Journal of Official Statistics, Vol.6, No.2, 1990. pp. 115131

Current Issue
Personal Reference Library (PRL)
Personal Page

Comparisons of Variance Estimators in Stratified Random and Systematic Sampling

Properties of stratified ratio and regression estimators and associated variance estimators can differ considerably under systematic as opposed to random sampling. Large sample properties under the sampling plans are compared theoretically and empirically. A simulation study of estimator performance is presented which examines the effect of different values of certain population parameters. The study contrasts properties under stratified random and systematic sampling and illustrates advantages of the separate regression estimator together with the jackknife variance estimator under a systematic plan. The separate and combined ratio estimators and the combined regression estimator, on the other hand, are less satisfactory. Each requires a more restrictive model in order to be conditionally unbiased than does the separate regression estimator. Even in populations in which the biases of the ratio estimators and the combined regression estimator can be controlled by systematic sampling, the variance estimators studied here can be substantial overestimates.

Jackknife variance estimator; separate ratio estimator; separate regression estimator; superpopulation model.

Copyright Statistics Sweden 1996-2018.  Open Access
ISSN 0282-423X
Created and Maintained by OKS Group