Impacts of Mentoring As an Enhancement to Education and Training Programs for Young Parents: Preliminary Impact Findings from the Young Parents Demonstration Evaluation
*Names in bold indicate Presenter
The YPD grantees are required to develop an enhanced service intervention, either mentoring or employment and training services, in addition to a set of existing basic services. Individuals are randomly assigned with equal probability to either the basic service group (the control group) or a treatment group that receives the enhanced mentoring and basic services. Since the inception of the demonstration in 2009 (through the end of enrollment in March 2015), over 3,500 young and expectant parents were randomly assigned to treatment and control groups.
This paper will provide impact estimates for the 14 YPD grantees that concluded random assignment in December 2012, with a particular focus on estimating impacts (by site and in aggregate) for YPD sites providing mentoring services as the enhanced service intervention. Impacts will be estimated on participant employment, earnings, and hourly wage rates up to 24 months after random assignment. The use of random assignment to treatment status permits the use of analysis of variance for estimating program impacts, but regression analysis and logistic analysis (for dichotomous outcomes such as employment status) will also be used to improve precision. Estimates will also be made for subgroups defined by participant characteristics, such as sex, race, ethnicity, and age, and the mentoring approach used.
A unique feature of the evaluation is that impact estimates are developed from three separate data sources – administrative program records, telephone surveys of participants, and nationally available state wage records collected for the unemployment insurance system. Thus, the paper not only identifies the impacts of the mentoring programs, it also enable us to see how outcome data and estimated impacts vary by source of data.
 In a forthcoming Evaluation Review article, Barnow and Greenberg show that earnings impact estimates can vary by as much as 500 percent depending on the source of data. See Burt S. Barnow and David Greenberg (forthcoming). Do Estimated Impacts on Earnings Depend on the Source of the Data Used to Measure Them? Evidence from Previous Social Experiments. Evaluation Review.