Saved in:
Bibliographic Details
Main Authors: Azriel, David, Krieger, Abba M., Kapelner, Adam
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2402.07247
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929578191093760
author Azriel, David
Krieger, Abba M.
Kapelner, Adam
author_facet Azriel, David
Krieger, Abba M.
Kapelner, Adam
contents We consider the general performance of the difference-in-means estimator in an equally-allocated two-arm randomized experiment under common experimental endpoints such as continuous (regression), incidence, proportion, count and uncensored survival. We consider two sources of randomness: the subject-specific assignments and the contribution of unobserved subject-specific measurements. We then examine mean squared error (MSE) performance under a new, more realistic "simultaneous tail criterion". We prove that the pairwise matching design of Greevy et al. (2004) performs best asymptotically under this criterion when compared to other blocking designs. We also prove that the optimal design must be less random than complete randomization and more random than any deterministic, optimized allocation. Theoretical results are supported by simulations in all five response types.
format Preprint
id arxiv_https___arxiv_org_abs_2402_07247
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle The Pairwise Matching Design is Optimal under Extreme Noise and Assignments
Azriel, David
Krieger, Abba M.
Kapelner, Adam
Methodology
We consider the general performance of the difference-in-means estimator in an equally-allocated two-arm randomized experiment under common experimental endpoints such as continuous (regression), incidence, proportion, count and uncensored survival. We consider two sources of randomness: the subject-specific assignments and the contribution of unobserved subject-specific measurements. We then examine mean squared error (MSE) performance under a new, more realistic "simultaneous tail criterion". We prove that the pairwise matching design of Greevy et al. (2004) performs best asymptotically under this criterion when compared to other blocking designs. We also prove that the optimal design must be less random than complete randomization and more random than any deterministic, optimized allocation. Theoretical results are supported by simulations in all five response types.
title The Pairwise Matching Design is Optimal under Extreme Noise and Assignments
topic Methodology
url https://arxiv.org/abs/2402.07247