Download Full Version ebook - Visit ebooknice.

com

(Ebook) Survival Analysis with Interval-Censored

Data: A Practical Approach with Examples in R,
SAS, and BUGS by Kris Bogaerts, Arnost Komarek,
Emmanuel Lesaffre ISBN 9781420077476, 1420077473
https://ebooknice.com/product/survival-analysis-with-
interval-censored-data-a-practical-approach-with-examples-
in-r-sas-and-bugs-6838858

Click the button below to download

DOWLOAD EBOOK

Discover More Ebook - Explore Now at ebooknice.com

 Instant digital products (PDF, ePub, MOBI) ready for you
Download now and discover formats that fit your needs...

Start reading on any device today!

(Ebook) Biota Grow 2C gather 2C cook by Loucas, Jason;

Viles, James ISBN 9781459699816, 9781743365571,
9781925268492, 1459699815, 1743365578, 1925268497
https://ebooknice.com/product/biota-grow-2c-gather-2c-cook-6661374

ebooknice.com

(Ebook) Matematik 5000+ Kurs 2c Lärobok by Lena

Alfredsson, Hans Heikne, Sanna Bodemyr ISBN 9789127456600,
9127456609
https://ebooknice.com/product/matematik-5000-kurs-2c-larobok-23848312

ebooknice.com

(Ebook) SAT II Success MATH 1C and 2C 2002 (Peterson's SAT

II Success) by Peterson's ISBN 9780768906677, 0768906679

https://ebooknice.com/product/sat-ii-success-
math-1c-and-2c-2002-peterson-s-sat-ii-success-1722018

ebooknice.com

(Ebook) Master SAT II Math 1c and 2c 4th ed (Arco Master

the SAT Subject Test: Math Levels 1 & 2) by Arco ISBN
9780768923049, 0768923042
https://ebooknice.com/product/master-sat-ii-math-1c-and-2c-4th-ed-
arco-master-the-sat-subject-test-math-levels-1-2-2326094

ebooknice.com
(Ebook) Cambridge IGCSE and O Level History Workbook 2C -
Depth Study: the United States, 1919-41 2nd Edition by
Benjamin Harrison ISBN 9781398375147, 9781398375048,
1398375144, 1398375047
https://ebooknice.com/product/cambridge-igcse-and-o-level-history-
workbook-2c-depth-study-the-united-states-1919-41-2nd-edition-53538044

ebooknice.com

(Ebook) Doing Bayesian Data Analysis: A Tutorial

Introduction with R and BUGS by John K. Kruschke, Kruschke
John ISBN 9780123814852, 0123814855
https://ebooknice.com/product/doing-bayesian-data-analysis-a-tutorial-
introduction-with-r-and-bugs-2334782

ebooknice.com

(Ebook) IFRS 9 and CECL Credit Risk Modelling and

Validation: A Practical Guide with Examples Worked in R
and SAS by Tiziano Bellini ISBN 9780128149409, 012814940X
https://ebooknice.com/product/ifrs-9-and-cecl-credit-risk-modelling-
and-validation-a-practical-guide-with-examples-worked-in-r-and-
sas-10013860
ebooknice.com

(Ebook) Statistical Analysis of Financial Data: With

Examples In R by James Gentle ISBN 9781138599499,
1138599492
https://ebooknice.com/product/statistical-analysis-of-financial-data-
with-examples-in-r-11015738

ebooknice.com

(Ebook) Analysis of Correlated Data with SAS and R by

Mohamed M. Shoukri ISBN 9781138197459, 1138197459

https://ebooknice.com/product/analysis-of-correlated-data-with-sas-
and-r-7145340

