Search for EdWorkingPapers here by author, title, or keywords.
Program and policy effects
Tutoring—defined here as one-on-one or small-group instructional programming by teachers, paraprofessionals, volunteers, or parents—is one of the most versatile and potentially transformative educational tools in use today. Within the past decade, dozens of preK-12 tutoring experiments have been conducted, varying widely in their approach, context, and cost. Our study represents the first systematic review and meta-analysis of these and earlier studies. We develop a framework for considering different types of programs to not only examine overall effects, but also explore how these effects vary by program characteristics and intervention context. We find that tutoring programs yield consistent and substantial positive impacts on learning outcomes, with an overall pooled effect size estimate of 0.37 SD. Effects are stronger, on average, for teacher and paraprofessional tutoring programs than for nonprofessional and parent tutoring. Effects also tend to be strongest among the earlier grades. While overall effects for reading and math interventions are similar, reading tutoring tends to yield higher effect sizes in earlier grades, while math tutoring tends to yield higher effect sizes in later grades. Tutoring programs conducted during school tend to have larger impacts than those conducted after school.
Younger siblings take more advanced high school course end of year exams when their older siblings perform better in those same exams. Using a regression discontinuity and data from millions of siblings who take Advanced Placement (AP) exams, we show that younger siblings with older siblings who marginally “pass” an AP exam are more likely to take at least one AP exam, increase the total number of AP exams, and are more likely to take the same exam as their sibling. The largest impacts are found among sisters, but we do not see differential effects in coursework where females are underrepresented.
COVID has led colleges to brace for potential enrollment declines in the Fall, which would devastate budgets and potentially decrease the likelihood a student ever earns a degree. We take an early look at California’s FAFSA applications up through mid-June, to anticipate how students may be responding to this crisis. We find that COVID did not affect most of California’s “traditional” high school graduates due to an early deadline for financial aid, which exists in a number of states. From early March to mid-June, FAFSA applications among freshmen declined 18%, relative to prior years. Although there were initial declines in applications among more experienced students, these quickly rebounded and are now 9% higher relative to prior years. The largest FAFSA increases occurred in counties that saw the most dramatic increases in Unemployment Insurance claims.
Since their introduction in the 1990s, charter schools have grown from a small-scale experiment to a ubiquitous feature of the public education landscape. The current study uses the legislative removal of a cap on the maximum number of charters, and the weakening of regulations on these new schools, in North Carolina as a natural experiment to assess the intensive impacts of charter school growth on teacher quality and student composition in traditional public schools (TPS) at different levels of local market penetration. Using an instrumental variable difference-in-differences approach to account for endogenous charter demand, we find that intensive local charter entry reduces the inflow of new teachers at nearby TPS, leading to a more experienced and credentialed teaching workforce on average. However, we find that the entry of charters serving predominantly White students leads to reductions in average teacher experience, effectiveness, and credentials at nearby TPS. Overall these findings suggest that the composition of the teacher workforce in TPS will continue to change as charter schools further expand, and that the spillover effects of future charter expansion will vary by the types of students served by charters.
Much of the literature estimating disproportionality in special education identification rates has focused on socioeconomic status, race, and gender. However, recent evidence suggests that a student’s school starting age also impacts the likelihood they receive special education services, particularly in the early grades. I build on the evidence that the youngest students in a grade more likely to be diagnosed with Attention Deficit/Hyperactivity Disorder and more likely to be placed in special education by estimating the effect of school starting age on special education identification in Michigan. I also estimate heterogeneity in this effect by student characteristics and across school districts. Using a regression discontinuity design exploiting variation in kindergarten starting age generated by a statewide kindergarten entrance age policy, I find that the youngest students in a kindergarten cohort are 40% more likely (3.3 percentage points, p<0.001) to be placed in special education than are the oldest students, and that this effect persists through eighth grade. Despite little evidence of heterogeneity by gender, race, or socioeconomic status, I find some suggestive evidence that the effect is particularly large for white boys in the early elementary grades and for black girls in the later elementary grades. I find no evidence that these effects vary across school districts. Finally, I find exploratory evidence of variation by school cohort age composition, suggesting these effects are driven moreso by relative age comparisons than absolute age developmental differences. Given the importance of special education services to the academic success of children with disabilities, these findings have implications for schools and for policymakers seeking to improve special education program provision.
