POSDEM: The Selection Process Between Probability Sampling Plans

by Gonzalo Sánchez-Crespo and Alberto Lezcano .

Abstract: Sampling theory proves, under certain conditions, that if population has a linear trend, Yates's method of end corrections presents a least expected error. If it is a linear trend with periodic variation, then modified systematic sampling is better, and in case of a parabolic trend or autocorrelated population, the centered method is the best. Several alternatives of systematic sampling and unequal probabilities sampling are frequently used. However both may be disturbed by the structure of the population investigated. Trends, cycles and inappropriate relationship between variables are present more often that it seems, which could even lead to losses in accuracy. In both cases, it is necessary to study the relationship between the sampling plans and the population frame before applying these methods. The main goal of this research is that it has been done a new computer program in order to study a new systematic approach and a new indicator to evaluate sampling methods.

The posdem software is available; to be used in research projects, for educational purposes and in the preparatory legwork of the surveys based on samples. The help files could be used like as a book of sampling techniques. Allows one to carry out sampling plans adapted to each investigation and incorporating the expert sampling knowledge, without the person carrying out the sampling design, necessarily needing that knowledge. http://www.geocities.com/ResearchTriangle/Facility/1075/POSDEM/default.htm

Key Words: Systematic sampling, Variable sampling interval, Probabilities proportional to size, Superpopulation model, Computer application: POSDEM

Gonzalo Sánchez-Crespo, gsanchez@ine.es
Alberto Lezcano

Editor: Suojin Wang , sjwang@PICARD.tamu.edu

READING THE ARTICLE: You can read the article in portable document (.pdf) format (618987 bytes.)

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 2571 times since July 24, 2006.

Return to the InterStat Home Page.