Search EdWorkingPapers by author, title, or keywords.
Virtual charter schools are increasingly popular, yet there is no research on the long-term outcomes of virtual charter students. We link statewide education records from Oregon with earnings information from IRS records housed at the US Census Bureau to provide evidence on how virtual charter students fare as young adults. Virtual charter students have substantially worse high school graduation rates, college enrollment rates, bachelor's degree attainment, employment rates, and earnings than students in traditional public schools. Although there is growing demand for virtual charter schools, our results suggest that students who enroll in virtual charters may face negative long-term consequences.
The media discourse on student loans plays a significant role in the way that policy actors conceptualize challenges and potential solutions related to student debt. This study examines the racialized language in student loan news articles published in eight major news outlets between 2006 and 2021. We found that 18% of articles use any racialized language, though use has accelerated since 2018. This increase appears to be driven by terms that denote groups of people instead of structural problems, with 8% of articles mentioning “Black” but less than 1% mentioning “racism.” These findings emphasize the importance of treating the media as a policy actor capable of shaping the salience of racialization in discussions about student loans.
Efforts to attract and retain effective educators in high poverty public schools have had limited success. Dallas ISD addressed this challenge by using information produced by its evaluation and compensation reforms as the basis for effectiveness-adjusted payments that provided large compensating differentials to attract and retain effective teachers in its lowest achievement schools. The Accelerating Campus Excellence (ACE) program offers salary supplements to educators with records of high performance who are willing to work in the most educationally disadvantaged schools. We document that ACE resulted in immediate and sustained increases in student achievement, providing strong evidence that the multi-measure evaluation system identifies effective educators who foster the development of cognitive skills. The improvements at ACE schools were dramatic, bringing average achievement in the previously lowest performing schools close to the district average. When ACE stipends are largely eliminated, a substantial fraction of highly effective teachers leaves, and test scores fall. This highlights the central importance of the performance-based incentives to attract and retain effective educators in previously low-achievement schools.
We evaluate the effects of the 2020 student debt moratorium that paused payments for student loan borrowers. Using administrative credit panel data, we show that the payment pause led to a sharp drop in student loan payments and delinquencies for borrowers subject to the debt moratorium, as well as an increase in credit scores. We find a large stimulus effect, as borrowers substitute increased private debt for paused public debt. Comparing borrowers whose loans were frozen with borrowers whose loans were not frozen due to differences in whether the government owned the loans, we show that borrowers used the new liquidity to increase borrowing on credit cards, mortgages, and auto loans rather than avoid delinquencies. The effects are concentrated among borrowers without prior delinquencies, who saw no change in credit scores, and we see little effects following student loan forgiveness announcements. The results highlight an important complementarity between liquidity and credit, as liquidity increases the demand for credit even as the supply of credit is fixed.
Although learners are being connected 1:1 with instructors at an increasing scale, most of these instructors do not receive effective, consistent feedback to help them improved. We deployed M-Powering Teachers, an automated tool based on natural language processing to give instructors feedback on dialogic instructional practices —including their uptake of student contributions, talk time and questioning practices — in a 1:1 online learning context. We conducted a randomized controlled trial on Polygence, a re-search mentorship platform for high schoolers (n=414 mentors) to evaluate the effectiveness of the feedback tool. We find that the intervention improved mentors’ uptake of student contributions by 10%, reduced their talk time by 5% and improves student’s experi-ence with the program as well as their relative optimism about their academic future. These results corroborate existing evidence that scalable and low-cost automated feedback can improve instruction and learning in online educational contexts.
