Neighborhood Effects on the Long-Term Well-Being of Low-Income Adults From All Five Sites of the Moving to Opportunity Experiment, 2008-2010 [Public Use Data] (ICPSR 34563)

Time Period(s) View help for Time Period(s)

Date of Collection View help for Date of Collection

Data Collection Notes View help for Data Collection Notes

Sample View help for Sample

The sampling frame for selection into the long-term survey sample was one adult from each of the 4,604 households who were randomized through the MTO program between 1994 and 1998. One "sample adult" was identified from each household by giving priority to female adults who were more likely to be caretakers of the children. The priority order for selecting both the MTO interim and final evaluation samples was: female heads of the core household, female spouses of the core household head, female baseline heads, female spouses of the baseline head, and finally, non-female heads of the core households. Note that the baseline head is often but not always the same person as the sample adult. For the long-term survey, the sample adults from all low-poverty voucher and control group households were selected for interview and, for budgetary reasons, the sample adults from only a random two-thirds of the traditional voucher group households were selected for interview. Youth ages 10 to 20 in the households were also interviewed, however, the data used for the Science article only includes adult respondents.

Universe View help for Universe

The cells included in the files represent data from the 3,273 interviews conducted with adult participants in the MTO program as part of the MTO long-term survey (2008-2010) that were analyzed for the Science article. At the study's outset, these adults came from low-income families with children recruited from public housing in highly distressed neighborhoods in Baltimore, Boston, Chicago, Los Angeles, and New York City.

Data Type(s) View help for Data Type(s)

Response Rates View help for Response Rates

The overall effective response rate (ERR) for the adult survey was 89.6 percent, and the ERRs by MTO treatment group were similar: 90.8 percent for the low-poverty voucher group, 86.6 percent for the traditional voucher group, and 90.0 percent for the control group.

Presence of Common Scales View help for Presence of Common Scales

The DSM-IV depression and Generalized Anxiety Disorder measures included on the dataset were assessed via the World Health Organization's Composite International Diagnostic Interview (see Kessler and Ustun (2004) for more information). The survey also included the Kessler 6 (K6), which consists of six questions and is the most widely used scale of nonspecific psychological distress in the literature (see Furukawa et al. (2003) and Kessler et al. (2003) for more information). Furukawa, T. A., Kessler, R. C., Slade, T., and Andrews, G. (2003). The performance of the K6 and K10 screening scales for psychological distress in the Australian National Survey of Mental Health and Well-Being. Psychological Medicine, 33(2), 357-362. Kessler, R. C., Barker, P. R., Colpe, L. J., Epstein, J. F., Gfroerer, J. C., Hiripi, E., Howes, M. J., et al. (2003). Screening for serious mental illness in the general population. Archives of General Psychiatry, 60(2), 184-189. Kessler, R. C., and Ustün, T. B. (2004). The World Mental Health (WMH) Survey Initiative Version of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI). International Journal of Methods in Psychiatric Research, 13(2), 93-121.

Original Release Date View help for Original Release Date

2013-02-28

Version History View help for Version History

Weight View help for Weight

All analysis of the data should be weighted using the total survey weight. The cell-level file includes a separate weight for each outcome and mediator measure that is the sum of weights for all observations in the cell with valid data for the measure (for example, wt_happy_scale123_ad is the weight for the subjective well-being scale measure, mn_happy_scale123_ad. In the pseudo-individual file, mn_f_wt_totsvy is the average of the total survey weight variable for all observations in the cell.

In the original individual-level file, the total survey weight (f_wt_totsvy) is calculated as the product of three component weights:

1. Randomization ratio weight - At the start of the MTO program, random assignment (RA) ratios were set to produce equal numbers of leased-up families in the low-poverty and traditional voucher groups based on expected lease-up rates. The initial ratios were "8 to 3 to 5": eight low-poverty voucher group families to three traditional voucher families to five control families. During the demonstration program, these RA ratios were adjusted to accommodate higher than anticipated lease up rates among low-poverty voucher group families. This weight ensures that the proportion of families in a given site is the same across all three treatment groups. This component weight value ranges from 0.59 to 2.09.

2. Survey sample selection weight - For budgetary reasons, adults from only a random two-thirds of traditional voucher group households were selected for the long-term survey interview sample (while adults from all low-poverty voucher and control group families were selected), so this component weights up the selected traditional voucher group adults so that they are representative of all traditional voucher group adults. This weight component is equal to the inverse probability of selection into the subsample (~1.52).

3. Phase 2 subsample weight - The long-term survey data collection was completed as a two-phase process. In the first phase, we sought to interview all selected respondents. Phase 2 of fielding was triggered when the response rate reached approximately 74 percent. In the second phase, we subsampled a random 35 percent of the remaining sample and only sought interviews with the selected subsample of respondents. This weight component is equal to the inverse probability of selection into the subsample (~2.86).