Major philanthropic initiatives that incorporate features of venture-capital practices have become increasingly prominent, particularly in K-12 public education. In this study, we provide empirical evidence on the reach, character, and impact of the Broad Superintendents Academy, a prominent and controversial venture-philanthropic initiative designed to transform leadership in the nation’s largest school districts. Using a novel dataset on all Broad trainees and a linked panel data set of all large school districts over 20 years, we find that Broad superintendents have had extensive reach (e.g., serving nearly 3 million students at their peak). We also show that, within districts that hired Broad trainees, Broad superintendents were 40 percent more likely to be Black than their non-Broad peers, but also had tenures that were 18 percent shorter. Panel-based estimates provide evidence that Broad-trained leaders had no clear effects on several district outcomes such as enrollment, school closures, per-pupil instructional and support-service spending, and student completion rates. However, Broad-trained leaders initiate a trend towards an increased number of charter schools and higher charterschool enrollment.
Patience and risk-taking – two cultural traits that steer intertemporal decision-making – are fundamental to human capital investment decisions. To understand how they contribute to international differences in student achievement, we combine PISA tests with the Global Preference Survey. We find that opposing effects of patience (positive) and risk-taking (negative) together account for two-thirds of the cross-country variation in student achievement. In an identification strategy addressing unobserved residence-country features, we find similar results when assigning migrant students their country-of-origin cultural traits in models with residence-country fixed effects. Associations of culture with family and school inputs suggest that both may act as channels.
Despite calls for more evaluative research in teacher education, formal assessments of the effectiveness of novel teacher education practices remain rare. One reason is that we lack designs and measurement approaches that appropriately meet the challenges of causal inference in the field. In this article, we seek to fill this gap. We first outline the difficulties of doing evaluative work in teacher education. We then describe a set of replicable practices for developing measures of key teaching outcomes, and propose evaluative research designs that can be adapted to suit the needs of the field. Finally, we identify community-wide initiatives that are necessary to advance useful evaluative research.
The COVID-19 pandemic has put virtual schooling at the forefront of policy concerns, as millions of children worldwide shift to virtual schooling with hopes of “slowing the spread”. Given the emergency shift to online education coupled with the large increase in demand for virtual education over the last decade it is imperative to explore the impacts of virtual education on student outcomes. This paper estimates the causal effect of full-time virtual school attendance on student outcomes with important implications for school choice, online education, and education policy. Despite the increasing demand for K-12 virtual schools over the past decade little is known about the impact of full-time virtual schools on students’ cognitive and behavioral outcomes. The existing evidence on the impact of online education on students’ outcomes is mixed. I use a longitudinal data set composed of individual-level information on all public-school students and teachers throughout Georgia from 2007 to 2016 to investigate how attending virtual schools influences student outcomes. I implement a variety of econometric specifications to account for the issue of potential self-selection into full-time virtual schools. I find that attending a virtual school leads to a reduction of 0.1 to 0.4 standard deviations in English Language Arts, Mathematics, Science, and Social Studies achievement test scores for students in elementary and middle school. I also find that ever attending a virtual school is associated with a 10-percentage point reduction in the probability of ever graduating from high school. This is early evidence that full-time virtual schools as a type of school choice could be harmful to students’ learning and future economic opportunities, as well as a sub-optimal use of taxpayer money.
Enrollment in higher education has risen dramatically in Latin America, especially in Chile. Yet graduation and persistence rates remain low. One way to improve graduation and persistence is to use data and analytics to identify students at risk of dropout, target interventions, and evaluate interventions’ effectiveness at improving student success. We illustrate the potential of this approach using data from eight Chilean universities. Results show that data available at matriculation are only weakly predictive of persistence, while prediction improves dramatically once data on university grades become available. Some predictors of persistence are under policy control. Financial aid predicts higher persistence, and being denied a first-choice major predicts lower persistence. Student success programs are ineffective at some universities; they are more effective at others, but when effective they often fail to target the highest risk students. Universities should use data regularly and systematically to identify high-risk students, target them with interventions, and evaluate those interventions’ effectiveness.