EdWorkingPapers
The Returns to Private Tutoring Centers
Private tutoring centers have expanded rapidly across the United States, yet there is little evidence on whether these centers improve student outcomes. Using business records on tutoring establishments linked to administrative data covering over five million public school students in Texas, this paper estimates the short- and long-run impacts of tutoring centers. Exploiting variation from the… more →
How Much Do Principals Matter When Politicians Choose Them? Evidence from Dated Successions and Close Elections in Brazil
Estimates of how much school principals matter differ by an order of magnitude, in part because conventional estimators require principal identifiers and absorb sampling noise into the estimated dispersion. We identify the variance of principal-by-school effects without identifiers, using successions dated by principals' reported tenure in Brazil's 2011 national assessment, which imply known… more →
What Districts Aim to Do: A Computational Mixed Methods Analysis of School District Strategic Plans
School districts are a distinct but long-critiqued feature of the United States educational system. Amid these critiques and increased public dissatisfaction with education, we revisit a fundamental question: What do school districts aim to do? Using a computational mixed methods analysis of 624 district strategic plans, we find that districts a) commit to a broad range of goals beyond student… more →
Legislating Literacy: Estimating the Impact of State-Level Science of Reading Reforms on Reading Test Scores
This study examines the extent to which state adoption of Science of Reading (SoR)-aligned policies was associated with changes in fourth-grade reading achievement. Using longitudinal state-level panel data from 2003-2024 and quasi-experimental difference-in-differences (DiD) methods, the study estimates relationships between SoR policy adoption and NAEP Grade 4 reading outcomes. Findings suggest… more →
Screen or Barrier? Science of Reading Certification Exams and Teacher Supply in Texas
Twenty states now require prospective teachers to pass a standalone Science of Reading (SOR) exam. These exams restrict teacher supply as a barrier, and whether they are worth that cost depends on whether they also screen on teaching quality. This paper explores this tension in Texas' SOR exam, required since 2021 for new elementary and middle-grades candidates, using administrative data linking… more →
Integrated vs. Separated Schooling for Refugee Children: Long-Term Effects on Education and Assimilation
Does direct integration into native schools or temporary placement in separate schools better promote the education and cultural assimilation of refugee children? In the Netherlands, primary-school-aged refugee children are placed in either integrated or separated schools depending on their randomly assigned municipality. By contrast, all secondary-school-aged arrivals are initially separated. We… more →
The Time Costs of Academic Credit Hours
Do students give up free time with added classes? We estimate the effect of additional credits on leisure using campus departure logs linked to academic records of U.S. Air Force Academy students and two natural experiments. Leveraging pre-assigned schedules, each credit reduces leisure by 7 hours per semester, while unexpected changes to graduation requirements crowded out 17 hours. Foregone… more →
Designing Effective Public-Private Partnerships in Education
Publicly funded school choice is examined as a form of education public-private partnership (PPP), with cross-national differences interpreted through institutional design rather than private provision alone. A novel theoretical framework derives two indicators – alignment and coherence – from a formal utility-theoretic model of PPP maturity, applied here for the first time to education. Across… more →
One Fell Through the Safety Net: Embeddedness and Take Up of Social Benefits
The U.S. social safety net comprises a patchwork quilt of welfare programs, but not all eligible beneficiaries receive aid. Incomplete take-up suggests that some children and families are not receiving access to the education, healthcare, housing, and income needed to break the cycle of poverty and may reflect unequal burdens in accessing benefits. We examine incomplete take-up using the case of… more →
Teacher-Mediated Multimodal Platform and Math Achievement: Evidence from Goblins' Yearlong District Implementation
Despite the rapid adoption of AI-powered learning tools in K–12 classrooms, there is limited evidence about their effectiveness in improving student learning. This paper examines the usage of Goblins—a teacher-mediated multimodal AI math platform that allows students to type, draw, and speak while receiving personalized feedback—and how that usage associates with middle school math achievement.… more →
Career ready or not? The relationship of WorkKeys scores, ACT scores, and high school GPAs with college and early-career outcomes
