View article

Although there is a broad hi-partisan support for Head Start, the evidence of positive longterm effects of the program is not overwhelming. Using data from the National Longitudinal Survey's Child-Mother file, we examine the impact of the program on a range of child outcomes. We compare non-parametric estimates of program effects with estimates from parametric models that control for selection by including mother fixed effects. This comparison suggests that studies that ignore selection can be substantially misleading; it also suggests that the impact of selection differs considerably across racial and ethnic groups. After controlling for selection, we find positive and persistent effects of participation in Head Start on the test scores of white and Hispanic children. These children are also less likely to have repeated a grade. We find no effects on the test scores or schooling attainment of African-American children. White children who attend Head Start are more likely to receive a measles shot, while African-American enrollees receive measles shots at an earlier age. African-American children who attend Head Start are also taller than their siblings. In a sample of the children's mothers, we find evidence that whites who attended Head Start as children are taller and have higher AFQT scores than their siblings who did not