentals might reduce the noise level in the home. This improvement might persist with subsequent moves and lead to greater mental calm or improved sleep, but it may be a one-time shift with no additional gains over time.
Sample and Analytic Approach Our sample draws on the adults from the original MTO households. To measure MTO’s impacts over the long term, we selected up to one adult for interview in each MTO household.3 The Institute for Social Research (ISR) at the University of Michigan interviewed adults using a computer-assisted survey between June 2008 and April 2010 (10 to 15 years after families were randomized in the program). Interviewers asked questions about the adult’s health and economic circumstances, took physical measurements, and collected blood samples with a simple finger stick (McDade, Williams, and Snodgrass, 2007). ISR used a two-stage field design. In the first stage, ISR tried to interview as many adults in the survey sample as possible. After the response rate reached 75 percent for a site and sample release, ISR randomly selected 35 percent of the remaining, hard-to-reach respondents for the second stage of more intensive survey recruitment efforts. In all, ISR interviewed 3,273 adults and achieved an overall effective response rate of 90 percent (excluding deceased adults). In addition to collecting data from the survey, we gathered data from administrative records.
Exhibit 2 shows the baseline characteristics of the adults interviewed for the final impacts evaluation. At baseline, the vast majority (92 percent) of households were female headed and three-fourths of household heads were on welfare. The median household income of interviewed participants was $10,614 (in 2009 dollars) in the year preceding entry into MTO. Only about one-fourth of adults were working. Slightly more than one-third of adults in MTO families had graduated from high school. Nearly two-thirds were African American and most of the rest were Hispanic. The average age of our interviewed sample adults at the time they joined the program was 32.9, and the average family size was about 3.7 members.
In selecting the adult survey sample, we prioritized female adults and household heads from the core family, adopting the same approach used for the interim impacts evaluation (Orr et al., 2003). We selected for interview one adult from each family in the experimental group and the control group. Because of funding constraints, we were unable to interview adults from all families in the Section 8 group and instead randomly selected a 68-percent subsample of these families for adult interviews.
118 Moving to Opportunity The Long-Term Effects of Moving to Opportunity on Adult Health and Economic Self-Sufficiency
HUD asked families applying to the program about their primary and secondary reasons for wanting to move. By far, the most common reason for signing up for MTO was to get away from drugs and gangs (77 percent of adults listed this reason first or second), followed by finding better schools (49 percent) and finding a better apartment (44 percent; Sanbonmatsu et al., 2011, Exhibit 1.2). The importance of safety in motivating families to participate in MTO is perhaps not surprising, given that more than two of every five baseline respondents said that someone in their household had been the victim of a crime during the 6 months preceding the baseline survey.
We present two types of estimates for how MTO affected the life outcomes of participating adults.
The first estimate, the intention-to-treat (ITT) effect, represents the effect of being offered a housing voucher or certificate, which we generate by comparing the outcomes of all adults randomly assigned to the experimental or Section 8 group (regardless of whether those adults moved with a program voucher) with the outcomes of all adults assigned to the control group. We calculate the ITT estimate using an ordinary least squares regression in which the outcome of interest is the dependent variable being predicted on treatment group assignment and a series of baseline covariates. The basic regression equation is
where Yi is an outcome for MTO program participant i; Expi and S8i are binary indicator variables equal to 1 if participant i was randomly assigned to the experimental or Section 8 group (the control
group is the omitted reference group); and Xi represents a series of baseline covariates that Sanbonmatsu et al. (2011) described and that is similar to the covariates Orr et al. (2003) used. The coefficients on Expi and S8i capture the ITT estimates for the experimental and Section 8 groups, respectively. The equation weights the data to account for adjustments to the randomization ratios during the study enrollment period and the probability sampling.
If we are willing to assume that assignment to one of the treatment groups does not have much effect on those families who were offered a voucher but did not use it, we can also estimate the effect of actually moving using an MTO program voucher, known as the effect of treatment on the treated (TOT).4 We can calculate the TOT effect by dividing the ITT effect by the share of the experimental or Section 8 group that relocated with an MTO voucher (Angrist, Imbens, and Rubin, 1996; Bloom, 1984). Because approximately one-half of the families in the experimental group used the MTO program voucher, the estimated TOT effect will be about twice as large as the ITT effect (that is, TOT ≈ ITT/0.5 ≈ ITT × 2). The statistical significance of both the ITT and TOT estimates are identical under this calculation, because we scale up the standard error and the impact estimate by the same factor (1/voucher use rate).
Measures of Health and Economic Self-Sufficiency Mental Health Measures To assess adult mental health outcomes, we used responses on the survey and the structured diagnostic interview within the survey. We measured psychological distress with the Kessler 6 scale (K6), which consists of questions about sadness, nervousness, restlessness, hopelessness, feeling that everything is an effort, and feelings of worthlessness (Kessler et al., 2003). The raw scores from the K6 can range from 0 (no distress) to 24 (highest level of distress). We assessed major depressive disorder, GAD, and PTSD using the World Health Organization’s Composite International Diagnostic Interview (CIDI; Kessler and Ustün, 2004), which is designed to be consistent with the Diagnostic and Statistical Manual of Mental Disorders, 4th edition Task Force on DSM-IV (2000). To meet the criteria for major depression, a respondent’s depressed mood or loss of interest had to
last for a period of at least 2 weeks and be accompanied by at least five of the following symptoms:
depressed mood, diminished interest or pleasure, unintentional weight loss or gain, insomnia, restlessness or slowing down, fatigue, feelings of worthlessness or excessive guilt, diminished ability to concentrate, and recurrent thoughts of death. Furthermore, these symptoms had to cause significant distress or impair the respondent’s functioning at work or in social situations.