Search for EdWorkingPapers here by author, title, or keywords.
Recent interest to promote and support replication efforts assume that there is well-established methodological guidance for designing and implementing these studies. However, no such consensus exists in the methodology literature. This article addresses these challenges by describing design-based approaches for planning systematic replication studies. Our general approach is derived from the Causal Replication Framework (CRF), which formalizes the assumptions under which replication success can be expected. The assumptions may be understood broadly as replication design requirements and individual study design requirements. Replication failure occurs when one or more CRF assumptions are violated. In design-based approaches to replication, CRF assumptions are systematically tested to evaluate the replicability of effects, as well as to identify sources of effect variation when replication failure is observed. In direct replication designs, replication failure is evidence of bias or incorrect reporting in individual study estimates, while in conceptual replication designs, replication failure occurs because of effect variation due to differences in treatments, outcomes, settings, and participant characteristics. The paper demonstrates how multiple research designs may be combined in systematic replication studies, as well as how diagnostic measures may be used to assess the extent to which CRF assumptions are met in field settings.
Researchers are rarely satisfied to learn only whether an intervention works, they also want to understand why and under what circumstances interventions produce their intended effects. These questions have led to increasing calls for implementation research to be included in high quality studies with strong causal claims. Of critical importance is determining whether an intervention can be delivered with adherence to a standardized protocol, and the extent to which an intervention protocol can be replicated across sessions, sites, and studies. When an intervention protocol is highly standardized and delivered through verbal interactions with participants, a set of natural language processing (NLP) techniques termed semantic similarity can be used to provide quantitative summary measures of how closely intervention sessions adhere to a standardized protocol, as well as how consistently the protocol is replicated across sessions. Given the intense methodological, budgetary and logistical challenges for conducting implementation research, semantic similarity approaches have the benefit of being low-cost, scalable, and context agnostic for use. In this paper, we demonstrate how semantic similarity approaches may be utilized in an experimental evaluation of a coaching protocol on teacher pedagogical skills in a simulated classroom environment. We discuss strengths and limitations of the approach, and the most appropriate contexts for applying this method.
This study assesses the effects of two text-messaging programs for parents that aim to support the development of math skills in prekindergarten students. One program focuses purely on math, while the other takes an identical approach but focuses on a combination of math, literacy, and social-emotional skills. We find no evidence that the math-only program benefits children’s development. However, the combination program shows greater promise, particularly for girls. Quantile regressions indicate that the effects are concentrated in the lower half of the outcome distribution. Results imply that girls may have started the year behind boys in math and caught up to and even surpassed boys when their parents have access to the program that combines topics. Our results also provide evidence that the structure of behavioral interventions can affect who benefits from the program, sometimes in unexpected ways, to produce meaningful differences in outcomes. We discuss and provide evidence for various hypotheses that could explain these differences.
We conduct a comprehensive examination of the causal effect of charter schools on school segregation, using a triple differences design that utilizes between-grade differences in charter expansion within school systems, and an instrumental variable approach that leverages charter school opening event variation. Charter schools increase school segregation for Black, Hispanic, White, and Asian students. The effect is of modest magnitude; segregation would fall 6 percent were charter schools eliminated from the average district. Analysis across varied geographies reveals countervailing forces. In metropolitan areas, charters improve integration between districts, especially in areas with intense school district fragmentation.
Most public colleges and universities rely heavily on state financial support. As state budgets have tightened in recent decades, appropriations for higher education have declined substantially. Despite concerns expressed by policymakers and scholars that the declines in state support have reduced the return to education investment for public sector students, little evidence exists that can identify the causal effect of these funds on long-run outcomes. We present the first such analysis in the literature using new data that leverages the merger of two rich datasets: consumer credit records from the New York Fed's Consumer Credit Panel (CCP), sourced from Equifax, and administrative college enrollment and attainment data from the National Student Clearinghouse. We overcome identification concerns related to the endogeneity of state appropriation variation using an instrument that interacts the baseline share of total revenue that comes from state appropriations at each public institution with yearly variation in state-level appropriations. Our analysis is conducted separately for two-year and four-year students, and we analyze individuals into their mid-30s. For four-year students, we find that state appropriation increases lead to substantially lower student debt originations. They also react to appropriation increases by shortening their time to degree, but we find little effect on other outcomes. In the two-year sector, state appropriation increases lead to more collegiate and post-collegiate educational attainment, more educational debt consistent with the increased educational attainment, but lower likelihood of delinquency and default. State support also leads to more car and home ownership with lower adverse debt outcomes, and these students experience substantial increases in their credit score and in the affluence of the neighborhood in which they live. Examining mechanisms, we find state appropriations are passed on to students in the form of lower tuition in the four-year sector with no institutional spending response. For community colleges, we find evidence of both price and quality mechanisms, the latter captured in higher educational resources in key spending categories. These results are consistent with the different pattern of effects we document in the four-year and two-year sectors. Our results underscore the importance of state support for higher education in driving student debt outcomes and the long-run returns to postsecondary investments that students experience.
