Relative Accuracy of Systematic and Stratified Random Samples for a Certain Class of Populations

Abstract

Comparison is made of random, stratified random (one element per stratum) and systematic samples where serial correlation exists. Stratified random samples are at least as accurate as random samples. The comparison of systematic and random samples depends on the form of the population. No unbiased estimate of error can be made from a single systematic sample, nor from a stratified random sample with only one element per stratum.