A scoping review of 54 studies from 22 countries, synthesizing proposed AI competencies for undergraduate medical education into a taxonomy of seven domains, 37 competencies, and 170 learning objectives.

Tests which mechanisms drive the success of text-message parenting programs, comparing general information, actionable advice, and encouragement for preschoolers' literacy development.

Examines how Kansas block grant funding cuts affected student achievement, asking whether rural districts are more vulnerable to reductions in education funding.

Presents five guidelines for interpreting effect sizes of education interventions, arguing that Cohen's benchmarks misjudge the impacts of field-based programs.

Uses the first large-scale panel survey of social-emotional learning to trace how four constructs develop from grade 4 to 12 across student groups.

Contradictory evidence of the relationship between education funding and student achievement could reflect heterogeneous effects by revenue source or student characteristics. This study examines potential heterogeneous effects of a particular type of local revenue—bond funds for capital investments—on achievement by socioeconomic status.

Tests whether text-message parenting programs are more effective when timed to periods when parents most need support or have more time with their children.

Reviews recent rigorous studies of STEM instructional improvement programs to identify what makes teacher professional development and curriculum reforms work.

Applies text-as-data methods to thousands of Washington State school improvement reports to reveal the processes behind reform effects.

Uses the first large-scale panel surveys of social-emotional learning to produce school-level value-added measures for growth mindset, self-efficacy, self-management, and social awareness.

Provides empirical evidence on how China's shift in college admission mechanisms and exam score information affected the academic match between colleges and students.

Meta-analyzes 60 causal studies of teacher coaching, finding pooled effects of 0.49 standard deviations on instruction and 0.18 on student achievement.

Proposes a culturally sensitive framework for incorporating underrepresented populations into large-scale experimental prevention studies.

Explores how the relationship between observed quality of elementary mathematics instruction and student achievement differs across school districts.

Examines how teacher performance ratings change with evaluation system design choices such as measure weights and rating thresholds.

Characterizes intergenerational income mobility at every U.S. college using data on over 30 million students, documenting stark differences in access by parent income.

Uses Add Health sibling data to test whether genotype moderates the transmission of economic standing from parents to children.

Shows, using data from California's CORE districts, that surveys of social-emotional learning and school climate are reliable, valid, and informative measures of school performance.

Replies to Jesse Rothstein's replication and critique of Chetty, Friedman, and Rockoff's findings on teachers' long-term impacts.

Reports findings from a three-year evaluation of a widely available mathematics professional development program designed to build teachers' knowledge and elicit student thinking.

Exploits the timing of 19th-century compulsory schooling laws across states to estimate the causal effect of education on intergenerational occupational mobility.

Shows through Monte Carlo simulations that testing for balance in lagged outcomes does not reliably detect bias in value-added models.

Surveys 1,368 eighth graders on non-cognitive skills, finding that scales predict individual outcomes but behave paradoxically at the school level and in charter comparisons.

Presents new evidence and a new modeling approach for estimating how teacher productivity grows with experience, showing how identifying assumptions can bias results.

Develops methods for estimation and inference when instruments are correlated with an endogenous regressor but also directly affect the outcome.