Table of contents for Counterfactuals and causal inference : methods and principles for social research / Stephen L. Morgan and Christopher Winship.

Bibliographic record and links to related information available from the Library of Congress catalog.

Note: Contents data are machine generated based on pre-publication provided by the publisher. Contents may have variations from the printed book or be incomplete or contain other coding.


Counter
Contents
I Counterfactual Causality and Empirical Research in
the Social Sciences 1
1 Introduction 1
1.1 The Counterfactual Model for Observational Data Analysis . . . 2
1.2 Causal Analysis and Observational Social Science . . . . . . . . . 4
1.2.1 Experimental Language in Observational Social Science . 4
1.2.2 ?The Age of Regression?. . . . . . . . . . . . . . . . . . . 8
1.3 Types of Examples Used Throughout the Book . . . . . . . . . . 11
1.3.1 Broad Examples from Sociology, Economics, and Political
Science . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.2 Narrow and Speci?c Examples . . . . . . . . . . . . . . . 15
1.4 Observational Data and Random-Sample Surveys . . . . . . . . . 19
1.5 Identi?cation and Statistical Inference . . . . . . . . . . . . . . . 20
1.6 Causal Graphs as an Introduction to the Remainder of the Book 22
2 The Counterfactual Model 29
2.1 Causal States and Potential Outcomes . . . . . . . . . . . . . . . 29
2.2 Treatment Groups and Observed Outcomes . . . . . . . . . . . . 32
2.3 The Average Treatment Eñect . . . . . . . . . . . . . . . . . . . 33
2.4 The Stable Unit Treatment Value Assumption . . . . . . . . . . . 35
2.5 Treatment Assignment and Observational Studies . . . . . . . . . 38
2.6 Average Causal Eñects and Naive Estimation . . . . . . . . . . . 40
2.6.1 Conditional Average Treatment Eñects . . . . . . . . . . 40
2.6.2 Naive Estimation of Average Treatment Eñects . . . . . . 42
2.6.3 Expected Bias of the Naive Estimator . . . . . . . . . . . 44
2.6.4 Estimating Causal Eñects Under Maintained Assumptions
About Potential Outcomes . . . . . . . . . . . . . . . . . 46
2.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.8 Appendix: Population and Data Generation Models . . . . . . . 49
2.9 Appendix: Extension of the Framework To Many-Valued Treat-
ments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
II Estimating Causal Eñects by Conditioning 57
3 Causal Graphs, Identi?cation, and Models of Causal Exposure 59
3.1 Causal Graphs and Conditioning as Back-Door Identi?cation . . 59
3.1.1 Basic Elements . . . . . . . . . . . . . . . . . . . . . . . . 60
3.1.2 Conditioning on Observable Variables . . . . . . . . . . . 62
3.1.3 Point-Identi?cation by Conditioning on Variables that Sat-
isfy the Back-Door Criterion . . . . . . . . . . . . . . . . 64
3.2 Models of Causal Exposure in the Counterfactual Tradition . . . 69
3.2.1 Treatment Assignment Modeling in Statistics . . . . . . . 70
3.2.2 Treatment Selection Modeling in Econometrics . . . . . . 73
3.3 Conditioning to Balance Versus Conditioning to Adjust . . . . . 76
3.4 Point-Identi?cation of Conditional Average Treatment Eñects via
Conditioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4 Matching Estimators of Causal Eñects 83
4.1 Origins of and Motivations for Matching . . . . . . . . . . . . . . 84
4.2 Matching as Conditioning via Strati?cation . . . . . . . . . . . . 86
4.2.1 Estimating Causal Eñects by Strati?cation . . . . . . . . 86
4.2.2 Overlap Conditions for Stratifying Variables . . . . . . . . 91
4.3 Matching as Weighting . . . . . . . . . . . . . . . . . . . . . . . . 94
4.3.1 The Utility of Known Propensity Scores . . . . . . . . . . 94
4.3.2 Weighting with Propensity Scores to Address Sparseness . 95
4.4 Matching as a Data Analysis Algorithm . . . . . . . . . . . . . . 101
4.4.1 Basic Variants of Matching Algorithms . . . . . . . . . . . 103
4.4.2 Which of These Basic Matching Algorithms Works Best? 105
4.4.3 Matching Algorithms That Seek Optimal Balance . . . . 109
4.5 Matching when Treatment Assignment is Nonignorable . . . . . . 111
4.6 Remaining Practical Issues in Matching Analysis . . . . . . . . . 112
4.6.1 Assessing the Region of Common Support . . . . . . . . . 112