ebooknice.com
 SURVIVAL ANALYSIS with
INTERVAL-CENSORED DATA
A Practical Approach with
Examples in R, SAS, and BUGS
CHAPMAN & HALL/CRC
Interdisciplinar y Statistics Series
Series editors: N. Keiding, B.J.T. Morgan, C.K. Wikle, P. van der Heijden
Published titles
AGE-PERIOD-COHORT ANALYSIS: NEW MODELS, METHODS, AND
EMPIRICAL APPLICATIONS Y. Yang and K. C. Land
ANALYSIS OF CAPTURE-RECAPTURE DATA R. S. McCrea and B. J.T. Morgan
AN INVARIANT APPROACH TO STATISTICAL ANALYSIS OF SHAPES
S. Lele and J. Richtsmeier
ASTROSTATISTICS G. Babu and E. Feigelson
BAYESIAN ANALYSIS FOR POPULATION ECOLOGY R. King, B. J.T. Morgan,
O. Gimenez, and S. P. Brooks
BAYESIAN DISEASE MAPPING: HIERARCHICAL MODELING IN SPATIAL
EPIDEMIOLOGY, SECOND EDITION A. B. Lawson
BIOEQUIVALENCE AND STATISTICS IN CLINICAL PHARMACOLOGY
S. Patterson and B. Jones
CAPTURE-RECAPTURE METHODS FOR THE SOCIAL AND MEDICAL
SCIENCES
D. Böhning, P. G. M. van der Heijden, and J. Bunge
CLINICAL TRIALS IN ONCOLOGY,THIRD EDITION S. Green, J. Benedetti, A. Smith,
and J. Crowley
CLUSTER RANDOMISED TRIALS R.J. Hayes and L.H. Moulton
CORRESPONDENCE ANALYSIS IN PRACTICE,THIRD EDITION M. Greenacre
THE DATA BOOK: COLLECTION AND MANAGEMENT OF RESEARCH DATA
M. Zozus
DESIGN AND ANALYSIS OF QUALITY OF LIFE STUDIES IN CLINICAL TRIALS,
SECOND EDITION D.L. Fairclough
DYNAMICAL SEARCH L. Pronzato, H. Wynn, and A. Zhigljavsky
FLEXIBLE IMPUTATION OF MISSING DATA S. van Buuren
GENERALIZED LATENT VARIABLE MODELING: MULTILEVEL, LONGITUDI-
NAL, AND STRUCTURAL EQUATION MODELS A. Skrondal and S. Rabe-Hesketh
GRAPHICAL ANALYSIS OF MULTI-RESPONSE DATA K. Basford and J. Tukey
INTRODUCTION TO COMPUTATIONAL BIOLOGY: MAPS, SEQUENCES, AND
GENOMES M. Waterman
MARKOV CHAIN MONTE CARLO IN PRACTICE W. Gilks, S. Richardson, and
D. Spiegelhalter
Published titles
MEASUREMENT ERROR ANDMISCLASSIFICATION IN STATISTICS AND EPIDE-
MIOLOGY: IMPACTS AND BAYESIAN ADJUSTMENTS P. Gustafson
MEASUREMENT ERROR: MODELS, METHODS, AND APPLICATIONS
J. P. Buonaccorsi
MEASUREMENT ERROR: MODELS, METHODS, AND APPLICATIONS
J. P. Buonaccorsi
MENDELIAN RANDOMIZATION: METHODS FOR USING GENETIC VARIANTS
IN CAUSAL ESTIMATION S.Burgess and S.G. Thompson
META-ANALYSIS OF BINARY DATA USINGPROFILE LIKELIHOOD D. Böhning,
R. Kuhnert, and S. Rattanasiri
MISSING DATA ANALYSIS IN PRACTICE T. Raghunathan
MODERN DIRECTIONAL STATISTICS C. Ley and T. Verdebout
POWER ANALYSIS OF TRIALS WITH MULTILEVEL DATA M. Moerbeek and
S. Teerenstra
SPATIAL POINT PATTERNS: METHODOLOGY AND APPLICATIONS WITH R
A. Baddeley, E Rubak, and R. Turner
STATISTICAL ANALYSIS OF GENE EXPRESSION MICROARRAY DATA T. Speed
STATISTICAL ANALYSIS OF QUESTIONNAIRES: A UNIFIED APPROACH
BASED ON R AND STATA F. Bartolucci, S. Bacci, and M. Gnaldi
STATISTICAL AND COMPUTATIONAL PHARMACOGENOMICS R. Wu and M. Lin
STATISTICS IN MUSICOLOGY J. Beran
STATISTICS OF MEDICAL IMAGING T. Lei
STATISTICAL CONCEPTS AND APPLICATIONS IN CLINICAL MEDICINE
J. Aitchison, J.W. Kay, and I.J. Lauder
STATISTICAL AND PROBABILISTIC METHODS IN ACTUARIAL SCIENCE
P.J. Boland
STATISTICAL DETECTION AND SURVEILLANCE OF GEOGRAPHIC CLUSTERS
P. Rogerson and I.Yamada
STATISTICAL METHODS IN PSYCHIATRY AND RELATED FIELDS:
LONGITUDINAL, CLUSTERED, AND OTHER REPEATED MEASURES DATA
R. Gueorguieva
STATISTICS FOR ENVIRONMENTAL BIOLOGY AND TOXICOLOGY A. Bailer
and W. Piegorsch
STATISTICS FOR FISSION TRACK ANALYSIS R.F. Galbraith
SURVIVAL ANALYSIS WITH INTERVAL-CENSORED DATA: A PRACTICAL
APPROACH WITH EXAMPLES IN R, SAS, AND BUGS
K. Bogaerts, A. Komárek, and E. Lesaffre
VISUALIZING DATA PATTERNS WITH MICROMAPS D.B. Carr and L.W. Pickle
 C hap m an & H al l / CR C
I n te rd is ci pl in ar y St a t i st i c s S e r i e s

SURVIVAL ANALYSIS with

INTERVAL-CENSORED DATA
A Practical Approach with
Examples in R, SAS, and BUGS

Kris Bogaerts
Arnošt Komárek
Emmanuel Lesaffre
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2018 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed on acid-free paper

Version Date: 20171012

International Standard Book Number-13: 978-1-4200-7747-6 (Hardback)

This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access
www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc.
(CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization
that provides licenses and registration for a variety of users. For organizations that have been granted
a photocopy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and
are used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
Contents

List of Tables xvii

List of Figures xxi

Notation xxvii

Preface xxix

I Introduction 1
1 Introduction 3

1.1 Survival concepts . . . . . . . . . . . . . . . . . . . . . . . . 3

1.2 Types of censoring . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Right censoring . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Interval and left censoring . . . . . . . . . . . . . . . . 4
1.2.3 Some special cases of interval censoring . . . . . . . . 5
1.2.4 Doubly interval censoring . . . . . . . . . . . . . . . . 7
1.2.5 Truncation . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Ignoring interval censoring . . . . . . . . . . . . . . . . . . . 8
1.4 Independent noninformative censoring . . . . . . . . . . . . . 13
1.4.1 Independent noninformative right censoring . . . . . . 13
1.4.2 Independent noninformative interval censoring . . . . 14
1.5 Frequentist inference . . . . . . . . . . . . . . . . . . . . . . 15
1.5.1 Likelihood for interval-censored data . . . . . . . . . . 15
1.5.2 Maximum likelihood theory . . . . . . . . . . . . . . . 17
1.6 Data sets and research questions . . . . . . . . . . . . . . . . 20
1.6.1 Homograft study . . . . . . . . . . . . . . . . . . . . . 20
1.6.2 Breast cancer study . . . . . . . . . . . . . . . . . . . 22
1.6.3 AIDS clinical trial . . . . . . . . . . . . . . . . . . . . 22
1.6.4 Sensory shelf life study . . . . . . . . . . . . . . . . . . 23
1.6.5 Survey on mobile phone purchases . . . . . . . . . . . 25
1.6.6 Mastitis study . . . . . . . . . . . . . . . . . . . . . . 27
1.6.7 Signal Tandmobiel study . . . . . . . . . . . . . . . . 28
1.7 Censored data in R and SAS . . . . . . . . . . . . . . . . . . 30
1.7.1 R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

