Panel: Heterogeneous Treatment Effects: Theory and Application

Saturday, November 10, 2012: 10:15 AM-11:45 AM
McKeldon (Sheraton Baltimore City Center Hotel)

*Names in bold indicate Presenter

Organizers:  Greg J. Duncan, University of California, Irvine
Moderators:  Hirokazu Yoshikawa, Harvard University
Chairs:  George Farkas, University of California, Irvine

The proposed panel features early results from a recently-awarded program project (P01) from NICHD which has the five-year goal of conceptualizing and testing for the heterogeneous nature of the impacts of various educational interventions across developmental stages, using the latest experimental and quasi-experimental methods. The grant brings together an interdisciplinary team of researchers from Education, Psychology, Economics, Sociology, and Criminology. The four papers in the panel address the intersection of policy and development to better understand how interventions affect youth across different developmental stages. They fit within the conference theme, “Policy Analysis and Public Management in an Age of Scarcity: The Challenges of Assessing Effectiveness and Efficiency”, by evaluating educational interventions using rigorous methods to determine how policies and educational interventions are the most effective for children and adolescents. Our common conceptual approach assumes that children and youth profit from education-related interventions to varying degrees, for two fundamental reasons. First is what we call stage/policy fit. Children in different developmental stages vary in their responses to policies because of differences in the fit between policy-induced changes in children’s immediate environments and the accomplishment of stage-salient developmental tasks. For example, Eccles et al. (1993) argue that the primary/middle-school model of education structure is inferior to an integrated K-8 structure because middle schools are ill-matched to the emerging developmental demands of children as they transition to adolescence. But there is also substantial variation in treatment impacts across children within a given stage (Imbens & Angrist, 1994), which we call child/policy fit. Early childhood interventions such as Head Start and Early Head Start are geared toward providing learning experiences to children whose family environments are unlikely to provide enough of them. Thus, they “fit” better, and likely generate larger impacts, for children from economically disadvantaged than advantaged circumstances. A contrasting view is the Cunha-Heckman skill production model and its “skill begets skill” assumption that education interventions are most productive for children with the highest levels of skill. The papers in the proposed panel address the different hypotheses mentioned above through a conceptual understanding of the heterogeneous treatment effects of educational interventions and by using rigorous quasi-experimental methods to evaluate treatment effects. Paper 1 is a conceptual paper that focuses on the different theories and heterogeneous treatment effects found in educational interventions across different developmental stages of childhood and adolescence. Paper 2 uses the Head Start Impact Study to examine treatment effects across achievement distributions in contrast to focusing on the average treatment effect of the intervention. Paper 3 seeks to understand the distributional effects of an “Algebra for All” curriculum requirement for all 8th graders in California on students’ achievement on the state high-school exit exam. Finally, paper 4 examines the effect of process child care quality, such as teacher-child interactions, on a low-income sample of preschool children’s academic achievement. All four papers address policy implications of interventions.

The Effect of Early-Grade Retention On Student Outcomes: Regression-Discontinuity Estimates From Florida's Test-Based Promotion Policy
Guido Schwerdt, LFO Institute for Economic Research and Martin West, Harvard University

Understanding Variation In the Impacts of Human Capital Interventions On Children and Youth
Greg J. Duncan, Deborah Lowe Vandell and Sara Wakefield, University of California, Irvine

Who Benefits From Detracking? A Distributional Analysis
Thurston Domina, Andrew M. Penner, Emily K. Penner, Marianne Bitler and AnneMarie Conley, University of California, Irvine

Child Care Quality and Academic Achievement: Results from PCER
Anamarie Auger1, George Farkas1, Greg J. Duncan1, Margaret Burchinal2 and Deborah Lowe Vandell1, (1)University of California, Irvine, (2)University of North Carolina at Chapel Hill

See more of: Education
See more of: Panel