4.6.2 The Expected Variance of Matching Estimates . . . . . . 113
4.6.3 Matching Estimators for Many-Valued Causes . . . . . . . 115
4.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
5 Regression Estimators of Causal Eñects 119
5.1 Regression as a Descriptive Tool . . . . . . . . . . . . . . . . . . 119
5.2 Regression Adjustment as a Strategy to Estimate Causal Eñects 125
5.2.1 Regression Models and Omitted Variable Bias . . . . . . . 126
5.2.2 Potential Outcomes and Omitted Variable Bias . . . . . . 128
5.2.3 Regression as Adjustment for Otherwise Omitted Variables132
5.3 The Connections Between Regression and Matching . . . . . . . 139
5.3.1 Regression as Conditional-Variance-Weighted Matching . 139
5.3.2 Regression as an Implementation of a Perfect Strati?cation148
5.3.3 Matching as Weighted Regression . . . . . . . . . . . . . . 149
5.3.4 Regression as Supplemental Adjustment when Matching . 153
5.4 Extensions and Other Perspectives . . . . . . . . . . . . . . . . . 156
5.4.1 Regression Estimators for Many-Valued Causes . . . . . . 156
5.4.2 Data Mining and the Challenge of Regression Speci?cation 158
5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
III Estimating Causal Eñects When Simple Condition-
ing is Ineñective 165
6 Identi?cation in the Absence of a Complete Model of Causal
Exposure 167
6.1 Non-ignorability and Selection on the Unobservables Revisited . 167
6.2 Sensitivity Analysis for Provisional Causal Eñect Estimates . . . 168
6.3 Partial Identi?cation with Minimal Assumptions . . . . . . . . . 170
6.3.1 No-Assumptions Bounds . . . . . . . . . . . . . . . . . . . 171
6.3.2 Bounds Under Additional Weak Assumptions . . . . . . . 174
6.4 Additional Strategies for the Point-Identi?cation of Causal Eñects 177
6.4.1 Conditioning on a ?Pre-Test? . . . . . . . . . . . . . . . . 177
6.4.2 Instrumental Variables and Naturally Occurring Variation 179
6.4.3 Mechanisms and the Front-Door Criterion . . . . . . . . . 180
6.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
6.6 Appendix: Latent Variable Selection Bias Models . . . . . . . . 182
7 Natural Experiments and Instrumental Variables 185
7.1 Causal Eñect Estimation with a Binary IV . . . . . . . . . . . . 185
7.2 Traditional IV Estimators . . . . . . . . . . . . . . . . . . . . . . 190
7.3 Recognized Pitfalls of Traditional IV Estimation . . . . . . . . . 195
7.4 Instrumental Variable Estimators of Average Causal Eñects . . . 198
7.4.1 IV Estimation as LATE Estimation . . . . . . . . . . . . 198
7.4.2 Implications of the LATE Perspective for Traditional IV
Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . 208
7.5 Two Additional Perspectives on the Identi?cation of Causal Ef-
fects via IVs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
7.5.1 Local IVs and Marginal Treatment Eñects . . . . . . . . . 211
7.5.2 Monotone IVs and Analyses of Bounds . . . . . . . . . . . 212
7.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
8 Mechanisms and Causal Explanation 217
8.1 The Dangers of Insuó ciently Deep Explanations . . . . . . . . . 218
8.2 Explanation and Identi?cation of Causal Eñects via Mechanisms 222
8.3 The Appeal for Generative Mechanisms . . . . . . . . . . . . . . 227
8.4 The Pursuit of Explanation via Mechanisms that Bottom Out . . 234
8.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
9 Repeated Observations and the Estimation of Causal Eñects 241
9.1 Interrupted Time Series Models . . . . . . . . . . . . . . . . . . . 242
9.2 Regression Discontinuity Designs . . . . . . . . . . . . . . . . . . 246
9.3 Panel Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
9.3.1 Traditional Adjustment Strategies . . . . . . . . . . . . . 249
9.3.2 Model-Based Approaches . . . . . . . . . . . . . . . . . . 258
9.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
IV Conclusions 271
10 Counterfactual Causality and Future Empirical Research in the
Social Sciences 273
10.1 Objections to Features of the Counterfactual Model . . . . . . . 274
10.2 Modes of Causal Inquiry in the Social Sciences . . . . . . . . . . 281

Library of Congress Subject Headings for this publication:

Social sciences -- Research.
Social sciences -- Methodology.
Causation.