Explanation in Causal Inference: Methods for Mediation and Interaction by Tyler VanderWeeleExplanation in Causal Inference: Methods for Mediation and Interaction by Tyler VanderWeele

Explanation in Causal Inference: Methods for Mediation and Interaction

byTyler VanderWeele

Hardcover | March 27, 2015

Pricing and Purchase Info

$84.56 online 
$110.95 list price save 23%
Earn 423 plum® points

Ships within 1-3 weeks

Ships free on orders over $25

Not available in stores


The book provides an accessible but comprehensive overview of methods for mediation and interaction. There has been considerable and rapid methodological development on mediation and moderation/interaction analysis within the causal-inference literature over the last ten years. Much of thismaterial appears in a variety of specialized journals, and some of the papers are quite technical. There has also been considerable interest in these developments from empirical researchers in the social and biomedical sciences. However, much of the material is not currently in a format that isaccessible to them. The book closes these gaps by providing an accessible, comprehensive, book-length coverage of mediation. The book begins with a comprehensive introduction to mediation analysis, including chapters on concepts for mediation, regression-based methods, sensitivity analysis, time-to-event outcomes, methods for multiple mediators, methods for time-varying mediation and longitudinal data, and relationsbetween mediation and other concepts involving intermediates such as surrogates, principal stratification, instrumental variables, and Mendelian randomization. The second part of the book concerns interaction or "moderation," including concepts for interaction, statistical interaction, confoundingand interaction, mechanistic interaction, bias analysis for interaction, interaction in genetic studies, and power and sample-size calculation for interaction. The final part of the book provides comprehensive discussion about the relationships between mediation and interaction and unites theseconcepts within a single framework. This final part also provides an introduction to spillover effects or social interaction, concluding with a discussion of social-network analyses.The book is written to be accessible to anyone with a basic knowledge of statistics. Comprehensive appendices provide more technical details for the interested reader. Applied empirical examples from a variety of fields are given throughout. Software implementation in SAS, Stata, SPSS, and R isprovided. The book should be accessible to students and researchers who have completed a first-year graduate sequence in quantitative methods in one of the social- or biomedical-sciences disciplines. The book will only presuppose familiarity with linear and logistic regression, and could potentiallybe used as an advanced undergraduate book as well.
Tyler J. VanderWeele, Ph.D., is a methodologist at Harvard University. He holds degrees in biostatistics, mathematics, finance, philosophy and theology and is currently Professor of Epidemiology in the Departments of Epidemiology and Biostatistics at the Harvard School of Public Health and a faculty affiliate of the Institute of Quanti...
Title:Explanation in Causal Inference: Methods for Mediation and InteractionFormat:HardcoverDimensions:728 pages, 9.41 × 6.5 × 1.69 inPublished:March 27, 2015Publisher:Oxford University PressLanguage:English

The following ISBNs are associated with this title:

ISBN - 10:0199325871

ISBN - 13:9780199325870

Look for similar items by category:


