Relatively few studies provide rigorous estimates of the long-term effects of large-scale public preschool programs, and their findings vary greatly. This study investigates the effects in third through tenth grade of New Jersey's Abbott preschool program which has many of the features and contexts hypothesized to mitigate fadeout. The program was designed and implemented in the context of a Court mandated systemic reform of education and its funding from preschool through high school. We describe in detail the program's features including an extensive, multitiered continuous improvement system. Sample size for analyses ranged from 426 to 785 depending on the grade and assessment. Participants were primarily Black and Hispanic students living in 31 communities with high concentrations of poverty. Inverse weighting by propensity scores was employed with multiple imputation for missing data to estimate effects on achievement, grade retention, and special education. Substantial positive effects were found in language arts and literacy, mathematics, and science on statewide assessments. Effects did not fade after grade 3. Achievement effects appear to be larger for 2-year than 1-year of the preschool program. Grade retention was significantly lower through grade 10. Effects on special education placement were imprecisely estimated but consistent with other findings of reduced special education. Results were robust with respect to alternative methods to control for measured and unmeasured differences between preschool and comparison groups and for missing data. This study adds to the evidence on preschool program features and contexts associated with long-term effects.

相对较少的研究提供了对大型公共学前教育计划的长期影响的严格估计，并且其发现差异很大。这项研究调查了新泽西州Abbott学前班计划三年级至十年级的影响，该计划具有许多可以减轻淡入淡出的功能和背景。该方案是在法院授权对教育及其从学龄前到高中的资金进行系统改革的背景下设计和实施的。我们将详细介绍该计划的功能，包括一个广泛的，多层的持续改进系统。分析的样本量范围从426到785，具体取决于等级和评估。参与者主要是居住在31个贫困集中的社区的黑人和西班牙裔学生。通过倾向得分的反加权，对缺失的数据进行多次插补，以估计对成就，成绩保持和特殊教育的影响。在全州评估中，语言艺术和读写能力，数学和科学领域都发现了巨大的积极影响。三年级以后，效果没有减弱。学前班2年的成绩似乎比1年的成绩要大。到10年级为止，年级保留率明显降低。对特殊教育安置的影响估计不精确，但与减少特殊教育的其他发现一致。在控制学前班和对照组之间测得的和未测得的差异以及数据丢失的替代方法方面，结果是可靠的。