Table of Contents
Acknowledgments ix
Introduction 1
What Is Causal Inference? 3
Do Not Confuse Correlation with Causality 6
Optimization Makes Everything Endogenous 8
Example: Identifying Price Elasticity of Demand 10
Conclusion 14
Probability and Regression Review 16
Directed Acyclic Graphs 96
Introduction to DAG Notation 97
Potential Outcomes Causal Model 119
Physical Randomization 123
Randomization Inference 148
Conclusion 174
Matching and Subclassification 175
Subclassification 175
Exact Matching 191
Approximate Matching 198
Regression Discontinuity 241
Huge Popularity of Regression Discontinuity 241
Estimation Using an RDD 252
Challenges to Identification 282
Replicating a Popular Design: The Close Election 289
Regression Kink Design 312
Conclusion 313
Instrumental Variables 315
History of Instrumental Variables: Father and Son 315
Intuition of Instrumental Variables 319
Homogeneous Treatment Effects 323
Parental Methamphetamine Abuse and Foster Care 329
The Problem of Weak Instruments 337
Heterogeneous Treatment Effects 346
Applications 352
Popular IV Designs 359
Conclusion 384
Panel Data 386
DAG Example 386
Estimation 388
Data Exercise: Survey of Adult Service Providers 396
Conclusion 405
Difference-in-Differences 406
John Snow's Cholera Hypothesis 406
Estimation 411
Inference 423
Providing Evidence for Parallel Trends Through Event Studies and Parallel Leads 425
The Importance of Placebos in DD 433
Twoway Fixed Effects with Differential Timing 461
Conclusion 509
Synthetic Control 511
Introducing the Comparative Case Study 511
Prison Construction and Black Male Incarceration 525
Conclusion 540
Bibliography 541
Permissions 555
Index 561