Atjaunināt sīkdatņu piekrišanu

E-grāmata: Statistical Methods in Psychiatry and Related Fields: Longitudinal, Clustered, and Other Repeated Measures Data

5.00/5 (1 ratings by Goodreads)
Citas grāmatas par šo tēmu:
  • Formāts - PDF+DRM
  • Cena: 60,10 €*
  • * ši ir gala cena, t.i., netiek piemērotas nekādas papildus atlaides
  • Ielikt grozā
  • Pievienot vēlmju sarakstam
  • Šī e-grāmata paredzēta tikai personīgai lietošanai. E-grāmatas nav iespējams atgriezt un nauda par iegādātajām e-grāmatām netiek atmaksāta.
Statistical Methods in Psychiatry and Related Fields: Longitudinal, Clustered, and Other Repeated Measures DataPalielināt
Citas grāmatas par šo tēmu:

DRM restrictions

  • Kopēšana (kopēt/ievietot):

    nav atļauts

  • Drukāšana:

    nav atļauts

  • Lietošana:

    Digitālo tiesību pārvaldība (Digital Rights Management (DRM))
    Izdevējs ir piegādājis šo grāmatu šifrētā veidā, kas nozīmē, ka jums ir jāinstalē bezmaksas programmatūra, lai to atbloķētu un lasītu. Lai lasītu šo e-grāmatu, jums ir jāizveido Adobe ID. Vairāk informācijas šeit. E-grāmatu var lasīt un lejupielādēt līdz 6 ierīcēm (vienam lietotājam ar vienu un to pašu Adobe ID).

    Nepieciešamā programmatūra
    Lai lasītu šo e-grāmatu mobilajā ierīcē (tālrunī vai planšetdatorā), jums būs jāinstalē šī bezmaksas lietotne: PocketBook Reader (iOS / Android)

    Lai lejupielādētu un lasītu šo e-grāmatu datorā vai Mac datorā, jums ir nepieciešamid Adobe Digital Editions (šī ir bezmaksas lietotne, kas īpaši izstrādāta e-grāmatām. Tā nav tas pats, kas Adobe Reader, kas, iespējams, jau ir jūsu datorā.)

    Jūs nevarat lasīt šo e-grāmatu, izmantojot Amazon Kindle.

Data collected in psychiatry and related fields are complex because outcomes are rarely directly observed, there are multiple correlated repeated measures within individuals, there is natural heterogeneity in treatment responses and in other characteristics in the populations. Simple statistical methods do not work well with such data. More advanced statistical methods capture the data complexity better, but are difficult to apply appropriately and correctly by investigators who do not have advanced training in statistics.

This book presents, at a non-technical level, several approaches for the analysis of correlated data: mixed models for continuous and categorical outcomes, nonparametric methods for repeated measures and growth mixture models for heterogeneous trajectories over time. Separate chapters are devoted to techniques for multiple comparison correction, analysis in the presence of missing data, adjustment for covariates, assessment of mediator and moderator effects, study design and sample size considerations. The focus is on the assumptions of each method, applicability and interpretation rather than on technical details.

Features





Provides an overview of intermediate to advanced statistical methods applied to psychiatry.





Takes a non-technical approach with mathematical details kept to a minimum.





Includes lots of detailed examples from published studies in psychiatry and related fields.





Software programs, data sets and output are available on a supplementary website.

The intended audience are applied researchers with minimal knowledge of statistics, although the book could also benefit collaborating statisticians. The book, together with the online materials, is a valuable resource aimed at promoting the use of appropriate statistical methods for the analysis of repeated measures data.

Ralitza Gueorguieva is a Senior Research Scientist at the Department of Biostatistics, Yale School of Public Health. She has more than 20 years experience in statistical methodology development and collaborations with psychiatrists and other researchers, and is the author of over 130 peer-reviewed publications.

