Journal of Official Statistics, Vol.3, No.3, 1987. pp. 251–265
Estimating the Variance of a Complex Statistic: A Monte Carlo Study of Some Approximate Techniques
Claes Andersson, Goösta Forsman, and Jan Wretman
Abstract:This paper describes a Monte Carlo study designed to illustrate the performance of four approximate techniques of estimating the sampling variance of a ratio between two estimated consumer price indices. The variance estimation techniques under study are: (1) traditional Taylor linearization, (2) Taylor linearization with a simplified variance estimation formula, (3) repeated random groups, and (4) jackknifing. These techniques are compared on the basis of observed relative bias, observed relative mean square error, and observed confidence interval coverage rate in a series of 1 000 samples drawn from the same population of stores. The sampling design is stratified sampling with PPS sampling within each stratum, and the data that are used are authentic price data from a price measurement study.
Keywords:Finite population sampling; variance estimation; complex statistic; consumer price index; Taylor linearization; repeated random groups; jackknife; Monte Carlo study.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group