Interrupted Time Series Analysis develops a comprehensive set of models and methods for drawing causal inferences from time series. It provides example analyses of social, behavioral, and biomedical time series to illustrate a general strategy for building AutoRegressive Integrated Moving Average (ARIMA) impact models. Additionally, the book supplements the classic Box-Jenkins-Tiao model-building strategy with recent auxiliary tests for transformation, differencing, and model selection. Not only does the text discuss new developments, including the prospects for widespread adoption of Bayesian hypothesis testing and synthetic control group designs, but it makes optimal use of graphical illustrations in its examples. With forty completed example analyses that demonstrate the implications of model properties, Interrupted Time Series Analysis will be a key inter-disciplinary text in classrooms, workshops, and short-courses for researchers familiar with time series data or cross-sectional regression analysis but limited background in the structure of time series processes and experiments.
|
|
|
xi | |
|
|
|
xv | |
| Acknowledgments |
|
xvii | |
|
|
|
1 | (5) |
|
|
|
6 | (3) |
|
1.2 A Short Note on Software |
|
|
9 | (2) |
|
|
|
11 | (3) |
|
2.1 White Noise Processes |
|
|
14 | (2) |
|
2.2 AR1 and Ma1 Processes |
|
|
16 | (6) |
|
|
|
22 | (2) |
|
2.4 Higher-Order and Mixed Processes |
|
|
24 | (5) |
|
2.5 Invertibility and Stationarity |
|
|
29 | (7) |
|
|
|
36 | (4) |
|
2.7 Stationarity Revisited |
|
|
40 | (3) |
|
|
|
43 | (3) |
|
|
|
46 | (2) |
|
3 The Noise Component: N(at) |
|
|
48 | (5) |
|
|
|
53 | (3) |
|
3.2 The Normality Assumption: A Digression |
|
|
56 | (3) |
|
3.3 Ar1 and Mai Time Series |
|
|
59 | (8) |
|
|
|
59 | (4) |
|
|
|
63 | (2) |
|
3.3.3 Pediatric Trauma Admissions |
|
|
65 | (2) |
|
3.4 Higher-Orderarma Processes |
|
|
67 | (10) |
|
3.4.1 Beveridge's Wheat Price Time Series |
|
|
67 | (7) |
|
3.4.2 Zurich Sunspot Numbers |
|
|
74 | (3) |
|
|
|
77 | (7) |
|
3.5.1 Kroeber's Skirt-Width Time Series |
|
|
78 | (3) |
|
3.5.2 Annual U.S. Tuberculosis Cases |
|
|
81 | (3) |
|
|
|
84 | (10) |
|
3.6.1 Anchorage Monthly Precipitation |
|
|
84 | (3) |
|
3.6.2 Monthly Atmospheric CO2 |
|
|
87 | (3) |
|
3.6.3 Australian Traffic Fatalities |
|
|
90 | (4) |
|
|
|
94 | (4) |
|
4 The Intervention Component: X(It) |
|
|
98 | (2) |
|
4.1 Abrupt, Permanent Impacts |
|
|
100 | (1) |
|
4.1.1 Rest Breaks and Productivity |
|
|
101 | (2) |
|
4.1.2 Prophylactic Vancomycin and Surgical Infection |
|
|
103 | (2) |
|
4.1.3 New Hampshire Medicaid Prescriptions |
|
|
105 | (3) |
|
4.1.4 Methadone Maintenance Treatments |
|
|
108 | (4) |
|
4.2 Gradually Accruing Impacts |
|
|
112 | (10) |
|
4.2.1 Australian Traffic Fatalities |
|
|
112 | (4) |
|
4.2.2 British Traffic Fatalities |
|
|
116 | (4) |
|
4.2.3 "Talking Out" Incidents |
|
|
120 | (2) |
|
|
|
122 | (6) |
|
4.3.1 Self-Injurious Behavior |
|
|
125 | (3) |
|
|
|
128 | (4) |
|
4.4.1 Decriminalization of Public Drunkenness |
|
|
129 | (3) |
|
|
|
132 | (4) |
|
5 Auxiliary Modeling Procedures |
|
|
136 | (1) |
|
|
|
137 | (7) |
|
|
|
144 | (6) |
|
5.3 Co-Integrated Time Series |
|
|
150 | (2) |
|
|
|
152 | (2) |
|
|
|
154 | (1) |
|
6.1 Bayesian Hypothesis Testing |
|
|
155 | (6) |
|
6.2 Synthetic Control Designs |
|
|
161 | (8) |
|
|
|
169 | (2) |
| References |
|
171 | (6) |
| Index |
|
177 | |
David McDowall is Distinguished Teaching Professor at the University at Albany, State University of New York. He serves on the faculty of Albany's School of Criminal Justice, where he also co-directs the Violence Research Group. His research interests involve the social distribution of criminal violence, including trends and other temporal features in crime rates.
Richard McCleary is a professor at the University of California, Irvine. In addition to faculty appointments in Criminology, Law and Society, Environmental Health Sciences, and Planning, Policy and Design, he directs the Irvine Simulation Modeling Laboratory. His research interests include population forecast models, time series models, and survival models.
Bradley J. Bartos is a doctoral candidate in the Department of Criminology, Law and Society at the University of California, Irvine. Through his work with the Irvine Simulation Modeling Laboratory, he has developed discrete-event population projection models for
various criminal-justice and corrections systems in California. His research interests include mass incarceration, policy evaluation, time series models, and synthetic control group designs.
