Welcome!

Policymakers often choose a policy bundle that is a combination of different interventions in different dosages. We develop a new technique—treatment variant aggregation (TVA)—to select a policy from a large factorial design. TVA pools together policy variants that are not meaningfully different and prunes those deemed ineffective. This allows us to restrict attention to aggregated policy variants, consistently estimate their effects on the outcome, and estimate the best policy effect adjusting for the winner's curse. We apply TVA to a large randomized controlled trial that tests interventions to stimulate demand for immunization in Haryana, India. The policies under consideration include reminders, incentives, and local ambassadors for community mobilization. Cross-randomizing these interventions, with different dosages or types of each intervention, yields 75 combinations. The policy with the largest impact (which combines incentives, ambassadors who are information hubs, and reminders) increases the number of immunizations by 44% relative to the status quo. The most cost-effective policy (information hubs, ambassadors, and SMS reminders, but no incentives) increases the number of immunizations per dollar by 9.1% relative to the status quo.

The paucity of public health messages that directly address communities of color might contribute to racial and ethnic disparities in COVID-19–related knowledge and behavior. This randomized trial examined whether physician-delivered prevention messages affect knowledge and information-seeking behavior of Black and Latinx persons and if this differs according to the race/ethnicity of the physician and the tailored content.

In this paper, we describe a quantum algorithm for computing an isogeny between any two supersingular elliptic curves defined over a given finite field. The complexity of our method is in \tilde{O}(p^{1/4}) where p is the characteristic of the base field.  Our method is an asymptotic improvement over the previous fastest known method which had complexity \tilde{O}(p^{1/2}) (on both classical and quantum computers). We also discuss the cryptographic relevance of our algorithm

We study the instrumental value of conceptual knowledge when making statistical decisions. Such knowledge tells agents how unknown, payoff-relevant states relate. It is distinct from the statistical knowledge gained from observing signals of those states. We formalize this distinction in a tractable framework used by economists and statisticians. Conceptual knowledge is valuable because it empowers agents to design more informative signals. It is more valuable when states are more “reducible”: when they can be explained with fewer common concepts. Its value is non-monotone in the number of signals and vanishes when agents have infinitely many signals. Agents who know more concepts can attain the same payoffs with fewer signals. This is especially true when states are highly reducible.

In both observational data and randomized control trials, researchers select statistical models to articulate how the outcome of interest varies with combinations of observable covariates. Choosing a model that is too simple can obfuscate important heterogeneity in outcomes between covariate groups, while too much complexity risks identifying spurious patterns. In this paper, we propose a novel Bayesian framework for model uncertainty called Rashomon Partition Sets (RPSs). The RPS consists of all models that have posterior density close to the maximum a posteriori (MAP) model. We construct the RPS by enumeration, rather than sampling, which ensures that we explore all models models with high evidence in the data, even if they offer dramatically different substantive explanations. We use a l0 prior, which allows the allows us to capture complex heterogeneity without imposing strong assumptions about the associations between effects, showing this prior is minimax optimal from an information-theoretic perspective. We characterize the approximation error of (functions of) parameters computed conditional on being in the RPS relative to the entire posterior. We propose an algorithm to enumerate the RPS from the class of models that are interpretable and unique, then provide bounds on the size of the RPS. We give simulation evidence along with three empirical examples: price effects on charitable giving, heterogeneity in chromosomal structure, and the introduction of microfinance.

Learning about technologies in heterogeneous environments is difficult: what works for others may not work for oneself. We evaluate two approaches: delivering tailored information at varying resolutions and providing generalizable knowledge of the production function so farmers can adapt practices to their own contexts. These approaches may substitute for or complement each other. We cross-randomize soil-testing resolution (number of tests per village) and a generalizable soil-fertility training to measure effects on smallholders’ beliefs, fertilizer customization, and farming outcomes. By randomizing test resolution and giving each farmer recommendations from the nearest soil sample, we estimate how the “signal translation function” changes with contextual distance. By independently randomizing the training, we test whether stronger mental models enhance or substitute for data relevance, identifying the value of models by how they improve signal translation. We also inform policy: how many soil tests should be conducted at scale, and can farmer training complete the “last mile” of fertilizer customization?

Many models fit the data well. But each model embeds its own conceptual structure—and with it, its own normative implications. This paper introduces a framework for surfacing and reasoning across these competing stories. We define the Rashomon Concept Ensemble (RCE): the set of semantically distinct, near-optimal models, each aligned to a dictionary of interpretable concepts. While this might seem abstract, we show that the RCE is statistically well-defined, finitely enumerable under mild assumptions, and tractable to compute. Once this space is recovered, we evaluate policy fragility and define robust interventions that perform well across interpretive disagreement. We also provide concrete guidance for design: how to construct pilots that reveal conceptual structure, and how to use heterogeneity in observational data as a diagnostic for semantic instability. The result is a framework that makes normative disagreement legible—and offers a structured alternative to approaches that rely too heavily on any one model's story.

Using data that we collected from over fifty rural communities around the world, we analyze the relationship between material wealth, inequality in wealth, and a variety of social networks that connect "sharing units" (households).  Looking within sites, we find significant, positive relationships between sharing units' access to and provisioning of support, on the one hand, and their material wealth on the other. We also find a form of wealth homophily in that sharing units that have a greater fraction of their connections to wealthy sharing units are significantly wealthier themselves. On the second level, we examine how network features---as they vary across sites---predict which sites have greater wealth inequality. Sites that have less economic connectedness (links across wealth levels) have significantly greater wealth inequality. In contrast, we do not find a correlation between the variance in sharing units' access to and provisioning of support within a site and its wealth Gini.