Association for Public Policy Analysis & Management

One of the primary goals of Head Start is to help low-income parents foster their children’s development. Two important domains of parenting behaviors that are linked to positive child outcomes and might be enhanced by Head Start services include cognitive stimulation (enriching behaviors like reading and discussing math concepts) and socioemotional interactions (a lack of physical punishment). Studies suggest that participation in Head Start is associated with increased parent reading and math-related activities and lower levels of harsh punishment (Ansari et al., 2016; Gelber & Isen, 2013; Puma et al., 2010). However, prior studies have exclusively examined the average effect of Head Start on parenting, ignoring potential heterogeneity of effects across centers that would elucidate optimal program practices for enhancing parenting behavior.

The present study exploits the random assignment and multi-site design of the Head Start Impact Study to 1) quantify variation in program impacts on cognitive stimulation and socioemotional interactions with children and 2) investigate whether programs higher in parent engagement and classroom quality improve parenting practices more effectively. I examine variation in the effects of random assignment to Head Start (an average effect of “intent to treat” [ITT]) using a 2-level random coefficients model with a lagged dependent variable, which estimates variation in effects across 318 Head Start sites.

Findings replicate prior studies showing an average ITT effect of Head Start on examined parenting behaviors. Findings also reveal significant variation in program impacts across a range of parenting behaviors after one year of Head Start, including literacy- and math-related activities, reading frequency and duration, and discipline occurrence and frequency. For instance, across centers, parents randomly assigned to Head Start read for 1.25 additional minutes per sitting on average compared to parents randomly assigned to the control condition. Notably, this effect varied significantly across programs, with a significant standard deviation of 2.21 minutes. Assuming normality, the range of ITT effects for this outcome signifies that in 95% of cases, parents in Head Start centers read for 3.07 fewer to 5.58 additional minutes per sitting than control group parents – demonstrating wide variation in program impacts. Similar ranges were observed for other examined parenting outcomes.

These results demonstrate that some Head Start programs are more effective than others at achieving Head Start’s goal of helping parents to promote their children’s development. Next I ask, what features of Head Start account for variability in program impacts on parenting behaviors? Future analyses will explore whether centers that are high in parent engagement or have high classroom quality are more effective at improving parenting outcomes than programs moderate or low in these features. These questions will be addressed using both ITT models and models using an instrumental variables approach to estimate the local average treatment effect (LATE) among random assignment compliers. Results could be used to help inform Head Start decision-makers about the features of programs that are most salient for promoting positive parenting behaviors.