This paper provides the first analysis of the life course outcomes through their mid-fifties for the participants of the Perry Preschool Project; the authors discuss the experiment design and methodology, how new methods compare with standard inferential methods, compromises in and adjustments to the randomization protocol, the extent of knowledge about departures from the initial random assignment, and treatment effects.
The authors of this paper present an analysis of the life course outcomes through late midlife for the participants of the Perry Preschool Project, which was an experimental preschool program for disadvantaged African American children in the 1960s. The authors used a new survey, conducted more than ten years after the previous follow-up, examining outcome measures analyzed in past studies, such as self-reported employment outcomes and measures of criminal activity sourced from administrative records, in addition to a number of new measures collected to analyze cognition, personality, and biomarkers of health measured through epidemiological exams. The authors developed and applied new statistical methods for analyzing the data, including a formal model of the randomization protocol that captured their imperfect knowledge of it. They also conducted Monte Carlo experiments to compare the false rejection rate of their procedure in practice with that of the standard inferential methods used in previous studies that ignored the essential details of the experimental setup. The authors found long-term treatment effects on crime, employment, health, cognitive and non-cognitive skills, and other outcomes of the Perry participants. Results indicated that treatment effects were especially strong for males, and that improvements in childhood home environments and parental attachment seemed to be an important source of the long-term benefits of the Perry program.