A Geometric Approach to Finite Population Sampling
share
A new, geometric and easy to understand approach to finite population sampling is presented. In this approach, first-order inclusion probabilities (FIPs) are represented by bars in a two-dimensional coordinate system, and their different arrangements lead to different designs. Only based on the geometric approach, designs can be fully implemented, without the need for mathematical algorithms. An special arrangement of the bars, is equivalent to Madow 1949 systematic method, which easily, with rearranging the bars, while keeping the FIPs unchanged, results in different second-order inclusion probabilities (SIPs) equivalent to other famous finite population sampling designs, such as Poisson sampling, maximum entropy sampling, etc. Geometric visualization of sampling designs leads to increased creativity of researchers to provide new efficient designs. This approach opens a new gate to finite population sampling that can deal with problems such as optimal designs, implementation of maximum entropy sampling, etc.
READ FULL TEXT