Handbook of Matching and Weighting Adjustments for Causal Inference

An observational study infers the effects caused by a treatment, policy, program, intervention, or exposure in a context in which randomized experimentation is unethical or impractical. One task in an observational study is to adjust for visible pretreatment differences between the treated and control groups. Multivariate matching and weighting are two modern forms of adjustment. This handbook provides a comprehensive survey of the most recent methods of adjustment by matching, weighting, machine learning and their combinations. Three additional chapters introduce the steps from association to causation that follow after adjustments are complete. When used alone, matching and weighting do not use outcome information, so they are part of the design of an observational study. When used in conjunction with models for the outcome, matching and weighting may enhance the robustness of model-based adjustments. The book is for researchers in medicine, economics, public health, psychology, epidemiology, public program evaluation, and statistics who examine evidence of the effects on human beings of treatments, policies or exposures.

Part I: Conceptual issues Overview of methods for adjustment and applications in the social and behavioral sciences: The role of study designTing-Hsuan Chang and Elizabeth A. Stuart Propensity scorePaul R. Rosenbaum Generalization and TransportabilityElizabeth Tipton and Erin Hartman Part II: Matching Optimization techniques in multivariate matchingPaul R. Rosenbaum and José R. Zubizarreta Optimal Full matchingMark M. Fredrickson and Ben Hansen Fine balance and its variations in modern optimal matchingSamuel D. Pimentel Matching with instrumental variablesMike Baiocchi and Hyunseung Kang Covariate Adjustment in Regression Discontinuity DesignsMatias D. Cattaneo, Luke Keele, Rocío Titiunik Risk Set MatchingBo Lu and Robert A. Greevy, Jr. Matching with Multilevel DataSamuel D. Pimentel and Luke Keele Effect Modification in Observational StudiesKwonsang Lee and Jesse Y. Hsu Optimal Nonbipartite MatchingRobert A. Greevy, Jr. and Bo Lu Matching Methods for Large Observational StudiesRuoqi Yu Part III: Weighting Overlap WeightingFan Li Covariate Balancing Propensity ScoreKosuke Imai and Yang Ning Balancing Weights for Causal InferenceEric R. Cohn, Eli Ben-Michael, Avi Feller, and José R. Zubizarreta Assessing Principal Causal Effects Using Principal Score MethodsAlessandra Mattei, Laura Forastiere, Fabrizia Mealli Incremental Causal Effects: An Introduction and ReviewMatteo Bonvini, Alec McClean, Zach Branson and Edward H. Kennedy Weighting Estimators for Causal MediationDonna L. Coffman, Megan S. Schuler, Trang Q. Nguyen, and Daniel F. McCaffrey Part IV: Machine Learning Adjustments Machine Learning for Causal InferenceJennifer Hill, George Perrett and Vincent Dorie Treatment Heterogeneity with Survival OutcomesYizhe Xu, Nikolaos Ignatiadis, Erik Sverdrup, Scott Fleming, Stefan Wager, Nigam Shah Why Machine Learning Cannot Ignore Maximum Likelihood EstimationMark J. van der Laan and Sherri Rose Bayesian Propensity Score methods and Related Approaches for Confounding Adjustment Joseph Antonelli Part V: Beyond Adjustments How to Be a Good Critic of an Observational StudyDylan S. Small Sensitivity AnalysisC.B. Fogarty Evidence FactorsBikram Karmakar

José Zubizarreta, PhD, is an associate professor in the Department of Health Care Policy at Harvard Medical School and in the Department Biostatistics at Harvard University. He is a Fellow of the American Statistical Association, and is a recipient of the Kenneth Rothman Award, the William Cochran Award, and the Tom Ten Have Memorial Award. Elizabeth A. Stuart, Ph.D. is Bloomberg Professor of American Health in the Department of Mental Health, the Department of Biostatistics and the Department of Health Policy and Management at Johns Hopkins Bloomberg School of Public Health. She is a Fellow of the American Statistical Association, and she received the mid-career award from the Health Policy Statistics Section of the ASA, the Gertrude Cox Award for applied statistics, Harvard University’s Myrto Lefkopoulou Award for excellence in Biostatistics, and the Society for Epidemiologic Research Marshall Joffe Epidemiologic Methods award. Dylan Small, PhD is the Universal Furniture Professor in the Department of Statistics and Data Science of the Wharton School of the University of Pennsylvania. He is a Fellow of the American Statistical Association and an Institute of Mathematical Statistics Medallion Lecturer. Paul R. Rosenbaum is emeritus professor of Statistics and Data Science at the Wharton School of the University of Pennsylvania. From the Committee of Presidents of Statistical Societies, he received the R. A. Fisher Award and the George W. Snedecor Award. He is the author of several books, including Design of Observational Studies and Replication and Evidence Factors in Observational Studies.