Download PDF by M. Ghosh, G. Meeden (auth.): Bayesian Methods for Finite Population Sampling

By M. Ghosh, G. Meeden (auth.)

ISBN-10: 0412987716

ISBN-13: 9780412987717

ISBN-10: 1489934162

ISBN-13: 9781489934161

Show description

Read or Download Bayesian Methods for Finite Population Sampling PDF

Best nonfiction_11 books

New PDF release: Hysteresis Effects in Economic Models

This quantity is dedicated to the hysteresis phenomenon in monetary relationships. This subject has got renewed realization in economics specially within the overdue eigh­ ties. because the factor isn't settled there's nonetheless a becoming literature on it. the purpose of this quantity is to summarize the findings, current new effects, and to attract cognizance to extra study.

Additional resources for Bayesian Methods for Finite Population Sampling

Example text

Since the inequality holds for all y in the parameter space and is strict for at least one, the proof is complete. 0 Note that this result is quite general; there are no restrictions on the function to be estimated, the loss function and the design. L(y) denote the population mean. Here we consider the problem of estimating the population mean with squared error loss. Let 8sm denote the estimator which is just the sampie mean, Le. 5) The admissibility of the sampie mean will be demonstrated for two different parameter spaces.

K let cy(i) = number of yj's which equal bi , and for a sampie s cy(i, s) = number of yj's in y(s) which equal bi , and finally for a data point Z = (s, zs) cz(i, s) = number of zj's in Zs which equal bio In the first stage we take as the prior 7fl the distribution which puts mass l/k on the k points (b1, ... ,bdT, ... ,(bk, ... ,bkf and zero mass elsewhere. Under this prior the only points in our sampie space, Z(Y(b),p), which receive positive marginal probability are those where all the observed values are identical.

1. Z)} . 6) tEs Now if 7f is such that for a given Z and for each j f/. z) = Z = 8sm (z) then 8tr (z) = Z = 8sm (z). So if we can produce a finite sequence of priors such that at each step, and for every Z with positive probability at the given step, the conditional expectation of each unobserved unit given the observed data is just the sampie mean the admissibility of 8sm will be proved. The way the argument will proceed is that first we will take care of all data points where all the observed units take on the same value.

Download PDF sample

Bayesian Methods for Finite Population Sampling by M. Ghosh, G. Meeden (auth.)

by Mark

Rated 4.35 of 5 – based on 26 votes