Super-Population Modeling. Model-Assisted Approach. Asymptotics
Document Type
Book Chapter
Publication Title
Indian Statistical Institute Series
Abstract
Purely design-based classical procedures of survey sampling, though analytically impressive, are deficient because positively conclusive results are hard to come by, efforts mostly ending in wild goose chase. The vector Y̲=(y1,…,yi,…,yN) of a finite number of real values allowed to be totally unrestricted cannot yield serviceably best results in search for estimation of parameters. So, if it is considered right to take initially as a random vector with a ‘probability distribution’ admitting a theoretical population called a ‘super-population’ contrasting against the finite population U= (1, …, i, …, N), then an inference approach in an alternate way turns out feasible. The probability distribution of Y need not be too specific. It nay be a member in a wide “class” of probability distributions, called a “Super-population model” admitting only a few low order moments. Then it is possible to develop and work out procedures for estimating parameters with alternative criteria discussed in this chapter.
First Page
159
Last Page
167
DOI
10.1007/978-981-19-1418-8_7
Publication Date
1-1-2022
Recommended Citation
Chaudhuri, Arijit and Pal, Sanghamitra, "Super-Population Modeling. Model-Assisted Approach. Asymptotics" (2022). Book Chapters. 165.
https://digitalcommons.isical.ac.in/book-chapters/165