This study evaluates the association between state repeal of nonmedical vaccine exemptions and kindergarten vaccination rates. Using a staggered difference-in-differences design, we find that total repeal reduced overall exemption rates by 3.2 percentage points within 3 years, with vaccination rates rising by 4.1 points for DTaP, 2.8 for hepatitis B, 4.0 for MMR, and 3.8 for polio. Minimal substitution toward medical exemptions occurred. These findings suggest that nonmedical exemption repeal played a role in maintaining vaccination coverage during a period of heightened vaccine hesitancy.

Adaptive experiments such as multi-armed bandits offer efficiency gains over traditional randomized experiments but pose two major challenges: invalid inference on the Average Treatment Effect (ATE) due to adaptive sampling and low statistical power for sub-optimal treatments. We address both issues by extending the Mixture Adaptive Design framework. First, we propose MADCovar, a covariate-adjusted ATE estimator that is unbiased and preserves anytime-valid inference guarantees while substantially improving ATE precision. Second, we introduce MADMod, which dynamically reallocates samples to underpowered arms, enabling more balanced statistical power across treatments without sacrificing valid inference. Both methods retain MAD's core advantage of constructing asymptotic confidence sequences that allow researchers to continuously monitor ATE estimates and stop data collection once a desired precision or significance criterion is met. Empirically, MADCovar reduces CS width by up to 60% relative to MAD in simulations, and achieves similar precision gains in a large-scale political RCT with 32,000 participants. MADMod improves statistical power across all treatment arms, particularly for suboptimal treatments. Both methods are implemented through an open-source software package.

We use comprehensive administrative data from Rhode Island to measure the impact of early-life interventions for low birth weight newborns on later-life outcomes. We use a regression discontinuity design based on the 1,500-gram threshold for Very Low Birth Weight (VLBW) status. We show that threshold crossing causes more intense in-hospital care, in line with prior studies. Threshold crossing also causes a 0.34 standard deviation increase in test scores in elementary and middle school, a 17.1 percentage point increase in the probability of college enrollment, and $66,997 decrease in social program expenditures by age 14. We explore potential mechanisms driving these impacts.

AB2020 established a deposit-refund system in California, where consumers are reimbursed the California Redemption Value (CRV) upon recycling eligible containers at a recycling center. We study recycling under this policy, focusing on consumer convenience, reported recycling and diversion behavior, and responses to changes in the CRV amount. We find that consumers prefer nearby centers with flexible operating hours and short waiting times. While the CRV induces recycling, an increase in CRV would not lead to major recycling increases, given the limited number of containers entering trash streams. Finally, most diverted containers are taken from trash streams, not curbside recycling bins.

The COVID-19 public health emergency caused widespread economic shutdown and unemployment. The resulting surge in Unemployment Insurance claims threatened to overwhelm legacy systems. In Rhode Island, we developed a scalable cloud solution to collect Pandemic Unemployment Assistance claims, deployed within 10 days of the passage of the CARES Act — making Rhode Island the first state to collect, validate, and pay PUA claims. Our system delivered timely benefits with an estimated cost savings of $502,000, representing a 411% return on investment.