vii
viii Contents

1.7.2 SAS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

2 Inference for right-censored data 35

2.1 Estimation of the survival function . . . . . . . . . . . . . . . 35

2.1.1 Nonparametric maximum likelihood estimation . . . . 35
2.1.2 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.1.3 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 40
2.2 Comparison of two survival distributions . . . . . . . . . . . 41
2.2.1 Review of significance tests . . . . . . . . . . . . . . . 41
2.2.2 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.2.3 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 45
2.3 Regression models . . . . . . . . . . . . . . . . . . . . . . . . 46
2.3.1 Proportional hazards model . . . . . . . . . . . . . . . 46
2.3.1.1 Model description and estimation . . . . . . 46
2.3.1.2 Model checking . . . . . . . . . . . . . . . . . 47
2.3.1.3 R solution . . . . . . . . . . . . . . . . . . . 50
2.3.1.4 SAS solution . . . . . . . . . . . . . . . . . . 52
2.3.2 Accelerated failure time model . . . . . . . . . . . . . 53
2.3.2.1 Model description and estimation . . . . . . 53
2.3.2.2 Model checking . . . . . . . . . . . . . . . . . 55
2.3.2.3 R solution . . . . . . . . . . . . . . . . . . . 56
2.3.2.4 SAS solution . . . . . . . . . . . . . . . . . . 58

II Frequentist methods for interval-censored data 61

3 Estimating the survival distribution 63

3.1 Nonparametric maximum likelihood . . . . . . . . . . . . . . 63

3.1.1 Estimation . . . . . . . . . . . . . . . . . . . . . . . . 63
3.1.2 Asymptotic results . . . . . . . . . . . . . . . . . . . . 67
3.1.3 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.1.4 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 71
3.2 Parametric modelling . . . . . . . . . . . . . . . . . . . . . . 77
3.2.1 Estimation . . . . . . . . . . . . . . . . . . . . . . . . 78
3.2.2 Model selection . . . . . . . . . . . . . . . . . . . . . . 78
3.2.3 Goodness of fit . . . . . . . . . . . . . . . . . . . . . . 78
3.2.4 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.2.5 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 82
3.3 Smoothing methods . . . . . . . . . . . . . . . . . . . . . . . 85
3.3.1 Logspline density estimation . . . . . . . . . . . . . . 85
3.3.1.1 A smooth approximation to the density . . . 85
3.3.1.2 Maximum likelihood estimation . . . . . . . 86
3.3.1.3 R solution . . . . . . . . . . . . . . . . . . . 87
3.3.2 Classical Gaussian mixture model . . . . . . . . . . . 90
 Contents ix

3.3.3 Penalized Gaussian mixture model . . . . . . . . . . . 93

3.3.3.1 R solution . . . . . . . . . . . . . . . . . . . 99
3.4 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 104

4 Comparison of two or more survival distributions 105

4.1 Nonparametric comparison of survival curves . . . . . . . . . 105

4.1.1 Weighted log-rank test: derivation . . . . . . . . . . . 107
4.1.2 Weighted log-rank test: linear form . . . . . . . . . . . 109
4.1.3 Weighted log-rank test: derived from the linear
transformation model . . . . . . . . . . . . . . . . . . 109
4.1.4 Weighted log-rank test: the G%,γ family . . . . . . . . 111
4.1.5 Weighted log-rank test: significance testing . . . . . . 112
4.1.6 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 117
4.1.7 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 123
4.2 Sample size calculation . . . . . . . . . . . . . . . . . . . . . 127
4.3 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 128

5 The proportional hazards model 131

5.1 Parametric approaches . . . . . . . . . . . . . . . . . . . . . 132

5.1.1 Maximum likelihood estimation . . . . . . . . . . . . . 132
5.1.2 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.1.3 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 135
5.2 Towards semiparametric approaches . . . . . . . . . . . . . . 136
5.2.1 Piecewise exponential baseline survival model . . . . . 137
5.2.1.1 Model description and estimation . . . . . . 137
5.2.1.2 R solution . . . . . . . . . . . . . . . . . . . 138
5.2.1.3 SAS solution . . . . . . . . . . . . . . . . . . 140
5.2.2 SemiNonParametric approach . . . . . . . . . . . . . . 141
5.2.2.1 Model description and estimation . . . . . . 141
5.2.2.2 SAS solution . . . . . . . . . . . . . . . . . . 142
5.2.3 Spline-based smoothing approaches . . . . . . . . . . . 144
5.2.3.1 Two spline-based smoothing approaches . . . 144
5.2.3.2 R solution . . . . . . . . . . . . . . . . . . . 146
5.2.3.3 SAS solution . . . . . . . . . . . . . . . . . . 147
5.3 Semiparametric approaches . . . . . . . . . . . . . . . . . . . 149
5.3.1 Finkelstein’s approach . . . . . . . . . . . . . . . . . . 149
5.3.2 Farrington’s approach . . . . . . . . . . . . . . . . . . 150
5.3.3 Iterative convex minorant algorithm . . . . . . . . . . 153
5.3.4 Grouped proportional hazards model . . . . . . . . . . 153
5.3.5 Practical applications . . . . . . . . . . . . . . . . . . 154
5.3.5.1 Two examples . . . . . . . . . . . . . . . . . 154
5.3.5.2 R solution . . . . . . . . . . . . . . . . . . . 155
5.3.5.3 SAS solution . . . . . . . . . . . . . . . . . . 158
x Contents

5.4 Multiple imputation approach . . . . . . . . . . . . . . . . . 163

5.4.1 Data augmentation algorithm . . . . . . . . . . . . . . 163
5.4.2 Multiple imputation for interval-censored survival times 165
5.4.2.1 R solution . . . . . . . . . . . . . . . . . . . 167
5.4.2.2 SAS solution . . . . . . . . . . . . . . . . . . 168
5.5 Model checking . . . . . . . . . . . . . . . . . . . . . . . . . . 170
5.5.1 Checking the PH model . . . . . . . . . . . . . . . . . 170
5.5.2 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 172
5.5.3 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 173
5.6 Sample size calculation . . . . . . . . . . . . . . . . . . . . . 175
5.7 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 175

