“Elusive Longer-Run Impacts of Head Start: Replications Within and Across Cohorts”, 2020-08-24 (; backlinks; similar):
Using an additional decade of data from the National Longitudinal Survey of 1979 Children and Young Adults (CNLSY), this study replicated and extended 2009’s evaluation of Head Start’s life cycle skill formation impacts in 3 ways:
Extending the measurement interval for Deming’s adulthood outcomes, we found no statistically-significant impacts on earnings and mixed evidence of impacts on other adult outcomes.
Applying Deming’s sibling comparison framework to more recent birth cohorts born to CNLSY mothers revealed mostly negative Head Start impacts.
Combining all cohorts showed generally null impacts on school-age and early adulthood outcomes.
[Keywords: early childhood, evaluation, human development, longitudinal studies, regression analyses]
…We found that extending the measurement period for Deming’s cohorts and early-adult outcomes decreased the estimated impact on the adulthood summary index (ASI) of Head Start attendance relative to not attending any preschool program 0.23–0.17 SD, standard error (SE) = 0.07. Of the longer-run outcomes we were able to consider, the largest impact of attending Head Start was on years of completed schooling (0.30 years; SE = 0.15). This is notable and, taken by itself, could indicate a sizable return on investment for the program. However, we estimated relatively small, non-statistically-significant impacts on gains on other later life outcomes, including college graduation and earnings. 6 For the children born after Deming’s cohorts, Head Start impacts were mostly null and sometimes negative. In fact, positive impacts on ASI generated by Deming’s cohorts were matched by nearly symmetric negative impacts for the complement cohorts (−0.15 SD; SE = 0.07). For the final sample that combined the 2 sets of cohorts, the point estimate of Head Start’s impact on the ASI was close to zero and not statistically-significant.