附註:Expanded, up-to-date version of 1992 ed.
Includes bibliographical references (pages 389-398) and index.
Conditional Specification -- Basic Theorems -- Exact and Almost-Exact Compatibility in Discrete Distributions -- Distributions with Normal Conditionals -- Conditionals in Exponential Families -- Other Conditionally Specified Families -- Impossible Models -- Characterizations Involving Conditional Moments -- Multivariate Extensions -- Parameter Estimation in Conditionally Specified Models -- Simulations -- Marginal and Conditional Specification of Distributions -- Conditional Survival Models -- Bivariate Extreme Models Based on CS -- Bayesian Analysis Using Conditionally Specified Models -- Conditional Specification of Simultaneous Equation Models -- Other Conditional Specification Cases.
摘要:"Any efforts to visualize multivariate densities will necessarily involve the use of cross sections or, equivalently, conditional densities. Distributions that are completely specified in terms of conditional densities are the focus of this book. They form flexible families of multivariate densities that provide natural extensions of many classical multivariate models. They are also used in any modeling situation where conditional information is completely or partially available. In the context of eliciting appropriate priors for multiparameter problems in Bayesian analysis, conditionally specified distributions are particularly convenient. They are effectively tailor-made for Gibbs sampler posterior simulations. All researchers, not just Bayesians, seeking more flexible models than those provided by classical models will find conditionally specified distributions of interest." "This book assumes an introductory course in statistical theory and some familiarity with calculus of several variables, matrix theory, and elementary Markov chain concepts."--Jacket.