6 The accelerated failure time model 179

6.1 Parametric model . . . . . . . . . . . . . . . . . . . . . . . . 180

6.1.1 Maximum likelihood estimation . . . . . . . . . . . . . 181
6.1.2 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 187
6.1.3 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 193
6.2 Penalized Gaussian mixture model . . . . . . . . . . . . . . . 195
6.2.1 Penalized maximum likelihood estimation . . . . . . . 196
6.2.2 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 204
6.3 SemiNonParametric approach . . . . . . . . . . . . . . . . . 215
6.3.1 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 216
6.4 Model checking . . . . . . . . . . . . . . . . . . . . . . . . . . 216
6.5 Sample size calculation . . . . . . . . . . . . . . . . . . . . . 216
6.5.1 Computational approach . . . . . . . . . . . . . . . . . 216
6.5.2 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 218
6.6 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 219

7 Bivariate survival times 221

7.1 Nonparametric estimation of the bivariate survival function . 222

7.1.1 The NPMLE of a bivariate survival function . . . . . 222
7.1.2 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 225
7.1.3 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 228
7.2 Parametric models . . . . . . . . . . . . . . . . . . . . . . . . 229
7.2.1 Model description and estimation . . . . . . . . . . . . 229
7.2.2 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 230
7.2.3 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 231
7.3 Copula models . . . . . . . . . . . . . . . . . . . . . . . . . . 232
7.3.1 Background . . . . . . . . . . . . . . . . . . . . . . . . 232
7.3.2 Estimation procedures . . . . . . . . . . . . . . . . . . 236
7.3.3 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 239
7.4 Flexible survival models . . . . . . . . . . . . . . . . . . . . . 240
7.4.1 The penalized Gaussian mixture model . . . . . . . . 240
 Contents xi

7.4.2 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 242

7.5 Estimation of the association parameter . . . . . . . . . . . . 243
7.5.1 Measures of association . . . . . . . . . . . . . . . . . 244
7.5.2 Estimating measures of association . . . . . . . . . . . 245
7.5.3 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 248
7.5.4 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 250
7.6 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 250

8 Additional topics 253

8.1 Doubly interval-censored data . . . . . . . . . . . . . . . . . 254

8.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . 254
8.1.2 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 260
8.2 Regression models for clustered data . . . . . . . . . . . . . . 262
8.2.1 Frailty models . . . . . . . . . . . . . . . . . . . . . . 263
8.2.1.1 R solution . . . . . . . . . . . . . . . . . . . 265
8.2.1.2 SAS solution . . . . . . . . . . . . . . . . . . 268
8.2.2 A marginal approach to correlated survival times . . . 270
8.2.2.1 Independence working model . . . . . . . . . 270
8.2.2.2 SAS solution . . . . . . . . . . . . . . . . . . 271
8.3 A biplot for interval-censored data . . . . . . . . . . . . . . . 274
8.3.1 Classical biplot . . . . . . . . . . . . . . . . . . . . . . 274
8.3.2 Extension to interval-censored observations . . . . . . 275
8.3.3 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 278
8.4 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 279

III Bayesian methods for interval-censored data 281

9 Bayesian concepts 283

9.1 Bayesian inference . . . . . . . . . . . . . . . . . . . . . . . . 284

9.1.1 Parametric versus nonparametric Bayesian approaches 285
9.1.2 Bayesian data augmentation . . . . . . . . . . . . . . . 285
9.1.3 Markov chain Monte Carlo . . . . . . . . . . . . . . . 287
9.1.4 Credible regions and contour probabilities . . . . . . . 289
9.1.5 Selecting and checking the model . . . . . . . . . . . . 291
9.1.6 Sensitivity analysis . . . . . . . . . . . . . . . . . . . . 295
9.2 Nonparametric Bayesian inference . . . . . . . . . . . . . . . 295
9.2.1 Bayesian nonparametric modelling of the hazard
function . . . . . . . . . . . . . . . . . . . . . . . . . . 297
9.2.2 Bayesian nonparametric modelling of the distribution
function . . . . . . . . . . . . . . . . . . . . . . . . . . 299
9.3 Bayesian software . . . . . . . . . . . . . . . . . . . . . . . . 302
9.3.1 WinBUGS and OpenBUGS . . . . . . . . . . . . . . . 302
9.3.2 JAGS . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
xii Contents

9.3.3 R software . . . . . . . . . . . . . . . . . . . . . . . . . 306

9.3.4 SAS procedures . . . . . . . . . . . . . . . . . . . . . . 306
9.3.5 Stan software . . . . . . . . . . . . . . . . . . . . . . . 306
9.4 Applications for right-censored data . . . . . . . . . . . . . . 306
9.4.1 Parametric models . . . . . . . . . . . . . . . . . . . . 307
9.4.1.1 BUGS solution . . . . . . . . . . . . . . . . . 308
9.4.1.2 SAS solution . . . . . . . . . . . . . . . . . . 315
9.4.2 Nonparametric Bayesian estimation of a survival curve 316
9.4.2.1 R solution . . . . . . . . . . . . . . . . . . . 317
9.4.3 Semiparametric Bayesian survival analysis . . . . . . . 318
9.4.3.1 BUGS solution . . . . . . . . . . . . . . . . . 321
9.5 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 323

10 Bayesian estimation of the survival distribution for interval-

censored observations 325

10.1 Bayesian parametric modelling . . . . . . . . . . . . . . . . . 325

10.1.1 JAGS solution . . . . . . . . . . . . . . . . . . . . . . 326
10.1.2 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 331
10.2 Bayesian smoothing methods . . . . . . . . . . . . . . . . . . 335
10.2.1 Classical Gaussian mixture . . . . . . . . . . . . . . . 335
10.2.1.1 R solution . . . . . . . . . . . . . . . . . . . 339
10.2.2 Penalized Gaussian mixture . . . . . . . . . . . . . . . 343
10.3 Nonparametric Bayesian estimation . . . . . . . . . . . . . . 343
10.3.1 The Dirichlet Process prior approach . . . . . . . . . . 343
10.3.1.1 R solution . . . . . . . . . . . . . . . . . . . 345
10.3.2 The Dirichlet Process Mixture approach . . . . . . . . 347
10.3.2.1 R solution . . . . . . . . . . . . . . . . . . . 348
10.4 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 353

