Search EdWorkingPapers by author, title, or keywords.
Covid-19 Education Research for Recovery
With a goal of contextualizing teacher job dissatisfaction during the first full school year of the COVID-19 pandemic, we contrast teachers’ experiences to the decade and a half leading up to the pandemic. We draw on nationally representative data from the Schools and Staffing Survey and National Teacher and Principal Survey from the 2003-04 to 2020-21 school years. Through descriptive and regression analysis, we show that (1) teacher dissatisfaction has gradually been increasing over time, but did not decrease sharply in the 2020-21 school year, (2) levels of dissatisfaction during the pandemic were not equal across subpopulations of teachers or over time, and (3) positive working conditions consistently predicted lower job dissatisfaction, including in the 2020-21 school year.
The extent to which pandemic-induced public school enrollment declines will persist is unclear. Student-level data from Michigan through fall 2021 yields three relevant findings. First, relative to pre-pandemic trends, fall 2021 enrollment had partially recovered for low-income, Black, and Hispanic students, but had declined further for non-low-income, White, and Asian students. Second, annual public school exit rates remained elevated for elementary students and accelerated further for middle school students. Third, public school exit is sticky and varies by chosen alternative. Only 21 percent of those who left for private schools in fall 2020 had returned by fall 2021, while 50 percent of those who left for homeschooling had returned. These findings suggest that pandemic-driven public school enrollment declines may persist, and more so among higher income families.
Tutoring has emerged as an especially promising strategy for supporting students academically. This study synthesizes 33 articles on the implementation of tutoring, defined as one-to-one or small-group instruction in which a human tutor supports students grades K-12 in an academic subject, to better understand the facilitators and barriers to program success. We find that policies influenced tutoring implementation through the allocation of federal funding and stipulation of program design. Tutoring program launch has often been facilitated by strategic relationships between schools and external tutoring providers and strengthened by transparent assessments of program quality and effectiveness. Successful implementation hinged on the support of school leaders with the power to direct school funding, space, and time. Tutoring setting and schedule, recruitment and training, and curriculum influenced whether students are able to access quality tutoring and instruction. Ultimately, evidence suggests that tutoring was most meaningful when tutors fostered positive student-tutor relationships which they drew upon to target instruction toward students’ strengths and needs.
Four-day school weeks are becoming increasingly common in the United States, but their effect on students’ achievement is not well-understood. The small body of existing research suggests the four-day schedule has relatively small, negative average effects (~-0.02 to -0.09 SD) on annual, standardized state test scores in math and reading, but these studies include only a single state or are limited by using district-level data. We conduct the first multi-state, student-level analysis that estimates the effect of four-day school weeks on student achievement and a more proximal measure of within-year growth using NWEA MAP Growth assessment data. We conduct difference-in-differences analyses to estimate the effect of attending a four-day week school relative to attending a five-day week school. We estimate significant negative effects of the schedule on spring reading achievement (-0.07 SD) and fall-to-spring achievement gains in math and reading (-0.06 SD in both). The negative effects of the schedule are disproportionately larger in non-rural schools than rural schools and for female students, and they may grow over time. Policymakers and practitioners will need to weigh the policy’s demonstrated negative average effects on achievement in their decisions regarding how and if to implement a four-day week.
We analyze the impact of COVID-19 diagnoses on student grades, retention, and on-time graduation at a large public university. Even though COVID-19 rarely causes major health complications for a typical university student, diagnosis and quarantine may cause non-trivial disruptions to learning. Using event study analysis, we find that a COVID-19 diagnosis decreased a student's term grade point average (GPA) modestly by 0.08 points in the semester of diagnosis without significant effects afterward. The results were the most pronounced for male students, individuals with face-to-face instruction, and those with higher GPAs before the pandemic. We do not find a significant increase in the incidence of failing or withdrawing from a course due to diagnosis. In addition, we find no general evidence that the diagnoses delayed graduation or significantly altered first-year retention. However, the University experienced significant grade inflation during the pandemic, which exceeded the estimated effects of any COVID-19 diagnoses.
We study the distributional effects of remote learning. Our approach combines newly collected data on parental preferences with administrative data from Los Angeles. The preference data allow us to account for selection into remote learning while also studying selection patterns and treatment effect heterogeneity. We find a negative average effect of remote learning on reading (–0.14σ) and math (–0.17σ). Notably, we find evidence of positive learning effects for children whose parents have the strongest demand for remote learning. Our results suggest an important subset of students who currently sort into post-pandemic remote learning benefit from expanded choice.
We use linked individual-level data on school enrollment, physician services received, and prescription medications to measure the effect of the COVID-19 pandemic and associated disruptions on mental health treatment received by adolescents in British Columbia. We also investigate whether these effects are mediated by socioeconomic status and schooling mode. The results suggest substantial increases for non-Indigenous English home language girls in treatment for depression/anxiety, ADHD, eating disorders and other mental health conditions. Indigenous and non-English home language girls also show increases in treatment for depression/anxiety, and Indigenous girls show increases in treatment for ADHD. In contrast, boys show no change or even reductions in treatment for most mental health conditions. These effects vary somewhat by socioeconomic status, but we find no evidence that they vary substantially by schooling mode.
Hardware requirements are a barrier to widespread adoption of digital learning software among low-income populations. We investigate the demand among smallholder-farming households for a simple, adaptive math learning tool that can be accessed by widely available ``brick'' phones, and its effect on educational outcomes. Over a quarter of invited households used the tool, with greater demand among households lacking electricity, radios, or televisions. Usage was highest when schools were out of session. Engagement lapsed without regular reminders to use the service. Using random variation in access to the service, we find evidence that the platform increased test scores, school attendance, and grade attainment. Interpretation of these estimates is complicated by potentially endogenous outcome observation.
School closures induced by the COVID-19 pandemic led to concerns about student learning. This paper evaluates the effect of school closures on student learning in Uzbekistan, using a unique dataset that allows assessing change in learning over time. The findings show that test scores in math for grade 5 students improved over time by 0.29 standard deviation despite school closures. The outcomes among students who were assessed in 2019 improved by an average of 0.72 standard deviation over the next two years, slightly lower than the expected growth of 0.80 standard deviation. The paper explores the reasons for no learning loss.