Citation

Robbins HERBERT, Zhang CH. Biometrika 1991; 78(2): 349-354.

Copyright

(Copyright © 1991, Biometrika Trust, Publisher Oxford University Press)

DOI

10.1093/biomet/78.2.349

PMID

unavailable

Abstract

Suppose that, given θ, x is Poisson with mean θ, while, given θ, y is Poisson with mean αθ or βθ according as treatment 1 or treatment 2 is used. Moreover, suppose that treatment 1 is used if x < a, while treatment 2 is used if x≥ a. Under this biased allocation of treatments we give consistent estimators of the two treatment effects α and β. The method also applies when x is binomial and θ is the odds on success, and to other cases.