We replicate and extend prior work on Florida’s Bright Futures merit aid scholarship to consider its effect on college enrollment and degree completion. We estimate causal impacts using a regression discontinuity design to exploit SAT thresholds that strongly determine eligibility. We find no positive impacts on attendance or attainment, and instrumental variable results generally reject estimates as small as 1-2 percentage points. Across subgroups, we do find that eligibility slightly reduces six-year associate degree attainment for lower-SES students and may induce small enrollment shifts among Hispanic and White students. Our findings of these minimal-at-best impacts contrast those of prior works, attributable in part to methodological improvements and more robust data, and further underscore the importance of study replication. (JEL: H75, I21, I22, I23, I28)
Parental text messaging interventions are growing in popularity to encourage at-home reading, school-attendance, and other educational behaviors. These interventions, which often combine multiple components, frequently demonstrate varying amounts of effectiveness, and researchers often cannot determine how individual components work alone or in combination with one another. Using a 2x2x3 factorial experiment, we investigate the effects of individual and interacted components from three behavioral levers to support summer reading: providing updated, personalized information; emphasizing different reading views; and goal setting. We find that the personalized information condition scored 0.03 SD higher on fall reading assessments. Test score effects were enhanced by messages that emphasized reading being useful for both entertainment and building skills compared to skill building alone or entertainment alone.
The COVID-19 pandemic presents a unique opportunity to examine how local governments respond to a public health crisis amid high levels of partisan polarization and an increasing tendency for local issues to become nationalized. As an arena that has, in recent years, been relatively separate from national partisan divides, public schools provide a useful window into these dynamics. Leveraging the fact that all of the nation’s school districts had to adopt a reopening plan for the fall, we test what factors best predict whether a district chose to return students to the classroom or educate them remotely. Contrary to the conventional understanding of school districts as localized and non-partisan actors, we find evidence that politics, far more than science, shaped school district decision-making. Mass partisanship and teacher union strength best explain how school boards approached reopening. Additionally, we find evidence that districts are sensitive to the threat of private school exit. Districts located in counties with a larger number of Catholic schools were less likely to shut down and more likely to return to in-person learning. These findings have important implications for our understanding of education policy and the functioning of American local governments.
In March 2020, most schools in the United States closed their doors and transitioned to distance learning in an effort to contain COVID-19. During the transition a significant number of students did not fully engage in these learning opportunities due to resource or other constraints. An urgent question for schools around the nation is how much did the pandemic impact student academic and social-emotional development. This paper uses administrative panel data from California to approximate the impact of the pandemic by analyzing how absenteeism affects student outcomes. We show wide variation in absenteeism impacts on academic and social-emotional outcomes by grade and subgroup, as well as the cumulative effect of different degrees of absence. Student outcomes generally suffer more from absenteeism in mathematics than in ELA. Negative effects are larger in middle school. Absences negatively affect social-emotional development, particularly in middle school, with slight differences across constructs. Our results add to the emerging literature on the impact of COVID-19 and highlight the need for student academic and social-emotional support to make up for lost time.
Jackson, Wigger, and Xiong (2020a, JWX) provide evidence that education spending reductions following the Great Recession had widespread negative impacts on student achievement and attainment. This paper describes our process of duplicating JWX and highlights a variety of tests we employ to investigate the nature and robustness of the relationship between school spending reductions and student outcomes. Though per-pupil expenditures undoubtedly shifted downward due to the Great Recession, contrary to JWX, our findings indicate there is not a clear and compelling story about the impact of those reductions on student achievement. Moreover, we find that the relationship between K-12 spending and college-going rates is likely confounded with contemporaneous higher education funding trends. While we believe that K-12 spending reductions may have negative impacts on student outcomes, our results suggest that estimating generalizable causal effects remains a significant challenge.