|Title:||A Lottery-Based Efficacy Study of the District of Columbia Public Prekindergarten Program|
|Principal Investigator:||Greenberg, Erica||Awardee:||Urban Institute|
|Program:||Early Learning Programs and Policies [Program Details]|
|Award Period:||5 years (09/01/2021 – 08/31/2026)||Award Amount:||$3,800,000|
Co-Principal Investigators: Hawkinson, Laura; Kahn, Sherylls; Monarrez, Tomas; Weiland, Christina
Purpose: While the evidence on short- and longer-term prekindergarten efficacy has grown in recent years, nearly all existing research comes from studies of programs for 4-year-olds. Increasing policy interest in programs for 3-year-olds, persistent disparities in access, and theories and evidence from economics, neuroscience, and developmental psychology highlight the need for new research so that policymakers, practitioners, and the public can make informed choices about investments in children's success. The purpose of this study is to evaluate the District of Columbia's large-scale 3-year-old PreK program with rigorous methods based on the program's centralized admissions lottery.
Project Activities: Researchers will estimate causal impacts by leveraging a centralized admissions lottery that matches some children with seats in the program and places others on a waiting list. Researchers will examine the impacts of 3-year-old PreK (PK3) on key outcomes measured in the fall and spring of the 3-year-old year and spring of the 4-year- old year using valid and reliable assessments administered by the study team. Analyses will also extend from kindergarten through third grade using outcomes available in administrative data. The team will assess short-term outcomes using valid and reliable measures of children's early language, literacy, math, social-emotional, and executive function skills. They will also measure longer-term outcomes using administrative data on persistence in public schools, in-seat attendance, in-grade retention, special education placement, and third-grade math and English language arts scores. Researchers will include outcomes from a new measure of racial attitudes to gauge impacts of a diverse PreK program on children's racial attitudes. Short-term outcomes will be measured for two cohorts of children, while longer-term outcomes available in administrative data will involve additional cohorts for greater statistical power. Researchers will also measure impact variation by specified moderators and mediators, estimate the program's cost and cost effectiveness, and conduct qualitative interviews with key stakeholders following a mixed-methods approach.
Products: The research team will generate new evidence on the efficacy of PK3 at scale. A robust dissemination plan will include distributing this evidence to policymakers, practitioners, researchers, and families through products tailored to each audience, including stakeholder briefings; blog posts; policy briefs or fact sheets that distill reports and peer-reviewed journal articles; presentations at policy, practice, and research conferences; and a concluding convening aimed at bringing multiple audiences together in service of a stronger research-to-PreK policy pipeline.
Setting: This study will take place in the District of Columbia.
Sample: The sample includes approximately 1,200 children in each annual lottery who apply to DC PK3 and either win a seat (the treatment condition) or do not (the control condition). Children from two lotteries (2022 and 2023) will be observed during the 3- and 4-year-old PreK years; children from earlier lotteries (2016 through 2019) will be observed from kindergarten through third grade. The sample is sociodemographically diverse. The majority are children of color, nearly half are at-risk (meaning they receive public benefits, are homeless, or are in foster care), and smaller shares receive special education or qualify as English language learners. The sample is intended to represent all 3-year-olds district-wide, given universal eligibility for PK3.
Intervention: The intervention is authorized by the Pre-Kindergarten Enhancement and Expansion Amendment Act of 2008 and consists of a full school-day, full school-year, tuition-free early learning program with universal eligibility, based in public schools. In traditional public schools, lead teachers are required to hold at least a BA, and assistant teachers must hold a Child Development Associate (CDA) or equivalent. The maximum class size for 3-year-olds is 16. Programs typically use the Creative Curriculum and/or Tools of the Mind, but schools may select among 14 approved curricula in addition to Dual Language Immersion and Montessori models. In public charter schools, school operators set their own teacher education, curriculum, and assessment standards. All PreK teachers engage in ongoing professional development, undergo annual quality observations (using the CLASS), and benefit from pay parity with those in K-12. Eligibility for the program is universal, and more than 70% of District 3-year-olds enroll.
Research Design and Methods: Researchers will use a lottery-based design to evaluate program impacts through the end of third grade using study-collected and administrative data. They will leverage all random variation in enrollment generated by the lottery through a deferred acceptance method akin to a randomized controlled trial with imperfect compliance. Researchers will work with two cohorts of children, families, and other study participants. Impact, implementation, cost, and cost effectiveness analyses will begin in year 2 and continue through year 5. Researchers will recruit Cohort 1 children at the PreK3 lottery in spring of year 1. In year 2, researchers will collect data from the Cohort 1 sample in fall and spring and recruit Cohort 2 children at the PreK3 lottery in spring of year 2. In year 3, researchers will collect data from the Cohort 2 sample in fall and spring and collect follow-up data from the Cohort 1 sample. Researchers will pair child assessments with parent surveys, related provider surveys, and key stakeholder interviews to document PK3 moderators, mediators, and features of implementation fidelity critical for interpreting estimates of PK3 impact. Dissemination and stakeholder engagement will occur throughout the course of the project, with concentrated efforts in the final year.
Control Condition: The control condition is not receiving a match in the PreK admissions lottery and being placed on a waiting list. Control group students may be non-randomly matched to a PreK program at a later time (control crossovers), or find an outside program including center-based preschool, home-based childcare, informal care, or parental care (control compliers). Researchers will focus on effects for compliers.
Key Measures: To measure short-term outcomes at the end of PreK, the research team will administer early language, literacy, and math assessments (the Peabody Picture Vocabulary Test and Letter-Word Identification and Applied Problems subscales of the Woodcock-Johnson Test); measures of social-emotional and executive function skills (the Forward Digit Span, Pencil Tap task, and Task Orientation Questionnaire); and a new Racial Attitudes Index. Outcomes from kindergarten through third grade will be measured using administrative data on persistence in public schools, in-seat attendance, in-grade retention, special education placement, and third- grade math and English language arts PARCC scores.
Data Analytic Strategy: Researchers will conduct mixed-methods analysis of all study-collected and administrative data. Quantitative analysis will involve casual impact estimation in an instrumental variables (2-stage least squares) regression framework. They will estimate PK3 main effects and impact heterogeneity. Qualitative analysis will use the constant comparative method.
Cost Analysis: Researchers will conduct cost and cost-effectiveness analyses using the ingredients method and CostOut tool. Provider surveys and key stakeholder interviews, along with existing literature, will inform development of a list of ingredients in DC PK3 and counterfactual conditions. For the main analyses, researchers will use national prices for identified ingredients drawing from the CostOut database, and they will validate estimates using local pricing.