Growing efforts to ensure that students graduate high school both college- and career-ready raise a need for better understanding and measures of “career readiness.” We leverage data from one state (North Carolina) where nearly half of students take a college entrance exam (the ACT) in 11th grade and a career readiness exam (the WorkKeys) in 12th grade to examine how these different measures of… more →
Talent at the Top: Effects of Taxation on University Presidents’ Compensation and Priorities
College presidents are among the highest-paid individuals in higher education, and in 2017 the Tax Cuts and Jobs Act increased their cost to their institutions via a 21% excise tax on nonprofit executive compensation above $1 million. I examine how private colleges and universities and their presidents responded using an event study design. Presidential compensation fell by 13%, with the entire… more →
Improvements in Using Distance to Predict College Enrollment
Geographic proximity to college is a widely used instrument for estimating returns to postsecondary education. Most applications rely on geodesic ("crow-flies") distance, which is easy to calculate but a poor proxy for actual travel costs when terrain, water, or congestion intervene. We show that replacing geodesic distance with driving distance or driving time substantially strengthens… more →
The Impact of Exam Schools on Disadvantaged Students: Evidence from NYC’s Discovery Program
This paper studies the New York City Department of Education's Discovery Program, an affirmative action policy implemented to increase access to some of New York City’s most elite public high schools. The Discovery Program allocates reserve seats at New York City’s exam schools to disadvantaged students who successfully complete the course requirements of a summer program. Using a regression… more →
Beyond the Special Education Gap: Differences in Special Education Expenditures Between Charter Schools and Traditional Public Schools
Researchers and policymakers have long noted differences in disability enrollment and placement between charter schools and traditional public schools (TPSs), yet little is known about whether these differences correspond to special education expenditures. Using Texas administrative and financial data from 2019-20 to 2024-25, we compare special education expenditures, disability composition, and… more →
Defaults, Discretion, and Local Capacity: District Leaders’ Accounts of Automatic English Learner Reclassification
Automatic policies, which produce a preferred outcome unless implementers act otherwise, are rare in K-12 education and have been studied almost entirely using quantitative methods. In 2019, Michigan replaced its district-led procedure for reclassifying students out of English learner (EL) status with an automatic, state-led one. Drawing on interviews with 17 district EL leaders across 14… more →
Early Evidence on Employer Responses to College Student Use of Generative AI
We conduct a survey experiment on professionals with hiring experience to investigate how employers respond to students' use of generative AI in higher education. Prior to respondents reviewing two resumes, they were randomized into a control group or one of two treatment groups. One treatment involved framing AI use as assisting students while the other involved framing AI use as substituting… more →
Human-, GRM-, and LLM-Generated School-Improvement Leadership Scores from Missouri: Structural Bias and Concurrent and Predictive Validity
Head or assistant principal evaluation systems are widely deployed but rarely examined for validity, and AI scoring tools have entered K–12 workflows ahead of the evidence. We examine human supervisor ratings, Graded Response Model (GRM) rescaling, and large language model (LLM) scoring of 1,385 yearly leader-authored, standards-based, retrospective accounts of school improvement efforts and… more →
Why Does Education Policy so Frequently Disappoint? Lessons from Teacher Evaluation
Decades of K–12 education reform — much of it focused on teaching — have produced a persistent paradox: policies grounded in strong evidence disappoint when scaled, even as some implementations achieve striking success. Teacher evaluation is the clearest case. Cincinnati, Dallas, Chicago, and Washington, DC produced meaningful achievement gains, while forty-four state systems, Houston, and a… more →
A National Measure of School District Partisanship for the 2016 to 2024 Presidential Elections
Scholars increasingly interested in the governance of local school districts, which invites curiosity about the partisanship of their voting constituencies. Presently, researchers have to translate precinct-level or county-level votes to school district geographies using imprecise imputation strategies as the geographies are not coterminous. We review the recent methods researchers use to… more →
Staffing, Retention, and Expenditures in the Four-Day School Week
