Co-Principal Investigators: Engel, Mimi; Wong, Vivian

Purpose: High quality early childhood education (ECE) has been shown to have profound positive effects on child development, help narrow opportunity gaps, and yield substantial social returns. Yet, more recent research on the impacts of large-scale, present-day ECE programs have yielded mixed results, raising at least 3 key questions for ECE policymakers and researchers: First, there is a need to identify what particular features of preschool programs contribute to children's development. Second, there is concerning evidence that the initial benefits of ECE on academic skills may be short-lived, fading out as children progress through school. Third, understanding the pathway(s) through which a given programmatic feature affects children and on which dimensions—academic, socioemotional, and/or behavioral—is essential to building a robust research base to inform effective ECE policy. The researchers will conduct analyses to answer several research questions to help districts nationwide decide whether providing full-day pre-K is a promising, implementable change districts could make to improve outcomes and address opportunity gaps.

Project Activities: This is a follow-up study to a first-of-its-kind randomized control trial  (RCT) of full- vs. half-day pre-kindergarten (pre-K) in Westminster Public Schools (WPS)—a predominantly non-white, low income district near Denver, Colorado. The RCT randomly offered children a spot in full- or half-day pre-K classrooms for the 2017, 2018, and 2019 school years (N=799). The full-day pre-K (FDPK) study has two phases: Phase 1 was the successful randomization, implementation, and analysis of short-term impacts at the end of the pre-K year for Cohort 1. This follow-up study is for Phase 2, which has two overarching aims: (1) The research team will expand RCT-based causal analyses to Cohorts 2 and 3, add outcomes measuring socioemotional and family impacts, and estimate longer-term effects on academic and behavioral outcomes through at least grade 3. (2) The research team will conduct exploratory analyses of rich, multimodal data to deepen understanding of the causal mechanisms underlying this intervention.

Products: The follow-up study will provide evidence to date about whether   full-day pre-K has lasting effects. Researchers and policymakers will benefit in learning about impacts on academic outcomes, as well as on outcomes in the socioemotional, behavioral, and familial domains. Moreover, the project will yield evidence about whether any apparent effects fade out. Products will include information on whether full-day pre-K had lasting impacts and how additional time might be used.

Structured Abstract

Setting: The study will take place in a public school district located northwest of Denver that serves approximately 10,000 students annually.

Sample: The district's student population is largely non-white (83%), low-income (72%), and has a high percentage of non-native English speakers (34%).

Intervention: The treatment/intervention of interest is an offer of a spot in full-day pre-K vs. an offer in half-day. Fifty percent of the RCT sample was randomized to an offer of full-day.

Research Design and Methods: The FDPK Study employed a block RCT design in pre-K (within first choice of school site). The unit of randomization was the family (same age siblings were randomized together). In pre-K, the RCT achieved successful baseline balance on observed covariates; had low levels of crossover and missing data, neither of which were predicted by baseline observables; and had sufficient power to detect end-of-pre-K effects on academic outcomes in Cohort 1 alone. In this Follow-Up study, the researchers will continue to leverage the randomization to estimate longer term impacts through grade 3–5.

Control Condition: The counterfactual condition was an offer of a spot in a half-day classroom.  An offer of half-day pre-K represented business-as-usual in WPS prior to the onset of the full day pre-K study. While full-day pre-K met 6 hours per day, 5 days a week (30 hours/week), half-day pre-K continued to meet for 3 hours per day, 4 days a week (12 hours/week). Full-day more than doubled the number of hours per week and added over 600 hours in class time across the school year.

Key Measures: When the children were in pre-K, the research team administered the Peabody Picture Vocabulary Test (receptive vocabulary academic outcome), the Early Screening Inventory-Revised (special needs assessment), and Head-Toes-Knees-Shoulders (behavioral self-regulation socio-emotional outcome). In pre-K, the team also gathered over 65 video-recorded classroom observations, with running-record time-use data and Classroom Assessment Scoring System (CLASS©) scores. The team collected parent and teacher surveys and WPS administrative data (FRPL, special education, and ELL designations, as well as attendance). The state administers the Colorado Measures of Academic Success in grades 3-5 (reading, math, social studies, and science).

Data Analytic Strategy: Depending on the outcomes, the researchers will use OLS/ logistic regression models to estimate intent-to-treat (ITT) effects. Because some families (19%) do not have to take-up their lottery offers, the researchers will use two-stage least squares to estimate complier-average-treatment-effects (CATE).

Cost Analysis: Adding full-day pre-K programming requires significant up-front costs, and it is not obvious how public education agencies should invest their scarce resources. As a result, district and state leaders need information about the cost-effectiveness of offering full-day pre-K. The research team will use an ingredients approach to calculate costs. Once the total costs have been compiled, the researchers will connect this information with the longer-term impact estimates from the proposed Phase 2 Follow-Up Study to generate cost-effectiveness ratios, which measure the cost per unit of outcomes produced by the intervention. These ratios allow policymakers to make cost-effectiveness comparisons across interventions.