Does Simple Random Sampling Provide Adequate Balance?
W. G. Cumberland, R. M. Royall
Journal of the Royal Statistical Society: Series B (Statistical Methodology) · 1988 · Volume 50 · Issue 1
DOI: 10.1111/j.2517-6161.1988.tb01717.x
Abstract
Theoretical arguments imply that, as the sample size increases, simple random sampling does not force balance fast enough to provide robust inference for a finite population total using the ratio estimator. An empirical study shows that under simple random sampling, confidence intervals for the total can have conditional coverage probabilities that are persistently awful as the sample size grows.
Keywords
finite population inference, random sampling, ratio estimator, confidence intervals, robustness, prediction
Enable JavaScript for citation tools and additional paper interactions.