11 The Bayesian proportional hazards model 355

11.1 Parametric PH model . . . . . . . . . . . . . . . . . . . . . . 355

11.1.1 JAGS solution . . . . . . . . . . . . . . . . . . . . . . 356
11.1.2 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 365
11.2 PH model with flexible baseline hazard . . . . . . . . . . . . 368
11.2.1 Bayesian PH model with a smooth baseline hazard . . 368
11.2.1.1 R solution . . . . . . . . . . . . . . . . . . . 369
11.2.2 PH model with piecewise constant baseline hazard . . 372
11.2.2.1 R solution . . . . . . . . . . . . . . . . . . . 376
11.3 Semiparametric PH model . . . . . . . . . . . . . . . . . . . 380
11.4 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 380
 Contents xiii

12 The Bayesian accelerated failure time model 381

12.1 Bayesian parametric AFT model . . . . . . . . . . . . . . . . 381

12.1.1 JAGS solution . . . . . . . . . . . . . . . . . . . . . . 384
12.1.2 SAS solution . . . . . . . . . . . . . . . . . . . . . . . 387
12.2 AFT model with a classical Gaussian mixture as an error
distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
12.2.1 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 395
12.3 AFT model with a penalized Gaussian mixture as an error
distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
12.3.1 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 408
12.4 Bayesian semiparametric AFT model . . . . . . . . . . . . . 412
12.4.1 R solution . . . . . . . . . . . . . . . . . . . . . . . . . 418
12.5 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 421

13 Additional topics 423

13.1 Hierarchical models . . . . . . . . . . . . . . . . . . . . . . . 424

13.1.1 Parametric shared frailty models . . . . . . . . . . . . 424
13.1.1.1 JAGS solution . . . . . . . . . . . . . . . . . 426
13.1.1.2 SAS solution . . . . . . . . . . . . . . . . . . 429
13.1.2 Flexible shared frailty models . . . . . . . . . . . . . . 430
13.1.2.1 R solution . . . . . . . . . . . . . . . . . . . 436
13.1.3 Semiparametric shared frailty models . . . . . . . . . 440
13.2 Multivariate models . . . . . . . . . . . . . . . . . . . . . . . 440
13.2.1 Parametric bivariate models . . . . . . . . . . . . . . . 440
13.2.1.1 JAGS solution . . . . . . . . . . . . . . . . . 441
13.2.1.2 SAS solution . . . . . . . . . . . . . . . . . . 443
13.2.2 Bivariate copula models . . . . . . . . . . . . . . . . . 444
13.2.3 Flexible bivariate models . . . . . . . . . . . . . . . . 445
13.2.3.1 R solution . . . . . . . . . . . . . . . . . . . 448
13.2.4 Semiparametric bivariate models . . . . . . . . . . . . 451
13.2.4.1 R solution . . . . . . . . . . . . . . . . . . . 451
13.2.5 Multivariate case . . . . . . . . . . . . . . . . . . . . . 455
13.3 Doubly interval censoring . . . . . . . . . . . . . . . . . . . . 457
13.3.1 Parametric modelling of univariate DI-censored data . 457
13.3.1.1 JAGS solution . . . . . . . . . . . . . . . . . 458
13.3.2 Flexible modelling of univariate DI-censored data . . . 461
13.3.2.1 R solution . . . . . . . . . . . . . . . . . . . 462
13.3.3 Semiparametric modelling of univariate DI-censored
data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
13.3.3.1 R solution . . . . . . . . . . . . . . . . . . . 465
13.3.4 Modelling of multivariate DI-censored data . . . . . . 465
13.4 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . 466
xiv Contents

IV Concluding remarks 467

14 Omitted topics and outlook 469

14.1 Omitted topics . . . . . . . . . . . . . . . . . . . . . . . . . . 469

14.1.1 Competing risks and multistate models . . . . . . . . 470
14.1.2 Survival models with a cured subgroup . . . . . . . . . 471
14.1.3 Multilevel models . . . . . . . . . . . . . . . . . . . . . 472
14.1.4 Informative censoring . . . . . . . . . . . . . . . . . . 472
14.1.5 Interval-censored covariates . . . . . . . . . . . . . . . 474
14.1.6 Joint longitudinal and survival models . . . . . . . . . 475
14.1.7 Spatial-temporal models . . . . . . . . . . . . . . . . . 477
14.1.8 Time points measured with error . . . . . . . . . . . . 477
14.1.9 Quantile regression . . . . . . . . . . . . . . . . . . . . 478
14.2 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478

V Appendices 481
A Data sets 483

A.1 Homograft study . . . . . . . . . . . . . . . . . . . . . . . . . 483

A.2 AIDS clinical trial . . . . . . . . . . . . . . . . . . . . . . . . 483
A.3 Survey on mobile phone purchases . . . . . . . . . . . . . . . 485
A.4 Mastitis study . . . . . . . . . . . . . . . . . . . . . . . . . . 485
A.5 Signal Tandmobiel study . . . . . . . . . . . . . . . . . . . . 486

B Distributions 489

B.1 Log-normal LN (γ, α) . . . . . . . . . . . . . . . . . . . . . . 490

B.2 Log-logistic LL(γ, α) . . . . . . . . . . . . . . . . . . . . . . 490
B.3 Weibull W(γ, α) . . . . . . . . . . . . . . . . . . . . . . . . . 491
B.4 Exponential E(α) . . . . . . . . . . . . . . . . . . . . . . . . 492
B.5 Rayleigh R(α) . . . . . . . . . . . . . . . . . . . . . . . . . . 492
B.6 Gamma(γ, α) . . . . . . . . . . . . . . . . . . . . . . . . . . 493
B.7 R solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
B.8 SAS solution . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
B.9 BUGS solution . . . . . . . . . . . . . . . . . . . . . . . . . . 496
B.10 R and BUGS parametrizations . . . . . . . . . . . . . . . . . 497