School districts across the United States are increasingly adopting the four-day school week, often in the hope of addressing teacher staffing challenges. We examine the effects of the four-day week on non-teacher staffing and recruitment, and schoolwide expenditures in Missouri, where three in ten districts currently use it. Using a difference-in-differences design to identify the causal impact… more →
The Effects of Course-Integrated AI Tutoring on Student Performance and Engagement: A Randomized University Trial
We evaluate access to a course-integrated generative AI (GenAI) tutor in a randomized controlled trial involving 2,379 undergraduate students and 30 instructors across multiple disciplines in a large U.S. public university. Among sections of the same course, tutor access reduced final grades by 0.37 standard deviations and learning management system participation by 0.90 standard deviations.… more →
High School College-Conduciveness and Its Association with the Top Ten Percent Students’ Academic Profiles
High school college-conducive resources shape students’ college readiness, which may explain the relative disadvantage racialized minority and low-income students face in accessing selective universities. Texas Top Ten Percent Plan (TTP) aims to bridge this gap by conditioning admission on class rank, which accounts for between-school differences. However, under policy complementarity theory,… more →
Tutor Traits and Student Achievement: Evidence from a Randomized Experiment
Tutors are increasingly central to school-based instruction, with the potential to personalize learning at scale. Little is known about which tutor characteristics predict effectiveness, limiting providers' and schools' recruiting, hiring, and retention strategies. Using random assignment of tutors in a high-dosage elementary-grade online literacy tutoring program, we examine how literacy… more →
The Impact of International Students on US Colleges: Higher Education as a Service Export
Between 2003 and 2016, international enrollment in US higher education nearly doubled. We examine how trade shocks in education affect public universities’ decision-making. We construct a shift-share instrument exploiting institutions’ historical networks with different origins of international students, income growth, and exchange-rate fluctuations. Contrary to claims that US-born students are… more →
Estimating the Expenditures Associated with Special Education: Evidence from Michigan
Using a 13-year panel of school districts in Michigan, this paper estimates the average expenditures associated with different disability category labels. Increasing the number of students with disabilities by 1 student in the average school district is associated with an increase in expenditures between $1,998 for students with specific learning disabilities to more than a $23,000 increase in… more →
Non-traditional Student Advising Before College: Lessons from Navigate Reconnect
Prospective college students benefit from tailored advising and guidance about whether, where, and how to enroll. Support for college planning drops sharply after students leave high school, however, leaving non-traditional students largely on their own. We study Navigate Reconnect, a Tennessee advising program that intends to fill this gap. We evaluate how Navigate Reconnect access and… more →
Do Civic Infrastructures Predict Student Outcomes? Comparing Grassroots and Professional Education Nonprofits
Local civic infrastructures—the networks of organizations, relations, and community resources supporting schools—can shape education policy and practice, yet evidence of their relationship to student outcomes remains limited. We examine this relationship using a novel panel dataset of US education nonprofits from 2010 to 2019, geocoded and linked to over 10,000 school districts. Grassroots… more →
Can Early Exposure to the Teaching Profession Increase Teacher Supply and Diversity? New Evidence on the Teacher Pipeline
We examine whether early exposure to the teaching career has the potential to expand the size and diversity of new teacher supply. Exploiting the staggered expansion of high school education courses across Texas, we find little evidence that these courses increased new teacher supply or diversity despite course takers being substantially more diverse than the teacher workforce. High school… more →
Heterogeneous Fadeout of Cognitive Gains: Evidence from Head Start Preschool
Fadeout is defined as the decline in an intervention's initial treatment effect in subsequent follow-up periods. This paper uses the Head Start Impact Study to examine heterogeneity in the fadeout of cognitive gains from access to Head Start preschool. I find that Black children experience substantially less fadeout than non-Black children and retain a 0.23 standard deviation gain at the end of… more →
Why aren’t online learning programs working better in schools: Maybe they need to be more integrated with what is actually taught during whole classroom instruction?
