| |
| |
Preface | |
| |
| |
Acknowledgements | |
| |
| |
| |
Preliminaries | |
| |
| |
| |
Introduction | |
| |
| |
| |
From Imbalance to the Field of Missing Data Research | |
| |
| |
| |
Incomplete Data in Clinical Studies | |
| |
| |
| |
MAR, MNAR, and Sensitivity Analysis | |
| |
| |
| |
Outline of the Book | |
| |
| |
| |
Key Examples | |
| |
| |
| |
Introduction | |
| |
| |
| |
The Vorozole Study | |
| |
| |
| |
The Orthodontic Growth Data | |
| |
| |
| |
Mastitis in Dairy Cattle | |
| |
| |
| |
The Depression Trials | |
| |
| |
| |
The Fluvoxamine Trial | |
| |
| |
| |
The Toenail Data | |
| |
| |
| |
Age-Related Macular Degeneration Trial | |
| |
| |
| |
The Analgesic Trial | |
| |
| |
| |
The Slovenian Public Opinion Survey | |
| |
| |
| |
Terminology and Framework | |
| |
| |
| |
Modelling Incompleteness | |
| |
| |
| |
Terminology | |
| |
| |
| |
Missing Data Frameworks | |
| |
| |
| |
Missing Data Mechanisms | |
| |
| |
| |
Ignorability | |
| |
| |
| |
Pattern-Mixture Models | |
| |
| |
| |
Classical Techniques and the Need for Modelling | |
| |
| |
| |
A Perspective on Simple Methods | |
| |
| |
| |
Introduction | |
| |
| |
| |
Simple Methods | |
| |
| |
| |
Problems with Complete Case Analysis and Last Observation Carried Forward | |
| |
| |
| |
Using the Available Cases: a Frequentist versus a Likelihood Perspective | |
| |
| |
| |
Intention to Treat | |
| |
| |
| |
Concluding Remarks | |
| |
| |
| |
Analysis of the Orthodontic Growth Data | |
| |
| |
| |
Introduction and Models | |
| |
| |
| |
The Original, Complete Data | |
| |
| |
| |
Direct Likelihood | |
| |
| |
| |
Comparison of Analyses | |
| |
| |
| |
Example SAS Code for Multivariate Linear Models | |
| |
| |
| |
Comparative Power under Different Covariance Structures | |
| |
| |
| |
Concluding Remarks | |
| |
| |
| |
Analysis of the Depression Trials | |
| |
| |
| |
View 1: Longitudinal Analysis | |
| |
| |
| |
Views 2a and 2b and All versus Two Treatment Arms | |
| |
| |
| |
Missing at Random and Ignorability | |
| |
| |
| |
The Direct Likelihood Method | |
| |
| |
| |
Introduction | |
| |
| |
| |
Ignorable Analyses in Practice | |
| |
| |
| |
The Linear Mixed Model | |
| |
| |
| |
Analysis of the Toenail Data | |
| |
| |
| |
The Generalized Linear Mixed Model | |
| |
| |
| |
The Depression Trials | |
| |
| |
| |
The Analgesic Trial | |
| |
| |
| |
The Expectation-Maximization Algorithm | |
| |
| |
| |
Introduction | |
| |
| |
| |
The Algorithm | |
| |
| |
| |
Missing Information | |
| |
| |
| |
Rate of Convergence | |
| |
| |
| |
EM Acceleration | |
| |
| |
| |
Calculation of Precision Estimates | |
| |
| |
| |
A Simple Illustration | |
| |
| |
| |
Concluding Remarks | |
| |
| |
| |
Multiple Imputation | |
| |
| |
| |
Introduction | |
| |
| |
| |
The Basic Procedure | |
| |
| |
| |
Theoretical Justification | |
| |
| |
| |
Inference under Multiple Imputation | |
| |
| |
| |
Efficiency | |
| |
| |
| |
Making Proper Imputations | |
| |
| |
| |
Some Roles for Multiple Imputation | |
| |
| |
| |
Concluding Remarks | |
| |
| |
| |
Weighted Estimating Equations | |
| |
| |
| |
Introduction | |
| |
| |
| |
Inverse Probability Weighting | |
| |
| |
| |
Generalized Estimating Equations for Marginal Models | |
| |
| |
| |
Weighted Generalized Estimating Equations | |
| |
| |
| |
The Depression Trials | |
| |
| |
| |
The Analgesic Trial | |
| |
| |
| |
Double Robustness | |
| |
| |
| |
Concluding Remarks | |
| |
| |
| |
Combining GEE and MI | |
| |
| |
| |
Introduction | |
| |
| |
| |
Data Generation and Fitting | |
| |
| |
| |
MI-GEE and MI-Transition | |
| |
| |
| |
An Asymptotic Simulation Study | |
| |
| |
| |
Concluding Remarks | |
| |
| |
| |
Likelihood-Based Frequentist Inference | |
| |
| |
| |
Introduction | |
| |
| |
| |
Information and Sampling Distributions | |
| |
| |
| |
Bivariate Normal Data | |
| |
| |
| |
Bivariate Binary Data | |
| |
| |
| |
Implications for Standard Software | |
| |
| |
| |
Analysis of the Fluvoxamine Trial | |
| |
| |
| |
The Muscatine Coronary Risk Factor Study | |
| |
| |
| |
The Crepeau Data | |
| |
| |
| |
Concluding Remarks | |
| |
| |
| |
Analysis of the Age-Related Macular Degeneration Trial | |
| |
| |
| |
Introduction | |
| |
| |
| |
Direct Likelihood Analysis of the Continuous Outcome | |
| |
| |
| |
Weighted Generalized Estimating Equations | |
| |
| |
| |
Direct Likelihood Analysis of the Binary Outcome | |
| |
| |
| |
Multiple Imputation | |
| |
| |
| |
Concluding Remarks | |
| |
| |
| |
Incomplete Data and SAS | |
| |
| |
| |
Introduction | |
| |
| |
| |
Complete Case Analysis | |
| |
| |
| |
Last Observation Carried Forward | |
| |
| |
| |
Direct Likelihood | |
| |
| |
| |
Weighted Estimating Equations | |
| |
| |
| |
Multiple Imputation | |
| |
| |
| |
Missing Not at Random | |
| |
| |
| |
Selection Models | |
| |
| |
| |
Introduction | |
| |
| |
| |
The Diggle-Kenward | |