Abstract:
Causal theories in the social sciences often involve abstract concepts that could take on many forms when translated into testable hypotheses. In such tests, researchers rely on concrete realizations of the theoretical intervention, yet not all realizations may similarly move outcomes of interest. In this seminar, I propose an adaptive experimental framework that incorporates both theory testing and theory building. First, I illustrate how adaptive experiments allow us to strategically explore the "theory space" for more informative hypothesis tests. Second, I extend this framework to theory building, addressing cases where minimal theoretical grounding exists. Through adaptive designs, we can improve statistical power, manage multiple testing concerns, and guide experiments toward effective interventions. This approach offers a flexible, principled way to evaluate and refine theories in the face of uncertainty.
Molly Offer-Westort is an Assistant Professor in the Department of Political Science at The University of Chicago, with affiliations in the Department of Statistics, the Committee on Data Science, and the PhD program in political economy. Her research integrates machine learning methods with experimental design to answer causal questions, focusing on online behavior and how individuals respond to information and conversations in digital spaces.