Research

Working Papers

Evaluating Behavioral Interventions at Scale with AI

with Felix Chopra, Ingar Haaland and Christopher Roth

We test the effectiveness of different AI-delivered conversation protocols to increase people's motivation for change. In a large-scale experiment with 2,719 social media users, we randomly assign participants to a control conversation or one of three treatment arms: two Motivational Interviewing protocols promoting self-persuasion (change focus or decisional balance) and a direct persuasion protocol providing unsolicited advice and information. All conversations are led by an AI interviewer, enabling standardized delivery of each protocol at scale. Our results show that all three interventions significantly increase motivation for change and the perceived costs of social media use, with change-focused self-persuasion yielding the largest effects. These effects persist and translate into self-reported reductions in social media use more than two weeks after the intervention. Our findings illustrate how AI-led conversations can serve as a scalable platform both for delivering behavioral interventions and for testing what makes them effective by systematically varying how conversations are conducted.

ECONtribute_385_2026.pdf

The Social Desirability Atlas

with Leonardo Bursztyn, Ingar Haaland and Christopher Roth

Accepted, Journal of Political Economy Microeconomics

Social desirability bias (SDB) is a pervasive threat to the validity of survey and experimental data. Respondents might often misreport sensitive attitudes and behaviors to appear more socially acceptable. We begin by synthesizing empirical evidence on the prevalence and magnitude of SDB across various domains, focusing on studies with individual-level benchmarks. We then critically assess commonly used strategies to mitigate SDB, highlighting how they can sometimes fail by creating confusion or inadvertently increasing perceived sensitivity. To help researchers navigate these challenges, we offer practical guidance on selecting the most suitable tools for different research contexts. Finally, we examine how SDB can distort treatment effects in experiments and discuss mitigation strategies.

w33920.pdf

AI Against Anxiety: Efficacy, Mechanisms, and Demand for Voice-Based AI Therapy

with Bennet Feld

Cognitive Behavioral Therapy (CBT) is among the most effective treatments for depression and anxiety, yet its scalability is limited by cost, availability, and stigma—constraints which artificial intelligence (AI) may help overcome. We evaluate the efficacy of interactive CBT fully administered by a voice-based AI and its downstream effects on demand for traditional mental healthcare in a large-scale randomized controlled trial. First, findings support that AI can effectively deliver CBT. Treated participants show large reductions in symptoms of depression and anxiety relative to the control group. Second, treated participants reduce their demand for human therapy, reporting less fear of judgment and greater comfort disclosing to AI than to a human therapist. Thus, while expanding access to treatment, AI may reshape how patients seek and engage with mental health care.

Work in Progress

Discounting, Transaction Cost, and Information in Bargaining

with Dongkyu Chang, Peter Cramton, Jeongbin Kim and Axel Ockenfels

We study bilateral bargaining with different negotiation costs and information structures in a continuous-time Rubinstein--Cramton framework with per-second transaction costs and exponential discounting. Under one-sided incomplete information, equilibrium features two endogenous cutoffs that split buyer types into immediate terminators, delay-and-signal types, and immediate accepters. We test these predictions in a 2x2 laboratory experiment (N=384) that varies information (one- vs. two-sided private values) and transaction costs. In the one-sided treatments, the cutoff logic organises the data: low types are much more likely to terminate without trade, intermediate types trade at prices increasing in the buyer’s value, and high types frequently accept early at prices close to the predicted opening price. In the two-sided treatments, allocations concentrate near equal splits and efficiency is high once gains from trade are non-trivial. Transaction costs mainly reshape the route of bargaining rather than its destination: negotiations become shorter and involve fewer exchanges, while agreement frontiers and split-normalised allocations remain similar across cost regimes.

Google Sites

Report abuse