Association for Public Policy Analysis & Management

As governments ramp up their commitment to expanding opportunities for pre-k, a central policy question concerns whether to fund universal pre-k programs that serve all children or targeted programs that focus on disadvantaged children.  One essential lens through which to address this policy issue is a child development perspective.  Extensive literature documents the potential for classroom composition- that is, the characteristics of children in the classroom- to affect individual children’s learning and development, but the vast majority of this work assessed primary or secondary school students.  Our own earlier work found that young children in classrooms with peers with greater academic and language skills gained more in their own language and math skills through the preschool year.  In the current paper we use a larger and more representative sample to expand on this work, arguing that peer characteristics in preschool classrooms might affect individual children through various pathways.  We will test three models.  First, we hypothesize that greater classroom poverty will be associated with lower gains in learning by children through the preschool year.  Second, we argue that the negative effects of classroom poverty will function through a lower level of average cognitive skills among peers, leading to lower growth in children’s cognitive skills.  And third, we argue that greater classroom poverty and lower peer skills will be associated with lower quality instruction, in turn inhibiting individual children’s learning.

Data will be drawn from the Preschool Curriculum Evaluation Research (PCER) study, which followed approximately 3,000 children clustered within approximately 300 preschool classrooms (from Head Start, public, and private preschool centers) from the beginning of their preschool year through kindergarten. Data were derived from teacher and parent interviews, teacher surveys, classroom observations, and direct child assessments in the fall and spring of preschool and spring of kindergarten.  Children were racially and ethnically diverse, and primarily from poor, low, and middle income families.  Children’s language, reading, and math skills were assessed the fall and spring of preschool and spring of kindergarten using well-validated direct assessments.  Children’s family income and academic skill scores were used to create classroom composites of peer poverty and of peer skills for each child.  Teachers’ instructional quality was assessed through direct observations using the ECERS-R; based on recent research we will use the instructional practices subscale validated by Burchinal and Gordon and others. 

Analyses for this work incorporate multilevel autoregressive path models, adjusting for children’s prior academic skills and a host of child, family, teacher, and classroom covariates to help adjust for selection bias.  Initial results found that the proportion of poor children in the classroom was associated with lower growth in children’s skills through the preschool year.  Moreover, the academic skills of classroom peers were further associated with growth in individual children’s skills through the year.  Additional analyses will test for the significance of indirect effects and will assess the role of teacher instructional quality.  Results will have implications for policy decisions regarding targeted versus universal pre-k programs, and will expand theoretical models of classroom composition.