Favorable Long-Term Effects of Head Start

Summary of working paper 8054

Head Start generates long-term improvements in important outcomes such as schooling attainment, earnings, and crime reduction.

Begun in 1965 as a summer program for disadvantaged children, Head Start has grown into a school-year preschool program enrolling more than 800,000 disadvantaged children at a cost of about $5,400 per child. Because Head Start was designed as an early intervention to remedy deficiencies that handicap disadvantaged children when they start school, it addresses both academic and physical problems. The federal guidelines charge the over 1,400 local programs with maintaining a nurturing learning environment, providing nutritious foods, and "empowering" parents through conferences and home visits. The program also requires active monitoring of each child's medical, dental, and mental-health referrals and follow-ups. However, much of the current focus in policy discussions is on the question of whether Head Start does in fact enhance longer-term child outcomes, such as educational attainment and earnings.

Previous research suggests that Head Start improves early test scores but that these improvements tend to "fade out" by third grade. It is possible however, that Head Start could improve long-term outcomes even if it did not increase test scores, since success in life generally reflects more than cognitive ability. Little evidence has been available on this important question, however, because previous studies have not followed Head Start children long enough to assess long-term effects.

In Longer Term Effects of Head Start (NBER Working Paper No. 8054), authors Eliana Garces, Duncan Thomas, and Janet Currie find that Head Start generates long-term improvements in important outcomes such as schooling attainment, earnings, and crime reduction. They find that disadvantaged whites who had been enrolled in Head Start were more likely to graduate from high school and to have attended college than siblings who did not. White children of high school dropouts also had higher average earnings between the ages of 23 and 25 if they attended Head Start. African-Americans who attended Head Start were "significantly less likely to have been booked or charged with a crime" compared to siblings who did not participate in Head Start. Finally, the authors find that male African-Americans were more likely to complete high school and to participate in the labor force if they had attended Head Start.

The data for this study come from the Panel Study of Income Dynamics (PSID) which began in 1968 with a survey of 4,802 households containing 18,000 individuals. In 1995, adults in the PSID who were age 30 or younger were asked whether they had ever been enrolled in Head Start or any other preschool or daycare program. The adults in the PSID survey had been "followed" since childhood, and also answered questions about labor force participation, earnings, schooling, and criminal activity. There are roughly 4,000 respondents in the survey for whom information about both preschool experiences and these adult outcomes is available. These data offer a unique opportunity to assess the long-term effects of Head Start.

Still, the Head Start questions refer to events that took place many years ago. Aware that survey participants might have problems remembering preschool attendance, the authors compare self-reported PSID Head Start enrollment rates and the racial composition of enrollments in the PSID with figures reported by the Head Start Bureau. They find no evidence that poor memories contaminate their results.

Further, because Head Start respondents are not selected randomly, it is possible that the outcomes of Head Start children differ from those of other children because of their family backgrounds and characteristics, rather than because of participation in Head Start. Although controlling for such observable characteristics as parental education, attendance at other preschools, family structure, and gender eliminates some of this bias, there are important determinants of both child outcomes and Head Start participation that remain unobserved, the authors conclude. In general, children who attend Head Start come from backgrounds that engender worse outcomes than those of observationally similar children. So, in addition to controlling for observable characteristics, the authors compare children who attended Head Start with their own siblings who did not. Any unmeasured background characteristics shared by the siblings are controlled for by using this study design.

One potential drawback to sibling comparisons is that there may be "spillover" effects from one sibling to another. Indeed, the authors find that in families in which an older child attended Head Start and a younger one did not, the younger sibling was less likely to have been booked or charged with a crime. This finding suggests that sibling comparisons may provide a lower-bound estimate of the true effects of Head Start on crime reduction, since the program benefitted both the sibling who attended and the one who did not.

-- Linda Gorman