Recenzijas

"This is a comprehensive text describing a wide-range of statistical techniques, from basic to advanced, applicable to data generated from the field of Mental Health research. The field of Psychiatry is one of the very core subjects of Mental Health, along with Clinical Psychology and Psychiatric SocialWork. Hence, it is in fact an amalgamation of many cross linking fields of the social sciences. Therefore, coming out with a book that addresses various dimensions of science is a remarkable achievement. [ T]he book has three unique selling points addressing the issues of the different natures of data in psychiatry, post hoc analyses and adjustments for multiple comparisons, and study design and sample size determination. I very strongly advocate that researchers/academics have this book with them or in their library." Chandra Bhushan Tripathi, in ISCB News, December 2018

"This book will reach a wide audience since it gives a non-technical and comprehensible introduction for non-experts to a complicated topic, with a series of worked-through examples, while at the same time it provides the applied statistician with thorough guidance to the analysis of longitudinal data, in the traditional normal distribution setting as well as for non-normal distributions (binary, Poisson etc.). It also contains insightful discussions on more advanced topics, with good references for further reading. The book stands out in the discussion on multiple testing, with good advice in a jungle of possibilities and in the handling of missing values (a comprehensible explanation of the problems and pitfalls, together with a sober guidance to avoiding such pitfalls in various circumstances). Also, the section on causality is probably the best I ever came across.The guidance and summary sections at the end of each chapter will serve as a reminder on the important hints for this particular topic. An extremely nice addition to the book is the accompanying home pages with SAS programs for the various analyses performed. The combination of non-technical verbal explanations of the models, combined with the precise analysis provided by the code, makes the book very well suited for teaching as well as for self study." Lene Theil Skovgaard, Section of Biostatistics, University of Copenhagen

"This is an extremely useful book, especially for data analysts who want the "nuts and bolts" of conducting longitudinal and clustered data analysis. The logic and necessary steps are carefully laid out and explained in great detail. Also, there are several chapters that include material not often included in books on longitudinal data analysis, for example the treatment of trajectory and growth mixture models, mediator and moderator effects, study design and sample size considerations. The examples focus on psychiatric studies, so this book will be of most interest to researchers in psychiatry, psychology, and mental health. However, those in other fields can also gain a great deal by considering this book, which is exemplary in its thoroughness and clarity. Highly recommended for anyone wanting to learn about statistical methods for longitudinal and clustered data, and even the experts can learn a great deal by the careful treatment of these topics that Dr. Gueorguieva has provided." Donald Hedeker, University of Chicago