Computer-assisted learning (CAL)—educational software providing instruction to help students learn concepts and skills—is ubiquitous in US schools. Despite meta-analyses suggesting benefits, national test scores have declined during rapid CAL adoption. This may reflect insufficient research on how and when CAL complements instruction. Using a fixed effect model, we find 60 additional minutes of… more →
Immigration Enforcement Actions and Empty Desks: Persistent and Acute Attendance Effects
How do changes in immigration policy affect student attendance? We use student-by-day administrative records from a mid-sized school district to estimate the causal effect of heightened federal immigration enforcement following the January 2025 presidential inauguration on student attendance using a difference-in-differences design. We find that heightened enforcement and immigration enforcement… more →
Assistant Principals’ Career Aspirations: Exploring the Supply Side of the School Leadership Pipeline
Most assistant principals (APs) seek to become principals in the future. However, little is known about why some APs aspire to the principalship while others wish to remain in the AP role. This study examines the factors that predict APs’ aspirations to the principalship, which is key to understanding the supply of future principals. We use data from the 2024 Tennessee Educator Survey to examine… more →
Impact of States' Adoption of Response to Intervention (RTI) on Disability Identification
The rise of specific learning disability (SLD) identification from 1975 to the early 2000s prompted the 2004 reauthorization of the Individuals with Disabilities Education Act to permit states to use a student's response to intervention (RTI) in eligibility determination to improve identification. Single-state studies report that RTI adoption reduced SLD identification by 11% to 15%. We test… more →
Incidence and Effects of Reported University Student Absences
Despite the potential for illness or bereavement to significantly disrupt education, there is little evidence about the frequency and impact of absences in higher education. Using administrative data from a large public university in the United States, we analyze the incidence of reported student absences and their effects on grades and retention. We document significant heterogeneity in the use… more →
Persisting Income-Based Preschool Enrollment Gap in the U.S., 2000 to 2023
This study aims to understand whether access to early childhood education (ECE) has become more equitable over a period of substantial growth in public pre-K investment. This study describes and statistically tests the differential trends across income groups in center-based ECE enrollment among 3- and 4-year-olds. Across descriptive, regression, and decomposition analyses, the income-based… more →
Revisiting Why Superintendents Turn Over: Understanding Pathways Out of School District Leadership
Research has examined many facets of superintendent turnover but offers an incomplete understanding of superintendents’ pathways out of the role. We pair administrative data on Tennessee superintendents over 15 years with hand-collected information about the reported reasons for their departures and jobs they took after leaving. Our data show that interpreting exits from state administrative data… more →
Off the Books: School-Based Informal Diversion and Student and Community Outcomes
Efforts to reduce school-based arrest may reshape policing beyond school doors, yet their downstream effects are largely unknown. We study a New York City policy that encouraged informal diversion in place of arrest, using a difference-in-differences strategy. The probability of a juvenile crime record near a school fell by 6.9 percentage points---concentrated among misdemeanor offenses and… more →
Superintendent (In)Stability and Student Achievement in the United States
Superintendents shape conditions for teaching and learning, yet the limited studies of their impact—concentrated in a few states and focused on superintendent turnover—find small or null effects on student achievement. Using more than 37,000 district-year records from national superintendents and student achievement datasets, and drawing on organizational learning and managerial succession… more →
(How) Do We Teach Emotions?
