Rapid technological advances in devices used for data collection have led to the emergence of a new class of longitudinal data: intensive longitudinal data (ILD). Behavioral scientific studies now frequently utilize handheld computers, beepers, web interfaces, and other technological tools for collecting many more data points over time than previously possible. Other protocols, such as those used in fMRI and monitoring of public safety, also produce ILD, hence the statistical models in this volume are applicable to a range of data. The volume features state-of-the-art statistical modeling strategies developed by leading statisticians and methodologists working on ILD in conjunction with behavioral scientists. Chapters present applications from across the behavioral and health sciences, including coverage of substantive topics such as stress, smoking cessation, alcohol use, traffic patterns, educational performance and intimacy. Models for Intensive Longitudinal Data (MILD) is designed for those who want to learn about advanced statistical models for intensive longitudinal data and for those with an interest in selecting and applying a given model. The chapters highlight issues of general concern in modeling these kinds of data, such as a focus on regulatory systems, issues of curve registration, variable frequency and spacing of measurements, complex multivariate patterns of change, and multiple independent series. The extraordinary breadth of coverage makes this an indispensable reference for principal investigators designing new studies that will introduce ILD, applied statisticians working on related models, and methodologists, graduate students, and applied analysts working in a range of fields. A companion Web site at www.oup.com/us/MILD contains program examples and documentation.
Author(s): Theodore A. Walls, Joseph L. Schafer
Publisher: Oxford University Press, USA
Year: 2006
Language: English
Pages: 311
Contents......Page 6
Contributors......Page 10
Introduction: Intensive Longitudinal Data......Page 12
1.1 Behavioral Scientific Motivations for Collecting Intensive Longitudinal Data......Page 26
1.2 Overview of Multilevel Models......Page 28
1.3 Applying Multilevel Modeling to Intensive Longitudinal Data......Page 36
1.4 Application: Control and Choice in Indian Schoolchildren......Page 50
1.5 Summary......Page 56
2 Marginal Modeling of Intensive Longitudinal Data by Generalized Estimating Equations......Page 61
2.1 What Is GEE Regression?......Page 62
2.2 Practical Considerations in the Application of GEE......Page 71
2.3 Application: Reanalysis of the Control and Choice Data Using GEE......Page 78
3 A Local Linear Estimation Procedure for Functional Multilevel Modeling......Page 86
3.1 The Model......Page 88
3.2 Practical Considerations......Page 94
3.3 Application: Smoking Cessation Study......Page 95
3.4 Discussion......Page 103
4 Application of Item Response Theory Models for Intensive Longitudinal Data......Page 107
4.1 IRT Model......Page 108
4.2 Estimation......Page 115
4.3 Application: Adolescent Smoking Study......Page 117
4.4 Discussion......Page 126
5.1 Periodic and Nonperiodic Trends......Page 132
5.2 The Model......Page 135
5.3 Application: Personality Data......Page 143
5.4 Discussion......Page 145
6 Multilevel Autoregressive Modeling of Interindividual Differences in the Stability of a Process......Page 147
6.1 Defining Stability as Regularity in a Time Series......Page 148
6.2 Multilevel Models......Page 149
6.3 A Multilevel AR(1) Model......Page 154
6.4 Application: Daily Alcohol Use......Page 156
6.5 Estimating This Model in SAS PROC MIXED......Page 157
6.6 Predicting the Individual AR(1) Coefficients......Page 161
6.7 Discussion......Page 166
7 The State-Space Approach to Modeling Dynamic Processes......Page 171
7.1 Gaussian State-Space Models......Page 172
7.2 Some Special Cases of State-Space Models......Page 175
7.3 Parameter Estimation......Page 179
7.4 Application 1: Connectivity Analysis with fMRI Data......Page 182
7.5 Application 2: Testing the Induced Demand Hypothesis from Matched Traffic Profiles......Page 188
7.6 Conclusions......Page 193
8 The Control of Behavioral Input/Output Systems......Page 199
8.1 A Typical Input/Output System......Page 200
8.2 Modeling System Dynamics......Page 202
8.3 Controller Strategies to Meet an Output Target......Page 206
8.4 Fitting Dynamic Models to Intensive Longitudinal Data......Page 212
9.1 Self-Regulation and Intrinsic Dynamics......Page 218
9.2 Coupled Regulation and Coupled Dynamics......Page 223
9.3 Time-Delay Embedding......Page 226
9.4 Accounting for Individual Differences in Dynamics......Page 228
9.5 Application: Daily Intimacy and Disclosure in Married Couples......Page 229
9.6 Discussion......Page 238
10 Point Process Models for Event History Data: Applications in Behavioral Science......Page 242
10.1 Ecological Momentary Assessment of Smoking......Page 245
10.2 Point Process Models......Page 247
10.3 Application: An EMA Study of Smoking Data......Page 251
10.4 Discussion of Results......Page 265
10.5 Multivariate Point Patterns......Page 268
11 Emerging Technologies and Next-Generation Intensive Longitudinal Data Collection......Page 277
11.1 Intensive Data Collection Systems......Page 279
11.2 Statistical Issues for Intensive Longitudinal Measurement......Page 288
11.3 Summary......Page 297
B......Page 302
D......Page 303
F......Page 304
I......Page 305
M......Page 306
P......Page 307
R......Page 308
S......Page 309
T......Page 310
W......Page 311
