Details

<p>Preface to the second edition xiii</p> <p>Data acknowledgements xv</p> <p>Acknowledgements xvii</p> <p>Glossary xix</p> <p><b>Part I Foundations 1</b></p> <p><b>1 Introduction 3</b></p> <p>1.1 Reasons for missing data 5</p> <p>1.2 Examples 6</p> <p>1.3 Patterns of missing data 7</p> <p>1.4 Inferential framework and notation 10</p> <p>1.5 Using observed data to inform assumptions about the missingness mechanism 21</p> <p>1.6 Implications of missing data mechanisms for regression analyses 24</p> <p>1.7 Summary 34</p> <p><b>2 The Multiple Imputation Procedure and Its Justification 39</b></p> <p>2.1 Introduction 39</p> <p>2.2 Intuitive outline of the MI procedure 40</p> <p>2.3 The generic MI procedure 45</p> <p>2.4 Bayesian justification of mi 48</p> <p>2.5 Frequentist inference 50</p> <p>2.6 Choosing the number of imputations 55</p> <p>2.7 Some simple examples 56</p> <p>2.8 mi in more general settings 64</p> <p>2.9 Constructing congenial imputation models 72</p> <p>2.10 Discussion 73</p> <p><b>Part II Multiple Imputation for Simple Data Structures 79</b></p> <p><b>3 Multiple Imputation of Quantitative Data 81</b></p> <p>3.1 Regression imputation with a monotone missingness pattern 81</p> <p>3.2 Joint modelling 85</p> <p>3.3 Full conditional specification 90</p> <p>3.4 Full conditional specification versus joint modelling 92</p> <p>3.5 Software for multivariate normal imputation 93</p> <p>3.6 Discussion 93</p> <p><b>4 Multiple Imputation of Binary and Ordinal Data 96</b></p> <p>4.1 Sequential imputation with monotone missingness pattern 96</p> <p>4.2 Joint modelling with the multivariate normal distribution 98</p> <p>4.3 Modelling binary data using latent normal variables 100</p> <p>4.4 General location model 108</p> <p>4.5 Full conditional specification 108</p> <p>4.6 Issues with over-fitting 110</p> <p>4.7 Pros and cons of the various approaches 114</p> <p>4.8 Software 116</p> <p>4.9 Discussion 116</p> <p><b>5 Imputation of Unordered Categorical Data 119</b></p> <p>5.1 Monotone missing data 119</p> <p>5.2 Multivariate normal imputation for categorical data 121</p> <p>5.3 Maximum indicant model 121</p> <p>5.4 General location model 125</p> <p>5.5 FCS with categorical data 128</p> <p>5.6 Perfect prediction issues with categorical data 130</p> <p>5.7 Software 130</p> <p>5.8 Discussion 130</p> <p><b>Part III Multiple Imputation in Practice 133</b></p> <p><b>6 Non-linear Relationships, Interactions, and Other Derived Variables 135</b></p> <p>6.1 Introduction 135</p> <p>6.2 No missing data in derived variables 141</p> <p>6.3 Simple methods 143</p> <p>6.4 Substantive-model-compatible imputation 152</p> <p>6.5 Returning to the problems 165</p> <p><b>7 Survival Data 175</b></p> <p>7.1 Missing covariates in time-to-event data 175</p> <p>7.2 Imputing censored event times 186</p> <p>7.3 Non-parametric, or 'hot deck' imputation 188</p> <p>7.4 Case–cohort designs 191</p> <p>7.5 Discussion 197</p> <p><b>8 Prognostic Models, Missing Data, and Multiple Imputation 200</b></p> <p>8.1 Introduction 200</p> <p>8.2 Motivating example 201</p> <p>8.3 Missing data at model implementation 201</p> <p>8.4 Multiple imputation for prognostic modelling 202</p> <p>8.5 Model building 202</p> <p>8.6 Model performance 204</p> <p>8.7 Model validation 206</p> <p>8.8 Incomplete data at implementation 208</p> <p><b>9 Multi-level Multiple Imputation 213</b></p> <p>9.1 Multi-level imputation model 213</p> <p>9.2 MCMC algorithm for imputation model 224</p> <p>9.3 Extensions 231</p> <p>9.4 Other imputation methods 234</p> <p>9.5 Individual participant data meta-analysis 237</p> <p>9.6 Software 241</p> <p>9.7 Discussion 241</p> <p><b>10 Sensitivity Analysis: MI Unleashed 245</b></p> <p>10.1 Review of MNAR modelling 246</p> <p>10.2 Framing sensitivity analysis: estimands 249</p> <p>10.3 Pattern mixture modelling with mi 251</p> <p>10.4 Pattern mixture approach with longitudinal data via mi 263</p> <p>10.5 Reference based imputation 267</p> <p>10.6 Approximating a selection model by importance weighting 279</p> <p>10.7 Discussion 289</p> <p><b>11 Multiple Imputation for Measurement Error and Misclassification 294</b></p> <p>11.1 Introduction 294</p> <p>11.2 Multiple imputation with validation data 296</p> <p>11.3 Multiple imputation with replication data 301</p> <p>11.4 External information on the measurement process 307</p> <p>11.5 Discussion 308</p> <p><b>12 Multiple Imputation with Weights 312</b></p> <p>12.1 Using model-based predictions in strata 313</p> <p>12.2 Bias in the MI variance estimator 314</p> <p>12.3 MI with weights 317</p> <p>12.4 A multi-level approach 320</p> <p>12.5 Further topics 328</p> <p>12.6 Discussion 329</p> <p><b>13 Multiple Imputation for Causal Inference 333</b></p> <p>13.1 Multiple imputation for causal inference in point exposure studies 333</p> <p>13.2 Multiple imputation and propensity scores 338</p> <p>13.3 Principal stratification via multiple imputation 343</p> <p>13.4 Multiple imputation for IV analysis 346</p> <p>13.5 Discussion 350</p> <p><b>14 Using Multiple Imputation in Practice 355</b></p> <p>14.1 A general approach 355</p> <p>14.2 Objections to multiple imputation 359</p> <p>14.3 Reporting of analyses with incomplete data 363</p> <p>14.4 Presenting incomplete baseline data 364</p> <p>14.5 Model diagnostics 365</p> <p>14.6 How many imputations? 366</p> <p>14.7 Multiple imputation for each substantive model, project, or dataset? 369</p> <p>14.8 Large datasets 370</p> <p>14.9 Multiple imputation and record linkage 375</p> <p>14.10 Setting random number seeds for multiple imputation analyses 377</p> <p>14.11 Simulation studies including multiple imputation 377</p> <p>14.12 Discussion 381</p> <p><b>Appendix A Markov Chain Monte Carlo 384</b></p> <p>A.1 Metropolis Hastings sampler 385</p> <p>A.2 Gibbs sampler 386</p> <p>A.3 Missing data 387</p> <p><b>Appendix B Probability Distributions 388</b></p> <p>B.1 Posterior for the multivariate normal distribution 391</p> <p><b>Appendix C Overview of Multiple Imputation in R, Stata 394</b></p> <p>C.1 Basic multiple imputation using R 394</p> <p>C.2 Basic MI using Stata 395</p> <p>References 398</p> <p>Author Index 419</p> <p>Index of Examples 429</p> <p>Subject Index 431</p>