Emotional intelligence constitutes a key component of human capital, shaped partially by educational materials. Using machine learning and generative AI tools, we examine emotional content in public-school textbooks and children’s literature. A stark mismatch emerges: text exposes children to a broad emotional range, but images overwhelmingly depict happiness and calm, regardless of emotions… more →
"How You Spend Your Money is Your Strategy": Education Spending and the Academic Achievement of Students with and without Disabilities in Virginia, 2018-2025
Virginia experienced a boost in federal and state education funding from 2020-2025 and launched the Virginia Literacy Act in 2023. These policy events prompted our study of the impact of education spending by school divisions on growth in academic pass rates in reading and mathematics for students with and without disabilities in grades three through eight (2018-2025). To do this, we estimated… more →
Is Calculus a Gatekeeper in Selective Engineering Admissions? Insights from a Lab-in-the-Field Experiment with Admissions Officers
Lower-resourced high schools are far less likely to offer calculus, raising concerns that calculus availability may be a gatekeeper in selective college admissions. Using a lab-in-the-field experiment (292 admissions officers, 1768 total observations), we investigate whether engineering applicants are penalized for not taking calculus—even when it was unavailable to them. We also investigate… more →
Shine a Light: The U.S. Federal Government Should Increase Access to Educational Administrative Data
The Institute of Educational Sciences invested substantial resources into the Statewide Longitudinal Data Systems grant program. These grants helped states convert their educational records to longitudinal datasets for research and evaluation. However, external researchers still have unreliable access to these datasets. In this commentary, we argue that the U.S. federal government should use… more →
Post-pandemic achievement recovery based on student characteristics and district-level policies
It is well-established that student achievement has declined dramatically since the onset of the COVID-19 pandemic. However, less is known how these patterns may vary by developmental timing of exposure to the pandemic and how district-level policies may have influenced student recovery. This descriptive study examined patterns of achievement across middle childhood during and after the COVID-19… more →
State School Accountability Decisions and the Implications for School Ratings
State school accountability systems have been a constant in educational policy since before federal law required them. Currently, state systems must include multiple measures of school quality but states otherwise have considerable leeway in measuring schools. We simulate school ratings using eight years of data to demonstrate how state decisions about (a) weighting multiple measures, and (b) how… more →
Evaluation Scores and Student Learning under Compulsory Teacher Re-Evaluation
We examine how lower-performing public school teachers respond to more frequent formative evaluations. A 2011 reform in Chile required teachers rated "basic" (the second-lowest category) to be re-evaluated after two years rather than four. Using nationwide administrative data on primary school teachers and students from 2005-2015, we show that the reform increased two-year re-evaluation rates by… more →
Measures Still Matter: Replicating and Extending a Meta-Analysis of the Effects of Educational Apps on Children's Reading and Math Skills with Six Years of New Data
This study replicates and extends a meta-analysis of 36 studies of educational apps (Kim et al., 2021) with six years of new data. We synthesize 841 effect sizes from 141 studies evaluating the effectiveness of educational apps on reading and math skills of students ages 3-9. Replicating the original study, we find a moderate positive pooled effect size with substantial heterogeneity (d = .40, 95… more →
The Access–Enrollment Disconnect in Corequisite Math Reform: Causal Evidence and Stakeholder Perspectives from a Community College
This study examines a statewide developmental education reform in Virginia that allows students who would have been placed in standalone remedial math to enroll directly in college-level math with corequisite support. Although corequisite support has been widely adopted and generally promising, less is known about how these reforms operate when expanded access does not always translate into… more →
How Colleges and Universities are Adopting AI in Higher Education: A Global Comparative Analysis
Artificial intelligence (AI) and generative AI have become central strategic priorities for universities worldwide, yet the extent to which institutions formally embed AI into their governance, academic, and policy structures remains unclear. This study examines the institutionalization of AI across a global sample of research universities drawn from the Times Higher Education 2026 World… more →
Leveraging complementarities between teachers' content knowledge and pedagogical supports: Experimental evidence from India
We evaluate whether pairing strong subject-matter knowledge with pedagogical supports can improve learning in public primary schools. In Pune, India, we randomly assigned grades within 48 schools to an intervention that recruited college students in science, technology, engineering, and math (“fellows”), supported them with training, coaching, and scripts, and had them partly replace math and… more →