Probability sampling designs: Balancing and principles for choice of design
Date issued
December 20, 2017
In
Statistical Science
Vol
2
No
32
From page
176
To page
189
Reviewed by peer
1
Subjects
design-based entropy model-based pivotal
method spatial sampling
Abstract
In this paper, we first aim to formalize the choice of the sampling design for a particular estimation problem. Next several principles are proposed: randomization, over-representation and restriction. These principles are fundamental to assist in the determination of the most appropriate design.
A priori knowledge of the population can be also formalized by modelling the population, which can be helpful when choosing the sampling design. We present a list of sampling designs by specifying their corresponding models and the principles used to derive them. Emphasis is placed on new spatial sampling methods and their related models. A simulation shows the advantages of the different methods.
A priori knowledge of the population can be also formalized by modelling the population, which can be helpful when choosing the sampling design. We present a list of sampling designs by specifying their corresponding models and the principles used to derive them. Emphasis is placed on new spatial sampling methods and their related models. A simulation shows the advantages of the different methods.
Later version
https://arxiv.org/abs/1612.04965
Publication type
journal article