Instrumental Variable (IV) Analysis of Head Start Enrollment on Children's Comprehensive Outcomes

Purpose:

Views on the effectiveness of Head Start vary. A number of observational studies find substantial short- and long-run impacts on test scores and other outcomes. By contrast, a recent randomized evaluation—the Head Start Impact Study (HSIS)—found small impacts on test scores that fade quickly as well as significant impact on children’s limited outcomes. The purpose of this study was to determine (1) why children who were assigned to Head Start did not enroll Head Start but attended other care programs? (2) the causal impact of Head Start enrollment on children’s comprehensive outcomes (cognitive, social-emotional, health) with an instrumental variable (IV) analysis.

Methods:

The current study used 3780 children from the HSIS data. Children’s cognitive outcomes were Peabody Picture Vocabulary Test and Woodcock-Johnson III Tests of Achievement test scores. Social outcomes were child-teacher relationship skills (Pianta, 1992), social skills and positive approaches to learning, social competencies checklist, behavioral problems. Health outcomes were i) whether the child has a regular medical care provider, ii) whether the child has received dental care, hearing check-up, and vision check-up. We used the assignment to the Head Start program as our IV. First, to determine which baseline characteristics affected children’s compliances with the original assignment, several sets of logistic regression analysis were conducted including quality of care (for similar variable list, see Klein & Walters, 2016, pp.1803-1804): Second, to determine the causal impact of Head Start enrollment on children, instrumental variable [IV] analyses were conducted for children’s outcomes. The current study used STATA 14/SE.

Results: Among 3780 children who remained in the HSIS data after 1 year of Head Start enrollment, 3155 children complied with the original assignments and 625 children did not comply with the assignments. Head Start assigned children were more likely to comply with the original assignment (β= - 0.382, SE = 0.08^***). Older children (β=-0.248, SE = 0.06^***), children with special needs (β=-0.199 SE = 0.09^*), black children (β=-0.158, SE = 0.08^*), and children living in a higher family income (β=0.081 SE = 0.02^***) were less likely to comply with the original assignments. Quality (β=-0.199 SE = 0.03^***) was significantly associated with compliances. Head Start had a positive causal impact on children’s cognitive (PPVT, WJIII applied problem solving, pre-academic skills, spelling, and word identification; Beta (SE) = 0.180 (.04)***, 0.200 (.05)***, 0.315 (0.04)***, 0.224 (.05) ***, 0.330 (.05)***), social-emotional (behavioral problem = -.111 (.05)* and close child parent relationship = .114 (.05)*) and health outcomes (dental, vision and hearing) after 1 year of Head Start enrollment.

Conclusion/Implications: The instrumental variable (IV) analysis determined Head Start’s causal impacts on children even after taking care of observed and un-observed variances related to Head Start assignment and children’s outcomes for comprehensive outcomes (cognitive, social-emotional, and health). Equity in access to quality Head Start program is important to achieve the ultimate goal of Head Start program.  Further, children’s pre-school programs should consider the needs of child and family characteristics in the context of all available (Head Start and other alternative) programs.