Applied Longitudinal Data Analysis Modeling Change and Event Occurrence

by Singer, Judith D.; Willett, John B.

• ISBN13: 9780195152968
• ISBN10: 0195152964
• Format: Hardcover
• Copyright: 2003-03-27
• Publisher: Oxford University Press

Summary

Change is constant in everyday life. Infants crawl and then walk, children learn to read and write, teenagers mature in myriad ways, the elderly become frail and forgetful. Beyond these natural processes and events, external forces and interventions instigate and disrupt change: test scoresmay rise after a coaching course, drug abusers may remain abstinent after residential treatment. By charting changes over time and investigating whether and when events occur, researchers reveal the temporal rhythms of our lives. Applied Longitudinal Data Analysis is a much-needed professional bookfor empirical researchers and graduate students in the behavioral, social, and biomedical sciences. It offers the first accessible in-depth presentation of two of today's most popular statistical methods: multilevel models for individual change and hazard/survival models for event occurrence (inboth discrete- and continuous-time). Using clear, concise prose and real data sets from published studies, the authors take you step by step through complete analyses, from simple exploratory displays that reveal underlying patterns through sophisticated specifications of complex statistical models. Applied Longitudinal Data Analysis offers readers a private consultation session with internationally recognized experts and represents a unique contribution to the literature on quantitative empirical methods. Visit http://www.ats.ucla.edu/stat/examples/alda.htm for: BL Downloadable data sets BL Library of computer programs in SAS, SPSS, Stata, HLM, MLwiN, and more BL Additional material for data analysis

Table of Contents

PART I
A Framework for Investigating Change over Time	3	(13)
When Might You Study Change over Time?	4	(3)
Distinguishing Between Two Types of Questions about Change	7	(2)
Three Important Features of a Study of Change	9	(7)
Exploring Longitudinal Data on Change	16	(29)
Creating a Longitudinal Data Set	17	(6)
Descriptive Analysis of Individual Change over Time	23	(10)
Exploring Differences in Change across People	33	(8)
Improving the Precision and Reliability of OLS-Estimated Rates of Change: Lessons for Research Design	41	(4)
Introducing the Multilevel Model for Change	45	(30)
What Is the Purpose of the Multilevel Model for Change?	46	(3)
The Level-1 Submodel for Individual Change	49	(8)
The Level-2 Submodel for Systematic Interindividual Differences in Change	57	(6)
Fitting the Multilevel Model for Change to Data	63	(5)
Examining Estimated Fixed Effects	68	(4)
Examining Estimated Variance Components	72	(3)
Doing Data Analysis with the Multilevel Model for Change	75	(63)
Example: Changes in Adolescent Alcohol Use	76	(4)
The Composite Specification of the Multilevel Model for Change	80	(5)
Methods of Estimation, Revisited	85	(7)
First Steps: Fitting Two Unconditional Multilevel Models for Change	92	(12)
Practical Data Analytic: Strategies for Model Building	104	(12)
Comparing Models Using Deviance Statistics	116	(6)
Using Wald Statistics to Test Composite Hypotheses About Fixed Effects	122	(5)
Evaluating the Tenability of a Model's Assumptions	127	(5)
Model-Based (Empirical Bayes) Estimates of the Individual Growth Parameters	132	(6)
Treating TIME More Flexibly	138	(51)
Variably Spaced Measurement Occasions	139	(7)
Varying Numbers of Measurement Occasions	146	(13)
Time-Varying Predictors	159	(22)
Recentering the Effect of Time	181	(8)
Modeling Discontinuous and Nonlinear Change	189	(54)
Discontinuous Individual Change	190	(18)
Using Transformations to Model Nonlinear Individual Change	208	(5)
Representing Individual Change Using a Polynomial Function of Time	213	(10)
Truly Nonlinear Trajectories	223	(20)
Examining the Multilevel Model's Error Covariance Structure	243	(23)
The ``Standard'' Specification of the Multilevel Model for Change	243	(3)
Using the Composite Model to Understand Assumptions about the Error Covariance Matrix	246	(10)
Postulating an Alternative Error Covariance Structure	256	(10)
Modeling Change Using Covariance Structure Analysis	266	(39)
The General Covariance Structure Model	266	(14)
The Basics of Latent Growth Modeling	280	(15)
Cross-Domain Analysis of Change	295	(4)
Extensions of Latent Growth Modeling	299	(6)
PART II
A Framework for Investigating Event Occurrence	305	(20)
Should You Conduct a Survival Analysis? The ``Whether'' and ``When'' Test	306	(3)
Framing a Research Question About Event Occurrence	309	(6)
Censoring: How Complete Are the Data on Event Occurrence?	315	(10)
Describing Discrete-Time Event Occurrence Data	325	(32)
The Life Table	326	(4)
A Framework for Characterizing the Distribution of Discrete-Time Event Occurrence Data	330	(9)
Developing Intuition About Hazard Functions, Survivor Functions, and Median Lifetimes	339	(9)
Quantifying the Effects of Sampling Variation	348	(3)
A Simple and Useful Strategy for Constructing the Life Table	351	(6)
Fitting Basic Discrete-Time Hazard Models	357	(50)
Toward a Statistical Model for Discrete-Time Hazard	358	(11)
A Formal Representation of the Population Discrete-Time Hazard Model	369	(9)
Fitting a Discrete-Time Hazard Model to Data	378	(8)
Interpreting Parameter Estimates	386	(5)
Displaying Fitted Hazard and Survivor Functions	391	(6)
Comparing Models Using Deviance Statistics and Information Criteria	397	(5)
Statistical Inference Using Asymptotic Standard Errors	402	(5)
Extending the Discrete-Time Hazard Model	407	(61)
Alternative Specifications for the ``Main Effect of TIME''	408	(11)
Using the Complementary Log-Log Link to Specify a Discrete-Time Hazard Model	419	(7)
Time-Varying Predictors	426	(17)
The Linear Additivity Assumption: Uncovering Violations and Simple Solutions	443	(8)
The Proportionality Assumption: Uncovering Violations and Simple Solutions	451	(10)
The No Unobserved Heterogeneity Assumption: No Simple Solution	461	(2)
Residual Analysis	463	(5)
Describing Continuous-Time Event Occurrence Data	468	(35)
A Framework for Characterizing the Distribution of Continuous-Time Event Data	469	(6)
Grouped Methods for Estimating Continuous-Time Survivor and Hazard Functions	475	(8)
The Kaplan-Meier Method of Estimating the Continuous-Time Survivor Function	483	(5)
The Cumulative Hazard Function	488	(6)
Kernel-Smoothed Estimates of the Hazard Function	494	(3)
Developing an Intuition about Continuous-Time Survivor, Cumulative Hazard, and Kernel-Smoothed Hazard Functions	497	(6)
Fitting Cox Regression Models	503	(40)
Toward a Statistical Model for Continuous-Time Hazard	503	(13)
Fitting the Cox Regression Model to Data	516	(7)
Interpreting the Results of Fitting the Cox Regression Model to Data	523	(12)
Nonparametric Strategies for Displaying the Results of Model Fitting	535	(8)
Extending the Cox Regression Model	543	(64)
Time-Varying Predictors	544	(12)
Nonproportional Hazards Models via Stratification	556	(6)
Nonproportional Hazards Models via Interactions with Time	562	(8)
Regression Diagnostics	570	(16)
Competing Risks	586	(9)
Late Entry into the Risk Set	595	(12)
Notes	607	(6)
References	613	(14)
Index	627