Working Papers
When Gender Is Seen: How Visibility Shapes Collaboration with Alexander Coutts
Mounting evidence suggests that women face a collaboration penalty, shaping their willingness to collaborate in high-stakes settings. We study an experimental labor market in which workers can choose to work alone or collaborate, with the objective of being hired. We compare behavior across two common-knowledge evaluation regimes: gender-blind and gender-visible evaluation. Women are equally likely to collaborate when gender is hidden, but a gap emerges once gender is revealed. This gap is entirely driven by men, who become less collaborative when gender is visible. These findings challenge common intuitions about the role of gender visibility in labor markets.
Work in Progress
Algorithmic Mediation and Fair Allocation Models with Onur Kesten and Selçuk Özyurt
Mediation is widely used as an alternative to courts, yet legal theorists argue that power imbalances and biases can disadvantage less powerful parties. We study whether algorithmic mediation can produce more impartial outcomes than human mediation in otherwise identical disputes. Our experiments vary both the structure of the mediation mechanism and the visibility of gender to disputants and mediators. We expect fairness rates in algorithmic mediation to remain stable across conditions, while outcomes in human mediation vary systematically with gender visibility. The results speak to the potential of algorithmic tools to reduce persistent inequities in conflict resolution.
Measuring the Impacts of Conflict Resolution Skills on Academic Success with Selçuk Özyurt and Fulya Türk
We conduct a randomized controlled field experiment with secondary school students in Istanbul, TĂĽrkiye. The study evaluates whether training in interpersonal conflict management and dispute resolution skills improves academic performance, peer collaboration, and classroom behavior.
The Design and Pricing of Customized Data
Data is increasingly traded as a good, yet its value depends on buyers’ specific informational needs. We study how a monopolist should design and price information when buyers face the same uncertainty but care about different preferred states. We show that a shared experiment can maximize value and derive optimal pricing under monopoly, first-degree, and second-degree price discrimination. Fully revealing experiments are sometimes optimal, while in other cases deliberately distorted signals increase revenue. The results provide guidance for how credit reporting and analytics platforms should version and price data.
