YSPH Biostatistics Seminar: “Design-Based Weighted Regression Estimators for Average and Conditional Spillover Effects"

NOTE: BIS 525 students are required to attend in person. Others are invited to attend in person, but may also attend via Zoom.

SPEAKER: Fei Fang, PhD, Post Doctoral Associate, Department of Biostatistics, Yale University

TITLE: “Design-Based Weighted Regression Estimators for Average and Conditional Spillover Effects"

ABSTRACT: In this paper, we conceptualize general spillover estimands as weighted sums of unit-to-unit spillover effects with estimand-specific weights under partial interference. Building on these estimands, we develop design-based weighted least squares (WLS) estimators for both average and conditional spillover effects. Regression-type estimators are appealing because they are intuitive to construct, straightforward to implement, and—when carefully designed—can yield valid inference even when the underlying outcome structure is complex.

For the average-type estimands, we introduce three constructions of the estimators—the dyadic, sender, and receiver perspectives—which distribute the estimand weights differently across the outcome vector, design matrix, and weight matrix. We show that all three perspective estimators are equivalent to the Hájek estimator. To extend this framework to conditional spillover effects, we construct parametric WLS estimators and we establish conditions under which they are consistent for the target conditional spillover effects.

We further derive concentration inequalities, a central limit theorem, and conservative variance estimators for the three perspective estimators in an asymptotic regime where both the number of clusters and cluster sizes grow, thereby providing a unified theoretical framework for regression-based spillover estimation under partial interference. We use simulation studies to evaluate the performance of the proposed estimators for average spillover effects and to examine the strength of the theoretical conditions linking the different estimators to the conditional spillover estimands. Finally, we demonstrate the utility of our methods by re-analyzing a randomized experiment by Cai et al. (2015), which studied the conditional spillover effects of an information session on the uptake of weather insurance among rice farmers in China.

YSPH values inclusion and access for all participants. If you have questions about accessibility or would like to request an accommodation, please contact Charmila Fernandes at Charmila.fernandes@yale.edu. We will try to provide accommodations requested by November 24, 2025.