Bayesian Inference for Partially Identified Models: Exploring the Limits of Limited Data

Many observational studies in epidemiology and other disciplines face inherent limitations in study design and data quality, such as selection bias, unobserved variables, and poorly measured variables. Accessible to statisticians and researchers from various disciplines, this book presents an overview of Bayesian inference in partially identified models. It includes many examples to illustrate the methods and provides R code for their implementation on the book’s website. The author also addresses a number of open questions to stimulate further research in this area.

Table of Contents

Chapter 1: Introduction
Chapter 2: The Structure of Inference in Partially Identified Models
Chapter 3: Partial Identification versus Model Misspecification: Is Honesty Best?
Chapter 4: Further Examples: Models Involving Misclassification
Chapter 5: Further Examples: Models Involving Instrumental Variables
Chapter 6: Further Examples
Chapter 7: Further Topics
Chapter 8: Concluding Thoughts