C Prior distributions 501

C.1 Beta prior: Beta(α1 , α2 ) . . . . . . . . . . . . . . . . . . . . 502

C.2 Dirichlet prior: Dir (α) . . . . . . . . . . . . . . . . . . . . . 503
C.3 Gamma prior: G(α, β) . . . . . . . . . . . . . . . . . . . . . 504
C.4 Inverse gamma prior: IG(α, β) . . . . . . . . . . . . . . . . . 505
C.5 Wishart prior: Wishart(R, k) . . . . . . . . . . . . . . . . . 506
 Contents xv

C.6 Inverse Wishart prior: Wishart(R, k) . . . . . . . . . . . . . 507

C.7 Link between Beta, Dirichlet and Dirichlet Process prior . . 508

D Description of selected R packages 511

D.1 icensBKL package . . . . . . . . . . . . . . . . . . . . . . . . 511

D.2 Icens package . . . . . . . . . . . . . . . . . . . . . . . . . . . 512
D.3 interval package . . . . . . . . . . . . . . . . . . . . . . . . . 513
D.4 survival package . . . . . . . . . . . . . . . . . . . . . . . . . 515
D.5 logspline package . . . . . . . . . . . . . . . . . . . . . . . . 516
D.6 smoothSurv package . . . . . . . . . . . . . . . . . . . . . . 517
D.7 mixAK package . . . . . . . . . . . . . . . . . . . . . . . . . 518
D.8 bayesSurv package . . . . . . . . . . . . . . . . . . . . . . . . 519
D.9 DPpackage package . . . . . . . . . . . . . . . . . . . . . . . 522
D.10 Other packages . . . . . . . . . . . . . . . . . . . . . . . . . . 523

E Description of selected SAS procedures 525

E.1 PROC LIFEREG . . . . . . . . . . . . . . . . . . . . . . . . 525

E.2 PROC RELIABILITY . . . . . . . . . . . . . . . . . . . . . 527
E.3 PROC ICLIFETEST . . . . . . . . . . . . . . . . . . . . . . 527
E.4 PROC ICPHREG . . . . . . . . . . . . . . . . . . . . . . . . 530

F Technical details 535

F.1 Iterative Convex Minorant (ICM) algorithm . . . . . . . . . 535

F.2 Regions of possible support for bivariate interval-censored data 536
F.2.1 Algorithm of Gentleman and Vandal (2001) . . . . . . 536
F.2.2 Algorithm of Bogaerts and Lesaffre (2004) . . . . . . . 537
F.2.3 Height map algorithm of Maathuis (2005) . . . . . . . 538
F.3 Splines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539
F.3.1 Polynomial fitting . . . . . . . . . . . . . . . . . . . . 540
F.3.2 Polynomial splines . . . . . . . . . . . . . . . . . . . . 540
F.3.3 Natural cubic splines . . . . . . . . . . . . . . . . . . . 541
F.3.4 Truncated power series . . . . . . . . . . . . . . . . . . 541
F.3.5 B-splines . . . . . . . . . . . . . . . . . . . . . . . . . 541
F.3.6 M-splines and I-splines . . . . . . . . . . . . . . . . . . 543
F.3.7 Penalized splines (P-splines) . . . . . . . . . . . . . . . 543

References 547

Author Index 568

Subject Index 577

List of Tables

1.1 Taxonomy of interval-censored observations . . . . . . . . . 5

1.2 Simulation study illustrating the effect of mid-point imputa-
tion and interval censoring. . . . . . . . . . . . . . . . . . . 12
1.3 Data of breast cancer study . . . . . . . . . . . . . . . . . . 21
1.4 Data of sensory shelf life study . . . . . . . . . . . . . . . . 24
1.5 Signal Tandmobiel study . . . . . . . . . . . . . . . . . . . . 30
1.6 Interval censoring in R . . . . . . . . . . . . . . . . . . . . . 33
1.7 Interval censoring in SAS . . . . . . . . . . . . . . . . . . . . 34

2.1 Transformations for confidence intervals . . . . . . . . . . . 37

2.2 Observed counts at time tj for derivation of the log-rank test. 42
2.3 Homograft study. Two-sample tests comparing aortic donor
grafts with pulmonary donor grafts. . . . . . . . . . . . . . . 43
2.4 Homograft study. Cox proportional hazards model. . . . . . 48
2.5 Homograft study. Accelerated failure time model. . . . . . . 56
2.6 Homograft study. Weibull AFT model. . . . . . . . . . . . . 56

3.1 Breast cancer study. Regions of possible support and NPMLE

equivalence classes for the radiotherapy-only group. . . . . . 68
3.2 Signal Tandmobiel study. Parametric modelling of emergence
of tooth 44 of boys. . . . . . . . . . . . . . . . . . . . . . . . 80
3.3 Signal Tandmobiel study (boys). Estimated parameters of the
emergence distribution of tooth 44. . . . . . . . . . . . . . . 97

4.1 Signal Tandmobiel study. Two-sample tests. . . . . . . . . . 115

4.2 Signal Tandmobiel study. Tests comparing three samples. . . 117

5.1 Method of Farrington (1996). Definition of binary response yi

and auxiliary intervals Bi . . . . . . . . . . . . . . . . . . . 152
5.2 Sensory shelf life study . . . . . . . . . . . . . . . . . . . . . 176

6.1 Signal Tandmobiel study. Parametric AFT model for emer-

gence of tooth 44. Model fit statistics obtained with R package
survival. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
6.2 Signal Tandmobiel study. AFT model for emergence of tooth
44 with a normal error distribution. . . . . . . . . . . . . . . 183

xvii
xviii List of Tables

6.3 Signal Tandmobiel study. AFT models for emergence of tooth

44 with a PGM error distribution. Model fit statistics. . . . 200
6.4 Signal Tandmobiel study. Mean AFT model for emergence of
tooth 44 with a PGM error distribution. . . . . . . . . . . . 202
6.5 Signal Tandmobiel study. Mean-scale AFT model for emer-
gence of tooth 44 with a PGM error distribution. . . . . . . 202
6.6 Signal Tandmobiel study. Estimated mean emergence times
based on the PGM AFT models. . . . . . . . . . . . . . . . 203

