Association for Public Policy Analysis & Management

As federal, state, and local governments seek to expand access to publicly funded preschool programs in the United States, policymakers and program leaders are eager to determine what program features increase impacts on child development and/or long run success. Two program features that act through children’s classroom experiences rather than parenting environments—the degree to which programs focus on academic content and the quality of teacher child interactions—represent malleable program components that might be related to larger program impacts. Prior research has used non-experimental methods to try to establish a causal link between program inputs and child outcomes where child and family characteristics associated with selection into preschool programs confound interpretation of associations as impacts.

In this paper, we capitalize on the experimental design of the Head Start Impact Study (HSIS) to measure the impact of Head Start on equivalent predicted subgroups of treatment and control group members—both children predicted to participate in Head Start classrooms with high quality teacher-child interactions or high levels of academic content and children predicted to participate in Head Start classrooms with lower levels of these factors.  We then estimate impacts on these matched subsamples of treatment and control group members as with any experimental subgroup analysis. The symmetry of the prediction procedure ensures that equivalent subgroups are compared and guarantees that the resulting impact estimates for predicted subgroups are free from differential selection bias or other sources of internal bias.  We then convert findings into estimates of impact on actual high-input and low-input subgroups using plausible assumptions and compare results to estimate how much those inputs matter to impact magnitude.

We use a composite of 31 variables to measure the quality of teacher-child interactions in Head Start classrooms. Eight of these variables come from the Early Childhood Environment Rating Scale (ECERS) and 23 from the Arnett Caregiver Interaction Scale. Our measure of academic content draws on 19 teacher-reported variables to determine the frequency of academically-focused activities that children experience in the classrooms. Children’s outcomes include pre-academic skills measured using the Peabody Picture Vocabulary Test (PPVT-III, adapted), and the Woodcock-Johnson III (WJ3) Letter-Word Identification and Applied Problems subsets. Children’s social and emotional skills are measured using the Total Child Behavior Problems scale.  

We find some evidence that Head Start’s impact varies by both teacher-child interaction levels and levels of academic content. Exposure to programs with high levels of teacher child interactions leads to generally higher impacts on a range of measures of pre-academic skills.  Among children who enter Head Start at age 4 high levels of academic content produce stronger positive impacts on children’s preacademic skills but not children’s behavior.  Among three-year-old entrants high levels of academic content generate less favorable impacts on children’s problem behaviors and vocabulary development.  Thus, although high levels of teacher-child interactions appear to be beneficial for both ages of children, high levels of academic content appear to be beneficial to older children but detrimental to younger children.  These findings are discussed in the context of current policy and program improvement efforts.