Table of Contents

PART I: MEDIATION ANALYSIS1. Explanation and Mechanism1.1. Causal Inference and Explanation1.2. Forms of Explanation and Types of Mechanisms1.3. Motivations for Assessing Mediation, Interaction, and Interference1.4. Organization of the Book2. Mediation: Introduction and Regression-Based Approaches2.1. Classic Regression Approach to Mediation Analysis2.2. Counterfactual Approach to Mediation Analysis: Continuous Outcomes2.3. Assumptions about Confounding2.4. Binary and Count Outcomes2.5. Binary Mediators2.6. Comparison of Approaches: Product-of-Coefficient and Difference Methods2.7. Description of the SAS Macro2.8. Description of the SPSS Macro2.9. Description of the Stata Macro2.10. Hypothetical Example with Output2.11. Empirical Example in Genetic Epidemiology2.12. When to Include an Exposure-Mediator Interaction2.13. Proportion Mediated2.14. Proportion Eliminated2.15. Study Design and Mediation Analysis2.16. Counterfactual Notation for Natural Direct and Indirect Effects2.17. An Alternative Regression-Based Estimation Approach Using Simulations2.18. Code for the Simulation-Based Approach in R2.19. Discussion3. Sensitivity Analysis for Mediation3.1. Sensitivity Analysis for Unmeasured Confounding for Total Effects3.2. Sensitivity Analysis for Unmeasured Confounding for Controlled Direct Effects3.3. Sensitivity Analysis for Unmeasured Confounding for Natural Direct and Indirect Effects3.4. Sensitivity Analysis Using Two Trials3.5. Sensitivity Analysis for Direct and Indirect Effects in the Presence of Measurement Error3.6. Discussion4. Mediation Analysis with Survival Data4.1. Earlier Literature on Mediation Analysis with Survival Models4.2. Mediation Analysis with an Accelerated Failure-Time Model4.3. Mediation Analysis with a Proportional-Hazards Model4.4. Mediation with an Additive-Hazard Model4.5. A Weighting Approach to Direct and Indirect Effects with Survival Outcomes4.6. Sensitivity Analysis with Survival Data4.7. Discussion5. Multiple Mediators5.1. Regression-Based Approaches to Multiple Mediators5.2. A Weighting Approach to Multiple Mediators5.3. Controlled Direct Effects and Exposure-Induced Confounding5.4. Effect Decomposition with Exposure-Induced Confounding5.5. Path-Specific Effects5.6. Sensitivity Analysis for Exposure-Induced Confounding5.7. Discussion6. Mediation Analysis with Time-Varying Exposures and Mediators6.1. Notation and Definitions6.2. Controlled Direct Effects with Time-Varying Exposures and Mediators6.3. Natural Direct and Indirect Effects and Their Randomized Interventional Analogues withTime-Varying Exposures and Mediators6.4. Counterfactual Analysis of MacKinnon's Three-Wave Mediation Model6.5. Discussion7. Selected Topics in Mediation Analysis7.1. Other Estimation Approaches7.2. Ill-Defined Mediators and Multiple Versions of the Mediator7.3. Controversies over Assumptions and Alternative Interpretations of Effects7.4. Direct and Indirect Effects in Health-Disparities Research7.5. Rubin's Seemingly Problematic Examples7.6. A Three-Way Decomposition into Direct, Indirect, and Interactive Effects7.7. Alternative Identification Strategies Using Confounding Control7.8. Identification Using Baseline Covariates That Interact with Exposure7.9. Power and Sample-Size Calculations for Mediation Analysis7.10. Discussion8. Other Topics Related to Intermediates8.1. Principal Stratification8.2. Surrogate Outcomes8.3. Instrumental Variables8.4. Mendelian Randomization8.5. DiscussionPART II: INTERACTION ANALYSIS9. An Introduction to Interaction Analysis9.1. Measures of Interaction and Scale of Interaction9.2. Statistical Interactions and Statistical Inference9.3. Inference for Additive Interaction9.4. SAS and Stata Code for Additive Interaction from Logistic Regression9.5. Additive versus Multiplicative Interaction9.6. Confounding and the Interpretation of Interaction: Interaction versus Effect Heterogeneity9.7. Presenting Interaction Analyses9.8. Synergism and Mechanistic Interaction9.9. Interactions for Continuous Outcomes and Time-to-Event Outcomes9.10. Identifying Subgroups to Target Treatment9.11. Qualitative Interaction9.12. Attributing Effects to Interactions9.13. Discussion10. Mechanistic Interaction10.1. Sufficient Causes and Synergism10.2. Statistical Interaction with No Mechanistic Interaction10.3. Empirical Tests for Sufficient-Cause Synergism10.4. Sufficient-Cause Interaction and Statistical Interactions10.5. "Epistatic" or Singular Interactions10.6. Extensions to Ordinal Exposures10.7. Extensions to Three or More Exposures10.8. Other Extensions10.9. Antagonism10.10. Limits of Inference Concerning Biology10.11. Discussion11. Bias Analysis for Interactions11.1. Sensitivity Analysis and Robustness for Additive Interaction11.2. Sensitivity Analysis and Robustness for Multiplicative Interaction11.3. Sensitivity Analysis for the Relative Excess Risk Due to Interaction11.4. Measurement Error and Additive Interaction11.5. Measurement Error and Multiplicative Interaction11.6. Discussion12. Interaction in Genetics: Independence and Boosting Power12.1. Case-Only Estimators of Interaction12.2. Joint Tests for Interactions and Main Effects12.3. Multiple Testing12.4. Discussion13. Power and Sample-Size Calculations for Interaction Analysis13.1. Power and Sample-Size Calculations for Interaction for Continuous Outcomes13.2. Power and Sample-Size Calculations for Binary Outcomes: Multiplicative Interaction13.3. Power and Sample-Size Calculations for Binary Outcomes: Additive Interaction13.4. Power and Sample-Size Calculations for Binary Outcomes: Mechanistic Interaction13.5. Excel Spreadsheets for Sample-Size and Power Calculations for Additive and Multiplicative Interaction for a Binary Outcome13.6. DiscussionPART III: SYNTHESIS AND SPILLOVER EFFECTS14. A Unification of Mediation and Interaction14.1. Notation and Definitions14.2. Four-Fold Decomposition: The Unification of Mediation and Interaction14.3. Identification of the Effects14.4. Relation to Statistical Models14.5. Binary Outcomes and the Ratio Scale14.6. Illustration in Genetic Epidemiology14.7. Relation to Mediation Decompositions14.8. Relation to Interaction Decompositions14.9. SAS Code for the Four-Way Decomposition14.10. Discussion15. Social Interactions and Spillover Effects15.1. Notation and Definitions for Spillover Effects15.2. Basic Spillover and Individual/Direct Effects15.3. Assessing "Infectiousness" Effects15.4. Contagion versus Infectiousness Effects15.5. Tests for Specific Forms of Interference Using Causal Interactions15.6. Inferential Challenges with Many Individuals per Cluster15.7. Spillover Effects and Observational Data15.8. Spillover Effects and Social Networks15.9. Discussion16. Mediation and Interaction: Future and Context16.1. The Present State of Methods and Future Methodological Development16.2. Philosophical QuestionsAppendix. Technical Details and ProofsReferences

Editorial Reviews

"Mediation is about understanding pathways between a treatment and an outcome that lead to the outcome, i.e., mechanisms. Mechanisms are a central thing in science and statisticians have been providing new principled methods for studying these topics over especially the last 10 years.Especially in the social and behavioral sciences and in epidemiology there has been great interest in these methods, and the methodology the author wants to write about is the new stuff from the last 10 years [VanderWeele] is the key player in statistical literature these days. He's a goodcommunicator. Primary market: applied researchers doing mediation in epidemiology, social and behavioral sciences. Secondary market: applied statisticians teaching causal inference and/or working in the area. Yes, I think this might get some adoptions, and as the potential outcomes framework becomesmore established in disciplines such as epidemiology and psychology, more adoptions." --Michael Sobel Dept Sociology Columbia