<p><b>JAMES R. CARPENTER</b> is Professor of Medical Statistics at the London School of Hygiene & Tropical Medicine and Programme Leader in Methodology at the MRC Clinical Trials Unit at UCL, UK. <p><b>JONATHAN W. BARTLETT</b> is a Professor of Medical Statistics at the London School of Hygiene & Tropical Medicine, UK. <p><b>TIM P. MORRIS</b> is Principal Research Fellow in Medical Statistics at the MRC Clinical Trials Unit at UCL, UK. <p><b>ANGELA M. WOOD</b> is Professor of Health Data Science in the Department of Public Health and Primary Care, University of Cambridge, UK. <p><b>MATTEO QUARTAGNO</b> is Senior Research Fellow in Medical Statistics at the MRC Clinical Trials Unit at UCL, UK. <p><b>MICHAEL G. KENWARD</b> retired in 2016 after sixteen years as GlaxoSmithKline Professor of Biostatistics at the London School of Hygiene & Tropical Medicine, UK.

<p><b>Multiple Imputation and its Application</b> <p>Second Edition <p><b>The most up-to-date edition of a bestselling guide to analyzing partially observed data</b> <p>In this comprehensively revised Second Edition of <i>Multiple Imputation and its Application,</i> a team of distinguished statisticians delivers an overview of the issues raised by missing data, the rationale for multiple imputation as a solution, and the practicalities of applying it in a multitude of settings. <p>With an accessible and carefully structured presentation aimed at quantitative researchers, <i>Multiple Imputation and its Application</i> is illustrated with a range of examples and offers key mathematical details. The book includes a wide range of theoretical and computer-based exercises, tested in the classroom, which are especially useful for users of R or Stata. Readers will find: <ul><li>A comprehensive overview of one of the most effective and popular methodologies for dealing with incomplete data sets</li> <li>Careful discussion of key concepts</li> <li>A range of examples illustrating the key ideas</li> <li>Practical advice on using multiple imputation</li> <li>Exercises and examples designed for use in the classroom and/or private study</li></ul> <p>Written for applied researchers looking to use multiple imputation with confidence, and for methods researchers seeking an accessible overview of the topic, <i>Multiple Imputation and its Application</i> will also earn a place in the libraries of graduate students undertaking quantitative analyses.

GB