Randomization Inference for Composite Experiments with Spillovers and Peer Effects

TL;DR

This paper develops methods for valid inference and optimal design in group-formation experiments with spillovers and external interventions, extending existing nonparametric approaches to more complex, real-world settings.

Contribution

It introduces Fisherian and Neymanian inference techniques for composite experiments with spillovers, and proposes an optimal design framework for such experiments.

Findings

Constructed Fisherian randomization tests for composite experiments with spillovers.

Developed Neymanian asymptotic confidence intervals accommodating external interventions.

Provided a methodology for designing optimal composite experiments.

Abstract

Group-formation experiments, in which experimental units are randomly assigned to groups, are a powerful tool for studying peer effects in the social sciences. Existing design and analysis approaches allow researchers to draw inference from such experiments without relying on parametric assumptions. In practice, however, group-formation experiments are often coupled with a second, external intervention, that is not accounted for by standard nonparametric approaches. This note shows how to construct Fisherian randomization tests and Neymanian asymptotic confidence intervals for such composite experiments, including in settings where the second intervention exhibits spillovers. We also propose an approach for designing optimal composite experiments.