Premium
Relative precision of stratified sampling, sampling with probability of selection proportional to size, and simple random sampling with ratio estimation
Author(s) -
Jensen A. L.
Publication year - 1991
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/env.3770020406
Subject(s) - statistics , stratified sampling , sampling (signal processing) , mathematics , simple random sample , sampling design , variable (mathematics) , bivariate analysis , sample size determination , poisson sampling , slice sampling , population , importance sampling , monte carlo method , computer science , mathematical analysis , demography , filter (signal processing) , sociology , computer vision
Variably of estimates in natural resources and the environmental sciences is high, and methods for increasing precision are important. Three different approaches have been applied to increase precision of sample based estimates by including auxiliary variables: these are stratified sampling, sampling with probability of selection proportion to size, and ratio estimation. The objective of this study was to determine the conditions under which each of these methods had a small error of estimate. Bivariate normal populations were generated with different coefficients of variation for both the variable of interest and the auxiliary variable, and with different correlations between the auxilliary variable, and with different correlations between the auxiliary variable and the variable of interest. All three approaches for increasing precision were applied to estimate the population means and determine the precision of the estimates relative to that of simple random sampling. Stratified sampling was the most precise method unless the coefficients of variation of both the variable of interest and the auxiliary variable were small and the correlation was high. This condition does not occur often in natural resources or the environmental sciences, and stratified sampling most often would be the best method for increasing precision.