Educational Evaluation and Policy Analysis, Ahead of Print.
Novice teachers improve substantially in their first years on the job, but we know remarkably little about the nature of this skill development. Using data from Tennessee, we leverage a feature of the classroom observation protocol that asks school administrators to identify an item on which the teacher should focus their improvement efforts. This “area of refinement” overcomes a key measurement challenge endemic to inferring from classroom observation scores the development of specific teaching skills. We show that administrators disproportionately identify two skills when observing novice teachers: classroom management and presenting content. Struggling with classroom management, in particular, is associated with early career attrition. Using a returns to experience framework, we observe improvement in these skills among teachers who remain.

Educational Evaluation and Policy Analysis, Ahead of Print.
Retention of teachers of color remains a persistent educational concern, yet little research explores whether teachers of color are less likely to turnover when teaching in schools with principals and peer teaching staff of the same race/ethnicity. This study explores whether principal and peer teacher demographics predict teacher turnover in New York City, and whether they do so differently for teachers of color. We find that Black teachers are less likely to turnover when working in schools with a principal and a higher share of peer teachers of the same race/ethnicity. Results show similar and more consistent patterns for White teachers but no significant difference for Hispanic teachers. We conclude by discussing implications for educational policy, practice, and research.

Educational Evaluation and Policy Analysis, Ahead of Print.
Addressing the educational needs of students in math early on is critical given that early gaps in math skills widen further over the course of schooling. This study examines the effectiveness and costs of Kentucky’s Math Achievement Fund—a unique state-level program that combines targeted interventions, peer-coaching, and close collaboration among teachers to improve math achievement in grades K–3. The program is found to improve not only math achievement, but also reading test scores and non-test outcomes including student attendance and disciplinary incidents. The benefits exist across students from various socioeconomic backgrounds, and they are slightly higher for racial minorities.

Educational Evaluation and Policy Analysis, Ahead of Print.
Preschool has direct effects on attendees, but less is known about how public preschool expansions targeted to low-income children affect early educational experiences and school readiness in a broader community. We use Virginia administrative data (~630,000 students) and a discrete increase in targeted public pre-K (Virginia Preschool Initiative Plus [VPI+]) to estimate the effects of means-tested pre-K expansion on preschool participation patterns and school readiness. VPI+ expansion decreased the probability that children spend no time in licensed settings (6–7 percentage points), increasing the probability for public pre-K and private centers (5–8 and 2–3 percentage points, respectively). Kindergarten literacy skills improve (5%–6% of SD). Effects are larger among the target population but shared broadly (perhaps through private centers). These previously unmeasured benefits inform future policy and research.

Educational Evaluation and Policy Analysis, Ahead of Print.
Teacher evaluation systems and their associated incentives have produced fairly mixed results. Our analyses are motivated by theory and descriptive evidence that accountability systems are highly racialized, and that individuals are less likely to respond to incentives when they have low expectations of success (and vice versa). Using a regression discontinuity design, we find that Black novices in the District of Columbia Public Schools faced the most negative consequences (dismissal threats) and the least benefits (salary incentives), without responding to either. White novices, in contrast, exhibited high expectations of success and large behavior changes, particularly in response to dismissal threats (0.6 SD). We also find some evidence of heterogeneity in effects by task difficulty, often used as a proxy for expectancy, though these differences are less stark.