Helping Parents Earn More: A Randomized Evaluation of Holistic Wrap-around Services for Parents and their Children
How to cite
Tebes, Jonathan; Giles, Tyler. 2023. "Helping Parents Earn More: A Randomized Evaluation of Holistic Wrap-around Services for Parents and their Children." AEA RCT Registry. AEARCTR-0012684. https://doi.org/10.1257/rct.12684-2.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
Catholic Charities of Northern Nevada (CCNN) is a nonprofit organization dedicated to supporting families in need in the Diocese of Reno, Nevada. To support families in poverty, CCNN is launching Elevating Families, a holistic, case management program for parents with young children that is based on Economic Mobility Pathway’s (EMPath) Mobility Mentoring model. This program provides case management services meant to help families overcome short-term barriers and build income-generating capacity: these include holistic mentoring, incentives for accomplishing goals, free childcare, and classes on money management and career development. In order to quantify the effects of Elevating Families on key outcomes such as parental earnings, children’s school performance, and child protective services involvement, CCNN is partnering with the Wilson Sheehan Lab for Economic Opportunity (LEO) to conduct a randomized controlled trial of the program. LEO is seeking to enroll 600 households to the study over a period of two years, with 300 in treatment and 300 in control. LEO will then track outcomes for study participants for two more years in order to determine long-run program effects.
Intervention
The Elevating Families program provides participants with a bundle of services meant to alleviate generational poverty. Specifically, parent participants work with a Mobility Mentor to develop a Goal Action Plan that addresses five key areas: family stability, health and well-being, financial management, education and training, and employment and career. In order to help motivate their progress and recognize their achievements, participants receive earned incentives upon the completion of their goals. In addition to meetings with Mobility Mentors, participants will have access to free childcare services during mentoring sessions and classes on money management, career development, and parenting skills at the Elevating Families center on CCNN’s new family services campus.
In the first 4 weeks of program enrollment, participants and their Mobility Mentors are required to have completed their first Bridge to Self-Sufficiency: this a holistic evaluation of the barriers and advantages that each participant has in the five focus areas and includes a long-term goal setting component. Roughly two weeks later, they then draft their first Goal Action Plan, a detailed outline of how a participant plans to meet the goals they set in their Bridge to Self-Sufficiency. The action plan includes transitionary goals, pillars, action steps, dates, and incentive amounts. Meetings or contacts between Mobility Mentors and participants then occur every 2 weeks and continue throughout the program, with participants reviewing their progress via a new Bridge to Self-Sufficiency after 6 months. As participants make significant progress towards their goals and are able to demonstrate self-sufficiency, participation is titrated as participants move towards greater independence. When a participant household has been in the program for 12 months, CCNN and program participants begin making a plan for concluding supportive services and transitioning to life without Elevating Families. Participants then follow this plan for the next 12 months, after which point they have completed the program. However, even after they have exited, participants are still given access to limited support from their Mobility Mentors.
Experimental design
Our primary specification estimates the impact of the offer to enroll in Elevating Families or intention-to-treat (ITT) effects of the program on outcomes. The basic specification is
yi = βTreatmenti + γyi0 + αs(i) + εi (1)
where yi is an outcome for enrolled participant i, Treatmenti indicates whether participant i was randomly assigned to the treatment group, yi0 is the outcome of participant i at baseline, and αs(i) are strata fixed effects (i.e., randomization-pair indicators). Within each randomization block, we plan to randomize between pairs of parents that are sequentially matched on employment status, reported income, months worked in the past 12 months, and household income at baseline. The coefficient of interest – β – estimates the average difference in outcomes between treatment and control groups, controlling for baseline outcome levels. This is our preferred specification provided that treatment and control groups do not meaningfully differ (by chance) in observable baseline characteristics.
We will also estimate treatment effects conditional on control vector Xi′ to account for any sampling variation in the composition of treatment and control groups:
yi =βTreatmenti +Xi′δ+γyi0 +αs(i) +εi (2)
Xi′ includes household-level controls that will be selected from the following variables collected in the baseline survey: gender, educational attainment, age, race, ethnicity, home language, parental status, marital status, disability status, employment status, employment history, household income, individual earnings, individual weekly hours worked, and felony arrest history. As described in Chetty et al. (2018), we will select baseline covariates using “the state-of-art approach to penalize overfitting such as LASSO (at present) or the preferred machine learning approach for covariate selection available at the time the analysis is done”.
In addition to the reduced-form estimates obtained by Equations 1 and 2, we are also interested in estimating the causal impact of the program on those who received it, or treatment-on-treated (TOT) effects. To this end, we will instrument for different measures of program take-up using treatment status. Using 2SLS, we will estimate the system:
yi = βDi + γy0i + X′iδ + δs(i) + ϵi (3)
Di = πTreatmenti + ηy0i + X′iν + ζs(i) + υi (4)
where Di is a measure of program enrollment and/or engagement. These exact engagement measure will depend on the data that are collected by CCNN. Similar to Engle et al. (2022), we would ideally measure enrollment as the participant having completed a bridge assessment and set at least one goal with their mentor. Similarly, engagement could be measured as the fraction of months the participant was marked as “actively engaged” throughout the two-year program period. Under reasonable assumptions, β^Di captures the causal impact of engagement with the program on outcome yi. Note that this parameter equals the intent-to-treat parameter (β^ITT ) divided by the regression-adjusted take-up rate (π).
Citations and Footnotes:
Engle, L., Katz, L., & Tebes, J. (2022). Pre-Analysis Plan for “Supporting Pathways Out of Poverty: A Randomized Evaluation of AMP Up Boston”
Chetty, R., S. DeLuca, L. F. Katz, and C. Palmer (2018). Creating moves to opportunity in seattle and king county randomized controlled trial pre-analysis plan.
This approach relies on the assumption that there was no average effect of being offered program enrollment on those who did not take up the program and that the control group was not affected by losing the lottery.
Design details
Cited by
No citation data available for this registration yet.