Schweinhart, L. J., & Weikart, D. P. (1997). Lasting differences: The HighScope Preschool Curriculum Comparison study through age 23 (Monographs of the HighScope Educational Research Foundation, 12). Ypsilanti, MI: HighScope Press.
(Apparently the full report is a monograph that must be purchased online, but a free summary can be found here:
Although this study was focused on early-childhood education, the findings seem quite stark. Two quotes from the summary:
- "Scripted teacher-directed instruction, touted by some as the surest path to school readiness, seems to purchase a temporary improvement in academic performance at the cost of a missed opportunity for long-term improvement in social behavior."
- "These findings constitute evidence that early childhood education works better to prevent problems when it focuses not on scripted, teacher-directed academic instruction but rather on child-initiated learning activities...These findings suggest that the goals of early childhood education should not be limited to academic preparation for school, but should also include helping children learn to make decisions, solve problems, and get along with others."
There were two other long-term studies started in the 1970's:
 Karnes, M. B., Schwedel, A. M., & Williams, M. B. (1983). A comparison of five approaches for educating young children from low-income homes. In Consortium for Longitudinal Studies, As the twig is bent . . . lasting effects of preschool programs (pp. 133-170). Hillsdale, NJ: Erlbaum
 Miller, L. B., & Bizzell, R. P. (1983). The Louisville experiment: A comparison of four programs. In Consortium for Longitudinal Studies, As the twig is bent . . . lasting effects of preschool programs (pp. 171-199). Hillsdale, NJ: Erlbaum.
All three included the Direct Instruction model—which offered scripted, teacher-directed academic instruction—and a Nursery School model, in which children initiated their own learning activities with minimal teacher support. The High/Scope study included the High/Scope model, in which children initiated learning activities with substantial teacher support. The Louisville and Illinois studies included several additional teacher-directed models and the Montessori model, which encouraged child-initiated activities with didactic materials.
These three studies found that children in Direct Instruction programs intellectually outperformed children in child-initiated-activities programs during and up to a year after the preschool program, but not thereafter. In the Louisville study, the Nursery School children showed higher verbal-social participation and increased more in ambition and aggressiveness than did the Direct Instruction children, but both groups scored lower than their peers on inventiveness. In the Illinois study, 78% of the Nursery School group, but only 48% of the Direct Instruction group and 47% of the no-program group graduated from high school.Finally, a more recently concluded long-term study conducted in Milwaukee, Wisconsin, is described in:
 Dohrmann, K. R. Outcomes for students in a Montessori program: A longitudinal study of the experience in the Milwaukee Public Schools. (2003). Association Montessori Internationale/ USA, 410 Alexander Street, Rochester, New York. 14607-1028. This paper is freely available online at http://www.montessori-ami.org/research/outcomes.pdf
This study looks at 201 Montessori students who completed fifth grade between 1990-1994, and graduated in 1997-2005, compared with a carefully matched control sample of non-Montessori students. Montessori schools emphasize child-centered education with a strong emphasis on self-directed learning. Dohrmann writes:
A significant finding in this study is the association between a Montessori education and superior performance on the Math and Science scales of the ACT and WKCE. In essence, attending a Montessori program from the approximate ages of three to eleven predicts significantly higher mathematics and science standardized test scores in high school.These findings are further corroborated by a short-term study published in Science, in 2006, which also finds significant academic/cognitive as well as social/behavioral benefits to a Montessori education:
 Lillard A, Else-Quest N (September 2006). "The early years. Evaluating Montessori education". Science 313 (5795): 1893–4. (See http://www.montessori-science.org/montessori_science_journal.htm for link to full paper.)
I think that the fact that four long-term studies done on this topic (I am not aware of any other such studies and would welcome pointers) all show positive impacts of starting children on an autonomous and self-directed learning path is pretty compelling. Don't you?