7.1 AIDS clinical trial . . . . . . . . . . . . . . . . . . . . . . . . 226

8.1 Mobile study. Ad hoc analyses of doubly interval-censored

data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
8.2 Mastitis study. Shared frailty models . . . . . . . . . . . . . 266

9.1 Homograft study. Bayesian Weibull AFT model. . . . . . . . 308

9.2 Homograft study. Bayesian accelerated failure time model. . 308

10.1 Signal Tandmobiel study. Bayesian parametric modelling of

emergence of tooth 44 of boys. . . . . . . . . . . . . . . . . . 326
10.2 Signal Tandmobiel study (boys). Posterior summary statistics
for the parameters of the emergence distribution of tooth 44. 339

11.1 Breast cancer study. Bayesian Weibull PH model. . . . . . . 357

11.2 Signal Tandmobiel study. Comparison of models M1 and
model M3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375

12.1 Signal Tandmobiel study. Bayesian Weibull AFT model for

emergence of tooth 44. . . . . . . . . . . . . . . . . . . . . . 383
12.2 Signal Tandmobiel study. AFT model for emergence of tooth
44 with a CGM error distribution. . . . . . . . . . . . . . . 392
12.3 Signal Tandmobiel study. AFT model for emergence of tooth
44 with a PGM error distribution. . . . . . . . . . . . . . . . 406
12.4 Signal Tandmobiel study. Semiparametric Bayesian AFT
model for emergence of tooth 44 with log-normal baseline sur-
vival distribution. Posterior summary measures. . . . . . . . 416

13.1 Mastitis study. Bayesian shared frailty models . . . . . . . . 425

13.2 Signal Tandmobiel study. Bayesian flexible AFT random-
effects applied to emergence times of permanent first premo-
lars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
13.3 Signal Tandmobiel study. Bivariate AFT models applied to
emergence times of horizontally symmetric teeth 14, 24 and
vertically symmetric teeth 24, 34, respectively. . . . . . . . . 448
13.4 Mobile study. Posterior summary measures from two models
that predict time to buy a new mobile phone. . . . . . . . . 460
 List of Tables xix

A.1 Data of AIDS trial . . . . . . . . . . . . . . . . . . . . . . . 484

B.1 Parametrizations of the distributions of the event time T in

R and BUGS. . . . . . . . . . . . . . . . . . . . . . . . . . . 498
B.2 Parametrizations of the distributions of the log-event time Y
in R and BUGS. . . . . . . . . . . . . . . . . . . . . . . . . . 499

E.1 Supported distribution in PROC LIFEREG . . . . . . . . . . 526

List of Figures

1.1 Right, left and interval censoring. . . . . . . . . . . . . . . . 6

1.2 Doubly interval censoring. . . . . . . . . . . . . . . . . . . . 7
1.3 Breast cancer study of the radiotherapy-only group. Median
time to breast retraction. . . . . . . . . . . . . . . . . . . . . 9
1.4 One (true) simulated data set from either setting used for the
illustration of mid-point imputation. . . . . . . . . . . . . . 11
1.5 Breast cancer study. Observed intervals in months for time to
breast retraction of early breast cancer patients per treatment
group. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.6 AIDS clinical trial. Observed intervals in months for time to
CMV shedding and time to MAC colonization. . . . . . . . 23
1.7 Sensory shelf life study. Shelf life of yoghurt stored at 42◦ C in
hours. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.8 Mobile study. Interval-censored times of previous and of cur-
rent mobile phone purchase. . . . . . . . . . . . . . . . . . . 26
1.9 Mastitis study. Time from parturition to mastitis in days by
location. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.10 Signal Tandmobiel study. FDI numbering system of deciduous
and permanent teeth. . . . . . . . . . . . . . . . . . . . . . . 29

2.1 Homograft study. Kaplan-Meier curve of homograft failure ac-

cording to type of graft . . . . . . . . . . . . . . . . . . . . . 37
2.2 Homograft study. Estimated survival curves for 14-year-old
patients based on the PH model. . . . . . . . . . . . . . . . 49
2.3 Homograft study. Martingale residuals vs. linear predictor. . 50
2.4 Impact of a covariate on the hazard of a PH and AFT model 54

3.1 Determination of regions of possible support for the Turnbull

estimate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.2 Ordering of two interval-censored observations with tied end-
points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.3 Breast cancer study of the radiotherapy-only group. NPMLE
of the cumulative distribution and survival functions. . . . . 69
3.4 Signal Tandmobiel study. Log-normal model for emergence of
tooth 44 of boys . . . . . . . . . . . . . . . . . . . . . . . . . 79

xxi
xxii List of Figures

3.5 Signal Tandmobiel study (boys). Probability plot of log-

normal model for emergence of tooth 44. . . . . . . . . . . . 79
3.6 Breast cancer study (radiotherapy-alone group). Distribution
of the time to breast retraction estimated using the logspline
method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.7 Several densities expressed as two- or four-component Gaus-
sian mixtures. . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.8 Several densities expressed as homoscedastic Gaussian mix-
tures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.9 Signal Tandmobiel study (boys). Distribution of the time to
emergence of tooth 44 estimated using the penalized Gaussian
mixture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
3.10 Signal Tandmobiel study (boys). Distribution of the standard-
ized log-time to emergence of tooth 44 estimated using the
penalized Gaussian mixture compared to parametric densities. 102

4.1 Signal Tandmobiel study. NPMLE of the survival functions for

emergence of permanent tooth 44 in two groups according to
baseline DMF status of primary tooth 84 and in three groups
according to occlusal plaque status of permanent tooth 46. . 114

