Research

Responses to performance feedback play a critical role in shaping future outcomes in educational and professional contexts. This paper examines whether evaluator gender influences the likelihood that individuals contest feedback. Using an experimental setting in large introductory economics courses, we exploit the random assignment of evaluators with randomly assigned male- or female-sounding names to identify a systematic gender bias: individuals are significantly more likely to contest feedback when it is delivered by an evaluator with a female-sounding name than when similar feedback comes from a male-sounding evaluator. This gender disparity is most pronounced when evaluations fall below students' performance expectations, are more ambiguous, yet are scored neutrally relative to a “fair” assessment. These findings suggest that women in evaluative positions face disproportionate resistance when delivering negative assessments and have implications for their authority, credibility, and career advancement in both educational and workplace settings.

This paper evaluates the effects of universal free school meals (UFM) on the distribution of student achievement. Exploiting the timing of a pandemic-era USDA policy and data from Florida, I estimate UFM's effects on reading and math achievement and chronic absenteeism. Consistent with recent work, I find no evidence of improvements in overall student performance; however, this result masks heterogeneity across the distribution of student achievement. Specifically, I find that UFM significantly reduces the proportion of students performing at both the lowest (inadequacy) and highest (mastery) levels. I argue that these adverse effects among top performers are consistent with peer and class size effects induced by increased attendance of lower-achieving students.

Many educational and professional outcomes depend on performance evaluations. These evaluations might be in the form of admission or hiring decisions, or selection of an award or scholarship/grant, or even loan applications. We aim to identify and explore accuracy in performance evaluations using data from an incentivized peer grading assignment in a large introductory economics course. Through our experimental design, we are able to randomize the quality, perceived identity, and order of a series of assignments to be graded. We find strong evidence of sequential biases: assignments graded first and those graded after a particularly strong assignment tend to receive lower scores. Our findings are consistent with a contrast effect model in which evaluators enter the experiment with overly-optimistic beliefs of student performance. We simulate sequential biases in a structural model and propose policy interventions to limit these biases.

This paper studies changes in contraceptive behavior and fertility decisions following the Dobbs decision. Using Merative MarketScan data and state variation in post-Dobbs abortion bans, we estimate event-study and difference-in-differences models to evaluate changes in short-acting and long-acting contraceptive use, sterilization procedures (including spousal vasectomies), and conceptions.

We observe substantial heterogeneity by age and potential fertility preferences. In states that restricted abortion access, younger women (15–24) increase use of long-acting contraceptives and are less likely to conceive a pregnancy. Among women over age 24, we document a large increase in sterilization procedures within 6 months of Dobbs but no change in pregnancy rates. Interestingly, married perimenopausal women reduce their net contraceptive use and experience an increase in conception, while the opposite is true for pre-perimenopausal women.

To explain these patterns, we develop a model of contraceptive choice in which individuals weigh pregnancy risks, evolving fertility preferences, and access to abortion as a form of insurance against complicated and/or unwanted pregnancies. The model suggests that, for most women, abortion access and contraception function (weakly) as economic substitutes, and that reduced abortion access may lead to increased use of more effective contraceptive methods as well as shifts in the timing of fertility. These implications align with observed empirical patterns, particularly the increased conception rates among older women.

These results have implications for women's labor market outcomes, pharmaceutical policy, and the broader understanding of reproductive decision-making.

An extensive empirical literature on the provision of relative rank information in educational settings has produced mixed results, limiting the ability of researchers and practitioners to draw clear policy recommendations. Some studies find that telling students where they stand relative to their peers improves outcomes, while others find null or even negative effects. We explore a new dimension of this puzzle by shifting the focus from relative performance to relative effort. We develop a theoretical model in which students use feedback to update their beliefs about the returns to effort. In this framework, students face uncertainty about how productive their study activities are and rely on informational signals to calibrate their behavior. We show that relative effort information can be superior to relative performance information because it targets a margin over which students have direct control. Performance rank, by contrast, conflates effort with ability and other factors, making it a noisier signal for guiding future behavior. Our model generates testable predictions about which students are most likely to respond to each type of feedback and in which direction. To test these predictions, we conduct a randomized controlled trial across multiple large introductory economics courses. Students are randomly assigned to one of three conditions: receiving their rank in exam performance, receiving their rank in a quasi-distribution of effort constructed from e-learning platform data, or receiving no additional information beyond their exam score. The effort measure leverages granular learning analytics from the courses' online platforms, capturing engagement behaviors such as time on task, practice problem attempts, and resource usage. This approach allows us to construct a meaningful proxy for effort that is observable at scale and at near-zero marginal cost. Consistent with our theoretical findings, effort rank feedback significantly increases subsequent exam scores, with effects that span approximately the 10th through 70th performance percentiles. This broad range suggests that effort information is useful to a wide cross-section of the student population, including many students who are typically difficult to reach with conventional interventions. Performance rank feedback, by contrast, has no significant average effect and produces gains only at the top decile of the distribution, consistent with a model in which high-performing students are better positioned to extract actionable information from performance signals. Our results highlight how learning analytics generated by increasingly common e-learning platforms can be leveraged to deliver low-cost, scalable, and actionable feedback that meaningfully improves student outcomes. As universities continue to adopt digital learning tools, the data these platforms generate represent an underutilized resource for designing light-touch interventions grounded in behavioral insights.