A sound statistical account of the principles and methods for the design and analysis of observational studies. Readers are assumed to have a working knowledge of basic probability and statistics, but otherwise the account is reasonably self- contained. Throughout there are extended discussions of actual observational studies to illustrate the ideas discussed, drawn from topics as diverse as smoking and lung cancer, lead in children, nuclear weapons testing, and placement programs for students. As a result, many researchers will find this an invaluable companion in their work.

An Observational study is an empiric investigation of the effects caused by a treatment, policy , or intervention in which it is not possible to assign subjects at random to treatment or control, as would be done in a controlled experiment. Observational studies are common in most fields that study the effects of treatments on people. The second edition of ¿Observational Studies¿ is about 50 percent longer than the first edition, with many new examples and methods. There are new chapters on nonadditive models for treatment effects (Chapter 5) and planning observational studies (Chapter 11) and Chapter 9, on coherence, has been extensively rewritten. Paul R. Rosenbaum is Robert G. Putzel Professor, Department of Statistics, The Wharton School of the University of Pennsylvania. He is a fellow of the American Statistical Association.

1 Observational Studies 1 2 Randomized Experiments 19 3 Overt Bias in Observational Studies 71 4 Sensitivity to Hidden Bias 105 5 Models for Treatment Effects 171 6 Known Effects 205 7 Multiple Reference Groups in Case-Referent Studies 231 8 Multiple Control Groups 253 9 Coherence and Focused Hypotheses 277 10 Constructing Matched Sets and Strata 295 11 Planning an Observational Study 333 12 Some Strategic Issues 367 Index 370