A cost analysis of ranked set sampling to estimate a population mean

Ranked set sampling (RSS) can be a useful environmental sampling method when measurement costs are high but ranking costs are low. RSS estimates of the population mean can have higher precision than estimates from a simple random sample (SRS) of the same size, leading to potentially lower sampling costs from RSS than from SRS for a given precision. However, RSS introduces ranking costs not present in SRS; these costs must be considered in determining whether RSS is cost effective. We use a simple cost model to determine the minimum ratio of measurement to ranking costs (cost ratio) necessary in order for RSS to be as cost effective as SRS for data from the normal, exponential, and lognormal distributions. We consider both equal and unequal RSS allocations and two types of estimators of the mean: the typical distribution-free (DF) estimator and the best linear unbiased estimator (BLUE). The minimum cost ratio necessary for RSS to be as cost effective as SRS depends on the underlying distribution of the data, as well as the allocation and type of estimator used. Most minimum necessary cost ratios are in the range of 1–6, and are lower for BLUEs than for DF estimators. The higher the prior knowledge of the distribution underlying the data, the lower the minimum necessary cost ratio and the more attractive RSS is over SRS. Copyright © 2005 John Wiley & Sons, Ltd.

[1]  S. Stokes,et al.  Estimation of Variance Using Judgment Ordered Ranked Set Samples , 1980 .

[2]  Ganapati P. Patil,et al.  Environmental sampling with a concomitant variable: A comparison between ranked set sampling and stratified simple random sampling , 1996 .

[3]  Narayanaswamy Balakrishnan,et al.  Handbook of Tables for Order Statistics from Lognormal Distributions with Applications , 1999 .

[4]  Probability proportional to size (πps) sampling using ranks , 1990 .

[5]  Lynne Stokes,et al.  Parametric ranked set sampling , 1995, Annals of the Institute of Statistical Mathematics.

[6]  E. H. Lloyd.  LEAST-SQUARES ESTIMATION OF LOCATION AND SCALE PARAMETERS USING ORDER STATISTICS , 1952 .

[7]  Ganapati P. Patil,et al.  Optimal Allocation for Symmetric Distributions in Ranked Set Sampling , 2000 .

[8]  Ganapati P. Patil,et al.  UNEQUAL ALLOCATION MODELS FOR RANKED SET SAMPLING WITH SKEW DISTRIBUTIONS , 1997 .

[9]  P. Yu,et al.  Regression estimator in ranked set sampling. , 1997, Biometrics.

[10]  Vic Barnett,et al.  Best linear unbiased estimates in ranked-set sampling with particular reference to imperfect ordering , 1997 .

[11]  Z. Bai,et al.  Ranked set sampling , 2004 .

[12]  S. Stokes,et al.  Ranked set sampling with concomitant variables , 1977 .

[13]  Nicolle A. Mode,et al.  Incorporating prior knowledge in environmental sampling: ranked set sampling and other double sampling procedures , 2002 .

[14]  J. L. Clutter,et al.  Ranked Set Sampling Theory with Order Statistics Background , 1972 .

[15]  Nicolle A. Mode,et al.  Ranked set sampling for ecological research: accounting for the total costs of sampling , 1999 .

[16]  John R. Skalski,et al.  Detection of contaminated samples using a two‐way compositing scheme , 1994 .

[17]  David G. Hankin,et al.  Estimating Total Fish Abundance and Total Habitat Area in Small Streams Based on Visual Estimation Methods , 1988 .

[18]  A. E. Sarhan,et al.  Contributions to order statistics , 1964 .

[19]  G. McIntyre,et al.  A method for unbiased selective sampling, using ranked sets , 1952 .

[20]  Ramzi W Nahhas,et al.  Ranked set sampling: cost and optimal set size. , 2002, Biometrics.

[21]  Kazumasa Wakimoto,et al.  On unbiased estimates of the population mean based on the sample stratified by means of ordering , 1968 .

[22]  Hassen A. Muttlak.  Median ranked set sampling with concomitant variables and a comparison with ranked set sampling and regression estimators , 1998 .

[23]  N. Balakrishnan,et al.  Lognormal Distributions and Properties , 1999 .