Composite Sampling: Some Problems in Detection and Estimation
ABSTRACT:
Composite sampling techniques have been extensively used in the environmental
and medical literature as a cost efficient alternative to traditional sampling
plans. They have also been effectively used to determine prevalence rates of
diseases where it is necessary to protect the confidentiality of the
individual. A composite is formed by combining multiple sample units to create
a new larger unit on which subsequent measurements are made. The data on the
composites may then be used to estimate the fraction defective, or in
detection of all units possessing a certain trait. I will discuss some group
screening procedures and evaluate their performance relative to the standard
single-stage sampling plans.
About the Speaker
Dr Nandini Kannan is an Associate Professor in the Division of Mathematics
and Statistics at the University of Texas at San Antonio. She received her
Ph.D. degree in Statistics from Pennsylvania State University in 1992.
After receiving her Ph.D., she joined the faculty at San Antonio. Her
research interests include Signal Processing, Multivariate Analysis,
Survival Analysis, and Inference.
References
The following references give some background to the subject of
Dr. Kanaan's seminar. They are all available in the Thode Library.
V. Lancaster and S. Keller-McNulty (1998) A Review of Composite Sampling
Methods. Journal of the
American Statistical Association, 93, 1216-1230
N. Kannan and A. Raychaudhuri (1994) Composite Sampling: Effectiveness
Under a Two Stage Scheme. Communications in Statistics: Theory and
Methods,
23 2533-2540.
A. Raychaudhuri and N. Kannan (1996) Inverse Sampling: A Composite
Approach. Communications in Statistics: Theory and Methods,
25 1033-1042