Preface xv
1 Introduction
1(36)
1.1 Aspects of Repeated Measures Data
2(2)
1.1.1 Average (Mean) Response
2(1)
1.1.2 Variance and Correlation
3(1)
1.2 Types of Studies with Repeatedly Measured Outcomes
4(4)
1.3 Advantages of Collecting and Analyzing Repeatedly Measured Data
8(1)
1.4 Challenges in the Analysis of Correlated Data
9(1)
1.5 Data Sets
10(13)
1.5.1 Augmentation Treatment for Depression
10(2)
1.5.2 Sequenced Treatment Alternatives to Relieve Depression (STAR*D)
12(1)
1.5.3 Combined Pharmacotherapies and Behavioral Interventions for Alcohol Dependence (COMBINE) Study
13(4)
1.5.4 The Health and Retirement Study
17(2)
1.5.5 Serotonin Transport Study in Mother-Infant Pairs
19(1)
1.5.6 Meta-Analysis of Clinical Trials in Schizophrenia
19(2)
1.5.7 Human Laboratory Study of Menthol's Effects on Nicotine Reinforcement in Smokers
21(1)
1.5.8 Functional Magnetic Resonance Imaging (fMRI) Study of Working Memory in Schizophrenia
22(1)
1.5.9 Association between Unemployment and Depression
23(1)
1.6 Historical Overview of Approaches for the Analysis of Repeated Measures
23(3)
1.7 Basic Statistical Terminology and Notation
26(9)
1.7.1 Response
26(2)
1.7.2 Predictors
28(1)
1.7.3 Linear Model
28(1)
1.7.4 Average (Mean) Response
29(1)
1.7.5 Residual Variability
30(2)
1.7.6 Estimation
32(1)
1.7.7 Statistical Inference
33(1)
1.7.8 Checking Model Assumptions
34(1)
1.7.9 Model Fit and Model Selection
35(1)
1.8 Summary
35(2)
2 Traditional Methods for Analysis of Longitudinal and Clustered Data
37(22)
2.1 Endpoint Analysis and Analysis of Summary Measures
38(11)
2.1.1 Change from Baseline to Endpoint
38(2)
2.1.2 Group Comparison in Endpoint Analysis
40(3)
2.1.3 Multiple Group Comparisons in Endpoint Analysis
43(4)
2.1.4 Controlling for Baseline or Other Covariates
47(2)
2.1.5 Summary
49(1)
2.2 Analysis of Summary Measures
49(3)
2.2.1 Mean Response
49(1)
2.2.2 Slope over Time
50(1)
2.2.3 Peak Response and Area under the Curve
51(1)
2.2.4 Summary
52(1)
2.3 Univariate rANOVA Models
52(3)
2.4 Multivariate rMANOVA Models
55(2)
2.5 Summary
57(2)
3 Linear Mixed Models for Longitudinal and Clustered Data
59(44)
3.1 Modeling the Time Trend in Longitudinal Studies
61(5)
3.2 Random Effects for Individual Variability in Response
66(6)
3.2.1 Random Intercept Model
66(2)
3.2.2 Random Intercept and Slope Model
68(2)
3.2.3 More Complex Random Effects Models
70(2)
3.3 Multilevel Models
72(2)
3.4 Covariance-Pattern Models
74(3)
3.5 Combinations of Random Effects and Covariance-Pattern Models
77(1)
3.6 Estimation, Model Fit, and Model Selection
78(2)
3.7 Residuals and Remedial Measures
80(1)
3.8 Examples
81(20)
3.8.1 Augmentation Treatment for Depression
81(7)
3.8.2 Serotonin Levels in Mother-Infant Pairs
88(2)
3.8.3 fMRI Study of Working Memory in Schizophrenia
90(6)
3.8.4 Meta-Analysis of Clinical Trials in Schizophrenia
96(3)
3.8.5 Citalopram Effects on Depressive Symptom Clusters in the STAR*D Study
99(2)
3.9 Summary
101(2)
4 Linear Models for Non-Normal Outcomes
103(40)
4.1 Generalized Linear Models (GLM)
105(15)
4.1.1 Logistic Regression for Binary Data
106(2)
4.1.2 Poisson Regression for Count Data
108(4)
4.1.3 Generalized Linear Models
112(1)
4.1.3.1 Model Definition and Most Commonly Used Specifications
112(2)
4.1.3.2 Overdispersion
114(1)
4.1.3.3 Estimation and Assessment of Model Fit
115(1)
4.1.3.4 Zero-Inflated and Hurdle Models
116(1)
4.1.3.5 GLM Summary
117(1)
4.1.4 GLM Extensions for Ordinal and Nominal Data
117(1)
4.1.4.1 Cumulative Logit Model for Ordinal Data
117(3)
4.1.4.2 Baseline Category Logit Model for Nominal Data
120(1)
4.2 Generalized Estimating Equations (GEE)
120(9)
4.2.1 Modeling the Mean
121(1)
4.2.2 Specifying the Working Correlation Structure
122(2)
4.2.3 Estimation Process and Properties
124(1)
4.2.4 GEE Analysis of Count Data: Number of Drinking Days in the COMBINE Study
124(3)
4.2.5 GEE Analysis of Ordinal Data: Self-Rated Health in the Health and Retirement Study
127(2)
4.3 Generalized Linear Mixed Models (GLMM)
129(12)
4.3.1 Modeling the Mean
130(1)
4.3.2 Implied Variance-Covariance Structure
131(1)
4.3.3 Estimation, Model Fit, and Interpretation
132(1)
