Association for Public Policy Analysis & Management

Since its inception in the 1960s, Head Start has always had a whole-family, antipoverty approach. In hopes of reducing social disparities over time, Head Start sought to provide early childhood education (ECE) programming for children as well as a host of services to enhance parents’ childrearing skills and their own educational progress. Yet Head Start agencies have also had considerable flexibility in how they structure their dual-generation programming. In the face of limited resources, this study examines the extent to which Head Start makes tradeoffs between providing services for children versus parents and the implications for parent and child wellbeing.  

Administrative data for this study come from Head Start Program Information Reports (PIR), which are mandatory, agency-level, annual surveys. We include all agencies from 2002-2014, which results in approximately 2,700 Head Start grantees/agencies with a total of around 37,000 program-by-year observations. We will also link the PIR data to the National Longitudinal Survey 1979 using FIPS county code to attain parent and child outcomes. A similar approach was employed by Currie & Neidell (2007) and resulted in a sample of approximately 880 children who attended Head Start that matched to the PIR data.

We first characterize trends in parent services versus child services in Head Start over the past decade by conducting mean-level difference tests at the national level in each year. For parents, we focus primarily on programming that directly targets their skill development, including adult education, job training, English as a Second Language training, and parenting education. For children we focus on ECE quality, including staff-child ratios, staff qualifications, and teacher salaries.

We then examine whether investment in parent services is associated with reductions in child services and vice versa. To do so, we employ a within-program differences modelmodeling changes within agency over time, thus accounting for all unobserved, time-invariant factors that might bias estimates of the associations of interest. We will then use a similar approach to investigate how varying combinations of parent and child services differentially relate to child wellbeing (language skills, reading skills, mathematics and behavior problems), as well as and parent outcomes (mother’s level of education, income, job skills).

Preliminary analysis indicate that there were slight increases in the proportion of parents receiving services at Head Start and larger increases in classroom quality. For example, in 2014, 13% of parents received adult education services (vs. 11% in 2003) and the the proportion of teachers with a Bachelor’s degree or higher increased from approximately 30% in 2002 to over 70% in 2014. We find little evidence that support for one generation (i.e., parents) comes at a cost for supporting the other generation (i.e., children). The only exception is that agencies that increased the proportion of parents who participated in adult education services had increases in the proportion of teachers with Bachelor’s degree and decreases in overall teacher salary. Future work will investigate how varying combinations of parent and child services differentially relate to child and parent outcomes.