We investigated the effectiveness of a sustained and spiraled content literacy intervention that emphasizes building domain and topic knowledge schemas and vocabulary for elementary-grade students. The Model of Reading Engagement (MORE) intervention underscores thematic lessons that provide an intellectual framework for helping students connect new learning to a general schema in Grade 1 (animal survival), Grade 2 (how scientists study past events), and Grade 3 (our human body, a living system that helps us survive). A total of 30 elementary schools (N = 2,870 students) were randomized to a treatment or control condition. In the treatment condition (i.e., full spiral curriculum schools), students participated in content literacy lessons from Grades 1 to 3 during the school year and wide reading of thematically related informational texts in the summer following Grades 1 and 2. In the control condition (i.e., partial spiral curriculum schools), students participated in Grade 3 MORE lessons. Grade 3 lessons for both conditions were implemented online during the COVID-19 pandemic school year. Results reveal that treatment group students outperformed control students on science vocabulary knowledge across all three grades. Furthermore, we found positive transfer effects on Grade 3 science reading (ES = .14), domain-general reading comprehension (ES = .11), and mathematics achievement (ES = .12). Treatment impacts were sustained at 14-month follow-up on Grade 4 reading comprehension (ES = .12) and mathematics achievement (ES = .16). Findings indicate that a content literacy intervention that spirals topics and vocabulary across grades can improve students’ long-term academic achievement outcomes.
Expected earnings matter for college major choices, and majors differ in both their average earnings and the age profile of their earnings. We show that students' family background is strongly related to the earnings paths of the major they choose. Students with more educated parents, especially those who have graduate degrees, choose majors with lower early-career earnings but much faster earnings growth. They are also less likely to choose safe majors with little early-career earnings or unemployment downside. Parental income has a weaker relationship with major choice and operates mostly through the type of institution the student attends.
A fundamental question for education policy is whether outcomes-based accountability including comprehensive educator evaluations and a closer relationship between effectiveness and compensation improves the quality of instruction and raises achievement. We use synthetic control methods to study the comprehensive teacher and principal evaluation and compensation systems introduced in the Dallas Independent School District (Dallas ISD) in 2013 for principals and 2015 for teachers. Under this far-reaching reform, educator evaluations that are used to support teacher growth and determine salary depend on a combination of supervisor evaluations, student achievement, and student or family survey responses. The reform replaced salary scales based on experience and educational attainment with those based on evaluation scores, a radical departure from decades of rigid salary schedules. The synthetic control estimates reveal positive and significant effects of the reforms on math and reading achievement that increase over time. From 2015 through 2019, the average achievement for the synthetic control district fluctuates narrowly between -0.27 s.d. and -0.3 s.d., while the Dallas ISD average increases steadily from -0.28 s.d. in 2015 to -0.08 s.d. in 2019, the final year of the sample. Though the increase for reading is roughly half as large, it is also highly significant.
Temporary college closures in response to the COVID-19 pandemic created an exodus of students from college towns just as the decennial census count was getting underway. We use aggregate cellular mobility data to evaluate if this population movement affected the distributional accuracy of the 2020 Census. Based on the outflow of devices in late March 2020, we estimate that counties with a college were undercounted by two percent, likely affecting Congressional apportionment. For college towns, student populations can impact government funding allocations, policy program decisions, and planning for infrastructure, public health, and more. The Census Bureau is allowing governmental entities to request count reviews through June 2023. Colleges should cooperate with state and local government efforts to ensure an accurate count.
When analyzing treatment effects on test score data, education researchers face many choices for scoring tests and modeling results. This study examines the impact of those choices through Monte Carlo simulation and an empirical application. Results show that estimates from multiple analytic methods applied to the same data will vary because, as predicted by Classical Test Theory, two-step models using sum or IRT-based scores provide downwardly biased standardized treatment effect coefficients compared to latent variable models. This bias dominates any other differences between models or features of the data generating process, such as the variability of item discrimination parameters. An errors-in-variables (EIV) correction successfully removes the bias from two-step models. Model performance is not substantially different in terms of precision, standard error calibration, false positive rates, or statistical power. An empirical application to data from a randomized controlled trial of a second-grade literacy intervention demonstrates the sensitivity of the results to model selection and tradeoffs between model selection and interpretation. This study shows that the psychometric principles most consequential in causal inference are related to attenuation bias rather than optimal scoring weights.