arXiv · 2611.05456
Aggregating treatment effects across multiple outcomes
Abstract
Researchers often aggregate outcomes to estimate treatment effects on abstract concepts, such as "crime" or "wealth." First, I document that principal component analysis (PCA), inverse variance matrix (IVM) weighting, and standardized averaging (SA) are commonly used but have unexpected shortcomings. PCA can assign negative weights, need not maximize precision, depends on an arbitrary sign normalization, and can have nonstandard asymptotics. Unlike SA, IVM can assign negative weights, but neither necessarily maximizes precision. Second, I develop a statistical decision-theoretic approach that uses mean-squared error to choose nonnegative weights while accounting for outcome quality. I revisit a study on social distancing.
Explore related subjects
Keep this discovery
Explore connections, maps & timelines
Chun Pong Lau. 2026-11-03. Aggregating treatment effects across multiple outcomes. https://searcharxiv.com/abs/2611.05456
Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.