4.3.4 GLMM for Count Data: Number of Drinking Days in the COMBINE Study
132(1)
4.3.4.1 Random Intercept Model
133(1)
4.3.4.2 Random Intercept and Slope Model
134(4)
4.3.5 GLMM Analysis of Ordinal Data: Self-Rated Health in the Health and Retirement Study
138(3)
4.4 Summary
141(2)
5 Non-Parametric Methods for the Analysis of Repeatedly Measured Data
143(16)
5.1 Classical Non-Parametric Methods for Independent Samples
144(1)
5.2 Simple Non-Parametric Tests for Repeated Measures Data
145(2)
5.3 Non-Parametric Analysis of Repeated Measures Data in Factorial Designs
147(5)
5.4 Data Examples
152(5)
5.4.1 COMBINE Study
152(2)
5.4.2 Human Laboratory Study of Menthol's Effects on Nicotine Reinforcement in Smokers
154(3)
5.5 Summary
157(2)
6 Post Hoc Analysis and Adjustments for Multiple Comparisons
159(28)
6.1 Historical Overview of Approaches to Multiple Comparisons
161(1)
6.2 The Need for Multiple Comparison Correction
162(2)
6.2.1 Hypothesis Testing
162(2)
6.2.2 Confidence Intervals
164(1)
6.3 Standard Approaches to Multiple Comparisons
164(11)
6.3.1 The Bonferroni Multiple Correction Procedure
165(2)
6.3.2 Tukey's Multiple Comparison Procedure
167(3)
6.3.3 Scheffe's Multiple Comparison Procedure
170(1)
6.3.4 Dunnett's Multiple Comparison Procedure
171(1)
6.3.5 Other Classical Multiple Comparison Procedures
171(1)
6.3.6 Classical Multiple Comparison Procedures for Repeated Measures Data
172(1)
6.3.7 Families of Comparisons and Robustness to Assumption Violations
172(1)
6.3.8 Post Hoc Analyses in Models for Repeated Measures
173(2)
6.4 Stepwise Modifications of the Bonferroni Approach
175(2)
6.4.1 The Bonferroni-Holm Multiple Comparison Procedure
175(1)
6.4.2 Hochberg's Multiple Comparison Procedure
176(1)
6.4.3 Hommel's Multiple Comparison Procedure
177(1)
6.5 Procedures Controlling the False Discovery Rate (FDR)
177(2)
6.5.1 Benjamini-Hochberg's Multiple Comparison Procedure
178(1)
6.5.2 Benjamini-Yekutieli's Multiple Comparison Procedure
178(1)
6.5.3 Simultaneous Confidence Intervals Controlling the False Coverage Rate
179(1)
6.6 Procedures Based on Resampling and Bootstrap
179(1)
6.7 Data Examples
179(6)
6.7.1 Post Hoc Testing in the COMBINE Study
179(1)
6.7.1.1 Correction for Simultaneous Pairwise Comparisons on Different Outcome Measures
180(1)
6.7.1.2 Post Hoc Testing of Significant Main Effects and Interactions
181(2)
6.7.1.3 Multiple Comparison Adjustments for Post Hoc Analysis in Models for Repeated Measures Data
183(1)
6.7.2 Post Hoc Testing in the fMRI Study of Working Memory in Schizophrenia
184(1)
6.8 Guidelines to Multiple Comparison Procedures
185(1)
6.9 Summary
185(2)
7 Handling of Missing Data and Dropout in Longitudinal Studies
187(28)
7.1 Types of Missing Data
188(3)
7.2 Deletion and Substitution Methods for Handling Missing Data
191(3)
7.3 Multiple Imputation
194(3)
7.4 Full Information Maximum Likelihood
197(1)
7.5 Weighted GEE
198(1)
7.6 Methods for Informatively Missing Data
198(2)
7.7 Data Examples
200(11)
7.7.1 Missing Data Models in the Augmentation Depression Study
201(6)
7.7.2 Missing Data Models in the Health and Retirement Study
207(4)
7.8 Guidelines for Handling Missing Data
211(1)
7.9 Summary
212(3)
8 Controlling for Covariates in Studies with Repeated Measures
215(24)
8.1 Controlling for Covariates in Cross-Sectional and Simple Longitudinal Designs
216(6)
8.1.1 Steps in Classical ANCOVA
217(3)
8.1.2 Analysis of Covariance in Randomized Studies
220(1)
8.1.3 Analysis of Covariance in Observational Studies
221(1)
8.2 Controlling for Covariates in Clustered and Longitudinal Studies
222(2)
8.3 Propensity Scoring
224(3)
8.4 Data Examples
227(11)
8.4.1 ANCOVA of Endpoint Drinks per Day Controlling for Baseline Drinking Intensity in the COMBINE Study
227(3)
8.4.2 Analysis of Monthly Drinks per Day Controlling for Baseline Drinking Intensity in the COMBINE Study
230(1)
8.4.3 Mixed-Effects Analysis of Depression Trajectories during Recent Unemployment with a Time-Dependent Covariate
231(3)
8.4.4 Estimating the Effect of Transition to Retirement on Change in Self-Rated Health in the Health and Retirement Study
234(4)
8.5 Summary
238(1)
9 Assessment of Moderator and Mediator Effects
