Association for Public Policy Analysis & Management

Prior research suggests that having a curriculum is one of the most important aspects of quality in ECE (Duncan & Magnuson, 2013) because it provides a framework to guide interactions and activities in the classroom. It is important that curricula research discern the conditions under which certain curricula work best for certain children. Given the importance of preschool curricula in high-quality center-based case, it is important to consider not just its average impact, but the way it impacts the entire distribution of children’s outcomes.

We might expect several different distributional patterns for the effects of curricula, which may result from a number of characteristics including baseline characteristics of children, local area contexts, and program experiences. For example, curricula effects on posttest scores could be larger in the lower tail of the test score distribution than in the upper tail or vice versa, which could have important policy implications for understanding who benefits most from the curricula (Schochet, Puma, & Deke, 2014). This study will provide a new way for thinking about the effects of curricula on the distribution of children’s outcomes by reevaluating the same experimental data from the Preschool Curriculum Evaluation Research study to test whether the original null mean effects found are concealing important distributional differences. The experimental nature of the PCER study allow for causal estimates of this relationship. Prior research has only examined the average impacts of experimental curricula—this study uses distributional methods to identify variation in effects that have been previously hidden.

To examine the effect of academically targeted curricula on the distribution of children’s achievement, I will take advantage of randomized assignment to treatment (academically targeted curricula) and control (business-as-usual curricula) classrooms to estimate quantile treatment effects (QTE; Firpo, 2007). In the context of experimental data, QTE are estimated by calculating the difference in the two marginal distributions and are identified at each quantile in an analogous logic to average treatment effects (Bitler et al., 2014). By comparing test scores at a number of quantiles, I am able to observe the effect of academically targeted curricula on different portions of the distribution. To address the potential lack of balance on baseline scores in the fall of preschool, I use an inverse propensity score weighting approach as a nonparametric first step (Firpo, 2007), which allows me to balance baseline test scores across the two groups and to account for differences in the likelihood of being assigned to a classroom with academically targeted curricula or business as usual curricula. Preliminary results indicate that the curricular interventions in PCER have heterogeneous impacts rather than uniform impacts on preschool children and that the largest achievement gains are for children at the bottom of the distribution.

Understanding variation in curricula effects is critical for facilitating the most efficient use of limited resources, by informing decisions about how best to targeted specific curricular programs, and suggesting ways to improve the design or implementation of the programs for high quality center-based child care settings.