Causal mediation analysis / Geoffrey T. Wodtke, Xiang Zhou.

Format:

Book

Author/Creator:

Wodtke, Geoffrey T., author.

Zhou, Xiang, author.

Series:

Analytical methods for social research.

Analytical methods for social research

Language:

English

Subjects (All):

Social sciences--Statistical methods.

Social sciences.

Physical Description:

1 online resource (xiv, 435 pages) : illustrations (black and white), digital, PDF file(s).

Edition:

1st ed.

Place of Publication:

Cambridge : Cambridge University Press, 2026.

Summary:

Geoffrey T. Wodtke and Xiang Zhou's 'Causal Mediation Analysis' offer a comprehensive yet accessible guide to causal mediation analysis for social scientists. They explore why an exposure affects an outcome by quantifying the processes and mechanisms through which a causal effect operates. Covering everything from traditional methods through machine learning techniques and experimental designs for analysing mediation, the authors make these methods broadly accessible through clear explanations, practical examples, and the inclusion of extensive Stata and R code.

Contents:

Cover

Half-title

Series information

Title page

Copyright information

Dedication

Contents

Acknowledgments

1 Introduction

1.1 Mediation in the Social Sciences

1.1.1 The Effects of Education on Mental Health

1.1.2 Media Discourse, Public Opinion, and Immigration

1.1.3 Workforce Development Programs and Employment

1.2 The Promise of Causal Mediation Analysis

1.3 A Brief History

1.4 Our Distinctive Approach

1.5 An Outline of the Book

1.6 Statistical Software

2 Foundations of Causal Inference

2.1 Measures of Association

2.2 Measures of Causation

2.2.1 Potential Outcomes

2.2.2 Individual Causal Effects

2.2.3 Average Causal Effects

2.2.4 The Fundamental Problem of Causal Inference

2.3 Resolving the Fundamental Problem

2.3.1 Identifying Individual Causal Effects

2.3.2 Identifying Average Causal Effects

2.4 Directed Acyclic Graphs

2.4.1 Elements of a DAG

2.4.2 DAGs as Nonparametric Causal Models

2.4.3 DAGs, Interventions, and Potential Outcomes

2.4.4 Sources of Statistical Association in DAGs

2.4.5 Identification Analysis with DAGs

2.5 Direct and Indirect Causation

2.5.1 A Simple Model of Mediation

2.5.2 Nested and Cross-World Potential Outcomes

2.5.3 The Fundamental Problem of Causal Mediation

3 Mediation Analysis with Baseline Confounding

3.1 Graphical Mediation Models

3.2 Causal Estimands

3.2.1 The Average Total Effect

3.2.2 Natural Direct and Indirect Effects

3.2.3 The Controlled Direct Effect

3.3 Nonparametric Identification

3.3.1 Nonparametric Identification of Average Total Effects

3.3.2 Nonparametric Identification of Controlled Direct Effects

3.3.3 Nonparametric Identification of Natural Directand Indirect Effects

3.4 Nonparametric Estimation

3.5 Parametric Estimation.

3.5.1 Estimation with Linear Models

3.5.2 Estimation via Simulation

3.5.3 Estimation with Inverse Probability Weights

3.6 Statistical Inference

3.7 Sensitivity Analysis

3.7.1 Nonparametric Bias Formulas

3.7.2 Bias from Exposure-Outcome Confounding

3.7.3 Bias from Mediator-Outcome Confounding

3.7.4 Bias from Exposure-Mediator Confounding

3.7.5 Bias-Adjusted Effect Estimates

3.8 The Effect of Job Training on Employment

3.9 Summary

4 Mediation Analysis with Exposure-Induced Confounding

4.1 Graphical Models with Multiple Mediators

4.2 Limitations of the Natural Effects Decomposition

4.3 Interventional Direct and Indirect Effects

4.4 Nonparametric Identification

4.4.1 Nonparametric Identification of Controlled Direct Effects

4.4.2 Nonparametric Identification of Interventional Direct and Indirect Effects

4.5 Nonparametric Estimation

4.6 Parametric Estimation

