Comparison of Some Sample Selection Procedures In Probability Proportional to Size
No Thumbnail Available
Date
2016
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ICASTOR Journal of Engineering
Abstract
The importance of probability proportional to size sampling cannot be over emphasized. In recent
times, Statisticians often work on complex survey data which involve weighting, as a result of the unequal
probability nature of units. There is need to work on design weights, which is defined as the inverse of the
probability of inclusion of single unit in the sample. Two simulated data sets are generated using ܴin order to
obtain the probability of inclusion of single unit and also the joint probability of inclusion of units in the
sample. Auxiliary variables and variables of interest were generated in such a way that there exists reasonable
correlation between both. A procedure for selecting samples with probability proportional to size nature
without replacement was proposed. This procedure could be used with Horvitz and Thompson estimator of
variance. The variance estimates are obtained using the modified sample selection procedure and an existing
procedure, and results obtained. For the two data sets, it is observed that the modified procedure is more
efficient than the existing one.
Description
Keywords
Inclusion probability, probability sampling, sampling scheme, unequal probability sampling, sampling without replacement.