239(30)
9.1 Moderators
240(5)
9.1.1 Assessment of Moderator Effects
240(3)
9.1.2 Moderator Effects in Experimental and Observational Studies
243(1)
9.1.3 Moderator Effects in Longitudinal Studies
244(1)
9.1.4 Moderator Effects in Studies with Clustering
245(1)
9.1.5 Multiplicity Corrections for Moderator Analyses
245(1)
9.2 Data Examples of Moderator Analysis
245(3)
9.2.1 Moderation of Treatment Effects on Number of Drinking Days in the COMBINE Study
245(2)
9.2.2 Type of Cigarettes Smoked as a Moderator of Nicotine Effects in Smokers
247(1)
9.3 Mediators
248(13)
9.3.1 Assessment of Mediator Effects: The Baron and Kenny Approach
249(2)
9.3.2 Assessment of Mediator Effects: The Causal Inference Approach
251(6)
9.3.3 Mediators in Experimental and Observational Studies
257(1)
9.3.4 Multiple Mediators
258(1)
9.3.5 Mediator Effects in Longitudinal Studies
258(2)
9.3.6 Mediator Effects in Studies with Clustered Data
260(1)
9.3.7 Moderated Mediation and Mediated Moderation
261(1)
9.4 Data Examples of Mediation Analysis
261(5)
9.4.1 Improvement in Sleep as Mediator of the Effects of Modafinil on Cocaine Use in Cocaine-Dependent Patients
261(2)
9.4.2 Intent-to-Smoke as a Mediator of the Effect of a School-Based Drug Prevention Program on Smoking
263(2)
9.4.3 Mediator Effects in a Simulated Repeated Measures Data Set
265(1)
9.5 Summary
266(3)
10 Mixture Models for Trajectory Analyses
269(20)
10.1 Latent Class Growth Models (LCGM)
270(3)
10.2 Growth Mixture Models (GMM)
273(4)
10.3 Issues in Building LCGM and GMM
277(3)
10.3.1 Model Fitting
277(1)
10.3.2 Model Selection
278(1)
10.3.3 Guidelines for Model Selection
279(1)
10.4 Data Examples
280(7)
10.4.1 Trajectories of Heavy Drinking in COMBINE
280(4)
10.4.2 Trajectories of Depression Symptoms in STAR*D
284(3)
10.5 Summary
287(2)
11 Study Design and Sample Size Calculations
289(32)
11.1 Study Design Considerations
290(4)
11.1.1 Study Objectives
290(1)
11.1.2 Target Population
290(1)
11.1.3 Study Sample
290(1)
11.1.4 Outcome Measures
291(1)
11.1.5 Study Design
291(1)
11.1.6 Data Collection, Management, and Monitoring
291(1)
11.1.7 Statistical Analysis Plan
292(1)
11.1.8 Sample Size Estimation or Power Analysis
293(1)
11.1.9 Reporting Guidelines
293(1)
11.2 Repeated Measures Study Designs
294(2)
11.2.1 Commonly Used Experimental Designs
294(1)
11.2.2 Observational Study Designs
295(1)
11.3 Sample Size Calculations for Traditional Methods
296(9)
11.3.1 Power Calculations for Simple Hypothesis Tests
296(5)
11.3.2 Power Calculations for Confidence Intervals
301(1)
11.3.3 Example Power Calculations for a Two-Group Study
302(1)
11.3.3.1 Hypothesis Test for the Difference of Two Means
303(1)
11.3.3.2 Confidence Interval for the Difference of Two Means
304(1)
11.4 Sample Size Calculations for Studies with Repeated Measures
305(10)
11.4.1 Clustered Data
306(2)
11.4.2 Longitudinal Data
308(1)
11.4.2.1 Power Calculations for Summary Measures
309(1)
11.4.2.2 Power Calculations for Traditional Methods (rANOVA, rMANOVA)
310(2)
11.4.2.3 Power Calculations for Mixed-Effects Models
312(2)
11.4.2.4 Power Calculations for GEE Models
314(1)
11.5 Randomization Methods for Experimental Studies
315(3)
11.6 Summary
318(3)
12 Summary and Further Readings
321(8)
12.1 Models for Multiple Outcomes
322(1)
12.2 Non-Linear and Spline Modeling of Time Effects
322(1)
12.3 Transition Models
323(1)
12.4 Survival Analysis
324(1)
12.5 Joint Analysis of Survival Outcomes and Repeated Measures
324(1)
12.6 Models for Intensive Longitudinal Data
325(1)
12.7 Models for Spatial Data
325(1)
12.8 Bayesian Methods
326(1)
12.9 Software
326(1)
12.10 Concluding Remarks
327(2)
References 329(16)
Index 345
Ralitza Gueorguieva is a Senior Research Scientist at the Department of Biostatistics, Yale School of Public Health. She has more than 20 years experience in statistical methodology development and collaborations with psychiatrists and other researchers, and is the author of over 130 peer-reviewed publications.
Biežāk uzdotie jautājumi par e-grāmatām Biežāk uzdotie jautājumi par e-grāmatām
Permanent link: https://www.kriso.lv/db/97814987407772e.html
Keywords: