You-Gan Wang
CSIRO Mathematical and Information
Sciences, CSIRO Long Pocket Laboratories,
120 Meiers Road, Indooroopilly, Queensland
4068, Australia
This talk considers how to efficiently select
sampling sites or subsamples from a finite number of samples by
borrowing information from a concomitant variable (e.g. historical data) that are
correlated with the outcome variable of interest. When such concomitant
variables are available, the new design can reduce cost or increase
efficiency greatly at a fixed cost (sample size) compared with simple random
sampling. The merits of the methodology are illustrated by two
studies. Using the trawl survey data from the Northern Prawn Fisheries, the
relative efficiencies in terms of the variance and MSE of the estimated mean
abundance indices are estimated to range from 100% to 200%. Using
data from the aging study on Tenualosa
ilisha, we show that the efficiency of age
prediction using weight data is as high as 140% when
compared with simple random sampling.
This methodology has wide applications in envirometrics, genetics,
fisheries and other areas.
For more details about the speaker, please visit: http://www.cmis.csiro.au/You-Gan.Wang/