4.6.1 Estimation Using Linear Models: Regression-with-Residuals

4.6.2 Estimation via Simulation

4.6.3 Estimation with Inverse Probability Weights

4.7 Statistical Inference

4.8 Sensitivity Analysis

4.8.1 Nonparametric Bias Formulas

4.8.2 Bias from Exposure-Outcome Confounding

4.8.3 Bias from Mediator-Outcome Confounding

4.8.4 Bias from Exposure-Mediator Confounding

4.8.5 Bias-Adjusted Effect Estimates

4.9 The Effect of Plow Use on Female Political Participation

4.10 Summary

5 Mediation Analysis with Multiple Mediators

5.1 The One-Mediator-at-a-Time Approach

5.2 The Multiple-Mediators-as-a-Whole Approach

5.3 Effect Decomposition with Multiple Mediators

5.3.1 Path-Specific Effects

5.3.2 Connections with Natural Direct and Indirect Effects

5.4 Nonparametric Identification

5.5 Nonparametric Estimation

5.6 Parametric Estimation

5.6.1 Estimation with Linear Models.

5.6.2 Estimation with Inverse Probability Weights

5.6.3 Estimation with Regression Imputation

5.7 Generalization to K(≥2) Causally Ordered Mediators

5.8 Sensitivity Analysis

5.8.1 Nonparametric Bias Formulas

5.8.2 Bias from Exposure-Outcome Confounding

5.8.3 Bias from Mediator-Outcome Confounding

5.8.4 Bias from Exposure-Mediator Confounding

5.8.5 Bias-Adjusted Effect Estimates

5.9 The Effect of Media Framing on Immigration Attitudes

5.10 Summary

6 Mediation Analysis with Robust Estimation Methods

6.1 Robust Estimation of Average Total Effects

6.1.1 Limitations of Parametric Estimators

6.1.2 Doubly Robust Estimation

6.1.3 Debiased Machine Learning

6.1.4 Statistical Inference

6.1.5 The Effect of College Attendance on Depression

6.2 Robust Estimation of Natural Direct and Indirect Effects

6.3 Robust Estimation of Interventional Direct and Indirect Effects

6.4 Robust Estimation of Path-Specific Effects

6.5 The Multigenerational Effects of Political Violence on Attitudes

6.6 Summary

7 Mediation Analysis with (Quasi-)Experimental Methods

7.1 Conventional Experiments and Their Limitations

7.2 Experimental Designs for Analyzing Causal Mediation

7.2.1 Joint and Sequential Randomization Designs

7.2.2 A Parallel Randomization Design

7.2.3 A Multi-Arm Randomization Design

7.2.4 Limitations

7.3 Quasi-experimental Designs for Analyzing CausalMediation

7.3.1 Instrumental Variable (IV) Analyses with an Instrumented Exposure

7.3.2 IV Analyses with an Instrumented Mediator

7.3.3 Difference-in-Difference (DiD) Models for Total Effects

7.3.4 DiD Models for Direct and Indirect Effects

7.4 Summary

8 The Future of Causal Mediation Analysis

8.1 Alternative Estimands

8.2 Longitudinal Analyses

8.3 Complex Sample Designs

8.4 Measurement Error.

8.5 Robust Estimation Methods

8.6 Experimental and Quasi-experimental Methods

8.7 Conclusion

Appendix A Nonparametric Identification of the Average Total Effect

Appendix B Nonparametric Identification of the Controlled Direct Effect

Appendix C Nonparametric Identification of the Natural Direct Effect

Appendix D Nonparametric Identification of the Natural Indirect Effect

Appendix E Natural Effects under Linear and Additive Models

Appendix F Ratio of Mediator Probability Weighting

Appendix G Nonparametric Identification of CDEs with Exposure-Induced Confounding

Appendix H Nonparametric Identification of the Interventional Direct Effect

Appendix I Nonparametric Identification of the Interventional Indirect Effect

Appendix J Bias Formulas for PSEs with K≥2 Causally Ordered Mediators

References

Index.

Notes:

Includes bibliographical references and index.

Description based on online resource; title from PDF title page (viewed on March 24, 2026).

Description based on publisher supplied metadata and other sources.

ISBN:

1-009-22218-X

1-009-22221-X

1-009-22219-8

OCLC:

1574808302