Table 2.
Residential Mobility, Poverty Rate, and Census-Tract Characteristics, According to Study Group.*
| Variable | Control | Low-Poverty Voucher | Traditional Voucher | ||||
|---|---|---|---|---|---|---|---|
| Mean | Intention-to-Treat Estimate (95% CI)† |
P Value |
Mean | Intention-to-Treat Estimate (95% CI)† |
P Value |
Mean | |
| Mean no. of moves‡ | 2.1 | 0.57 (0.42 to 0.71) | <0.001 | 2.7 | 0.58 (0.38 to 0.79) | <0.001 | 2.7 |
| Poverty rate in census tract (%)§ | |||||||
| Baseline | 53.1 | −0.37 (−1.23 to 0.50) | 0.41 | 52.5 | −0.37 (−1.55 to 0.81) | 0.54 | 52.9 |
| At 1 yr | 50.0 | −17.06 (−18.57 to −15.56) | <0.001 | 32.7 | −13.50 (−15.33 to −11.67) | <0.001 | 36.6 |
| At 5 yr | 39.9 | −9.78 (−11.25 to −8.31) | <0.001 | 30.0 | −6.26 (−8.41 to −4.11) | <0.001 | 33.0 |
| At 10 yr | 33.0 | −4.86 (−6.23 to −3.48) | <0.001 | 28.3 | −2.87 (−4.80 to −0.95) | 0.003 | 29.2 |
| Mean census-tract characteristics (%)¶ | |||||||
| Poor‖ | 39.6 | −9.14 (−10.26 to −8.02) | <0.001 | 30.4 | −6.07 (−7.53 to −4.61) | <0.001 | 32.9 |
| Minorities | 88.0 | −6.23 (−7.58 to −4.89) | <0.001 | 81.9 | −0.99 (−2.88 to 0.90) | 0.30 | 85.8 |
| Household headed by a woman | 54.3 | −7.95 (−9.08 to −6.82) | <0.001 | 46.2 | −5.03 (−6.55 to −3.51) | <0.001 | 48.7 |
| College graduate | 16.1 | 4.49 (3.68 to 5.30) | <0.001 | 20.5 | 1.41 (0.29 to 2.52) | 0.01 | 18.4 |
| Respondents reporting collective efficacy (%)** | |||||||
| At 4–7 yr | 54.0 | 10.61 (6.46 to 14.76) | <0.001 | 65.4 | 5.30 (0.53 to 10.07) | 0.03 | 59.9 |
| At 10–15 yr | 58.9 | 8.20 (4.20 to 12.21) | <0.001 | 67.2 | 0.80 (−5.16 to 6.76) | 0.79 | 62.4 |
| Respondents reporting feeling safe or very safe on streets near home during the day (%) | |||||||
| At 4–7 yr | 74.9 | 9.14 (5.77 to 12.52) | <0.001 | 84.6 | 8.95 (5.16 to 12.73) | <0.001 | 84.4 |
| At 10–15 yr | 80.7 | 3.70 (0.52 to 6.87) | 0.02 | 84.2 | 5.00 (0.50 to 9.50) | 0.03 | 85.1 |
| Respondents reporting having at least one friend who graduated from college (%) | |||||||
| At 4–7 yr | 40.8 | 6.90 (2.63 to 11.17) | 0.002 | 48.0 | 4.55 (−0.22 to 9.33) | 0.06 | 45.3 |
| At 10–15 yr | 53.4 | 6.90 (2.74 to 11.06) | 0.001 | 60.4 | −2.11 (−8.33 to 4.11) | 0.51 | 53.2 |
| Respondents reporting access to local health care services, excluding emergency room (%) | |||||||
| At 4–7 yr | 89.7 | −1.35 (−4.13 to 1.43) | 0.34 | 88.8 | −0.21 (−3.15 to 2.73) | 0.89 | 89.5 |
| At 10–15 yr | 93.4 | −1.36 (−3.49 to 0.77) | 0.21 | 92.1 | 0.64 (−2.11 to 3.40) | 0.65 | 95.2 |
The analysis sample consisted of women with a valid BMI or glycated hemoglobin measurement. Analyses of number of moves and census-tract characteristics were further limited to participants with valid addresses at baseline and years 1, 5, and 10. The intention-to-treat estimates come from a regression that compares average outcomes across randomly assigned groups, with statistical control for baseline characteristics, which may differ slightly from the difference in raw group means presented here. See the Supplementary Appendix for the sample sizes used.
Intention-to-treat estimates compare the average of the outcomes for everyone assigned to the intervention group with the average of the outcomes for controls, with adjustment for the set of baseline covariates shown in Table 1 and indicators for survey-sample release (families were randomly selected with regard to the time at which they would first be contacted about participation in the long-term follow-up study), site, and random-assignment periods. The effects on continuous dependent variables were calculated with the use of linear regression; the effects on dichotomous variables were calculated with the use of logistic regression and are presented as average marginal effects.
The total number of moves is the number from the time of randomization (1994 through 1998) to the beginning of long-term follow-up (May 2008).
Census-tract characteristics were recorded as of the time when a family lived in the tract and were interpolated with the use of 1990 and 2000 decennial census data and data from the American Community Survey, 2005 to 2009.
Average duration-weighted census-tract characteristics give more weight to tracts in which families spent relatively more time during the study period.
The term “poor” is defined as having an annual income below the federal government’s poverty threshold.
Collective efficacy is defined as the likelihood that adults will take action in response to youth spraying graffiti on local buildings. See Sampson et al. for more details on collective efficacy.29