5.1 Breast cancer study. Validation of PH assumption for treat-

ment via a transformation of the survival function. . . . . . 173
5.2 Breast cancer study. Validation of PH assumption for treat-
ment via residuals. . . . . . . . . . . . . . . . . . . . . . . . 174
5.3 Sensory shelf life study. Baseline survival function for different
models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177

6.1 Signal Tandmobiel study. Parametric AFT model based sur-

vival functions for emergence of permanent tooth 44 in gender
by DMF groups. . . . . . . . . . . . . . . . . . . . . . . . . . 184
6.2 Signal Tandmobiel study. Parametric AFT model based sur-
vival functions for emergence of permanent tooth 44 compared
to NPMLE. . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
6.3 Signal Tandmobiel study. PGM AFT models based estimated
error densities compared to a density of the standard normal
distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
6.4 Signal Tandmobiel study. PGM AFT models based estimated
survival functions for emergence of permanent tooth 44 com-
pared to NPMLE. . . . . . . . . . . . . . . . . . . . . . . . . 203
6.5 Signal Tandmobiel study. Mean-scale AFT model for emer-
gence of tooth 44 with a PGM error distribution. Estimated
density of the standardized error term compared to paramet-
ric densities. . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
 List of Figures xxiii

6.6 Signal Tandmobiel study. The mean-scale PGM AFT model

based estimated survival and hazard functions. . . . . . . . 212

7.1 Graphical representation of 4 bivariate interval-censored ob-

servations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
7.2 Artificial example with more regions of support than observa-
tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.3 An artificial data set with 6 observed rectangles and their
corresponding 4 regions of support . . . . . . . . . . . . . . 223
7.4 Density plots of Clayton, normal and Plackett copula . . . . 235
7.5 Signal Tandmobiel study. Density of penalized Gaussian mix-
ture model for emergence of permanent teeth 14 and 24. . . 242
7.6 Signal Tandmobiel study. The estimated cross-ratio function
for the maxillar first premolars for boys . . . . . . . . . . . . 249

8.1 Three cases of doubly interval-censored survival times. . . . 256

8.2 Mobile study. Forest plot showing the impact of gender, house-
hold size and age on the time to change phone . . . . . . . . 258
8.3 Signal Tandmobiel study. Biplot . . . . . . . . . . . . . . . . 277

9.1 Illustration of gamma process: Ten realizations of G(cH ∗ , c) 298

9.2 Illustration of the Dirichlet process: Ten realizations of
DP{c Weibull(1.5, 7)} . . . . . . . . . . . . . . . . . . . . . . 300
9.3 Homograft study. MCMC diagnostics . . . . . . . . . . . . . 313
9.4 Homograft study (aortic homograft patients). Nonparametric
Bayesian estimate of survival function . . . . . . . . . . . . 317
9.5 Homograft study. Semiparametric PH model Bayesian esti-
mates of survival functions . . . . . . . . . . . . . . . . . . . 320

10.1 Signal Tandmobiel study. Diagnostic plots of Bayesian analy-

sis using the SAS procedure LIFEREG . . . . . . . . . . . . . 334
10.2 Signal Tandmobiel study (boys, emergence of tooth 44). Pos-
terior densities of selected parameters, distribution of the
emergence time estimated using the classical Gaussian mix-
ture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
10.3 Breast cancer study of the radiotherapy-only group. Nonpara-
metric Bayesian estimate of survival function. . . . . . . . . 346
10.4 Signal Tandmobiel study. Estimated density, survival distribu-
tion and hazard function from DPMdencens of the emergence
time for tooth 44 of boys for two precision parameters . . . 348
10.5 Signal Tandmobiel study. Imputed emergence times for tooth
44 of boys from 2nd solution obtained from the R package
DPpackage together with observed intervals . . . . . . . . . 352

11.1 Breast cancer study. Diagnostic plots of Bayesian analysis us-

ing runjags. . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
xxiv List of Figures

11.2 Breast cancer study. Q-Q plots to contrast the ‘true la-
tent’ survival times with the ‘model-based replicated’ survival
times. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
11.3 Breast cancer study. PPCs corresponding to the range and
max gap test. . . . . . . . . . . . . . . . . . . . . . . . . . . 364
11.4 Breast cancer: Smooth and Weibull survival functions . . . . 371
11.5 Signal Tandmobiel study. Estimated piecewise constant dy-
namic regression coefficients obtained from dynsurv. . . . . . 374
11.6 Signal Tandmobiel study. Estimated piecewise constant base-
line hazard functions obtained from dynsurv. . . . . . . . . . 376
11.7 Signal Tandmobiel study. Frequency of jump points obtained
from dynsurv. . . . . . . . . . . . . . . . . . . . . . . . . . . 379

12.1 Signal Tandmobiel study. CGM AFT model based posterior

predictive error density compared to a density of the standard
normal distribution. . . . . . . . . . . . . . . . . . . . . . . . 393
12.2 Signal Tandmobiel study. CGM AFT model based poste-
rior predictive survival functions for emergence of permanent
tooth 44 compared to NPMLE. . . . . . . . . . . . . . . . . 394
12.3 Signal Tandmobiel study. CGM AFT model based posterior
predictive hazard functions for emergence of permanent tooth
44. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
12.4 Signal Tandmobiel study. PGM AFT model based posterior
predictive error density compared to a density of the standard
normal distribution. . . . . . . . . . . . . . . . . . . . . . . . 407
12.5 Signal Tandmobiel study. PGM AFT model based poste-
rior predictive survival functions for emergence of permanent
tooth 44 compared to CGM AFT based estimates. . . . . . 408
12.6 Signal Tandmobiel study. Semiparametric Bayesian AFT
model trace plots and marginal posterior densities. . . . . . 415
12.7 Signal Tandmobiel study. Posterior predictive error distribu-
tion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
12.8 Signal Tandmobiel study. Posterior predictive survival distri-
bution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418

13.1 Mastitis study. Log-normal frailty distribution with Weibull

baseline hazard: Estimated random effects and true survival
times. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
13.2 Signal Tandmobiel study. Estimated error (left) and random
effects density (right) for Norm-PGM and PGM-Norm models
to evaluate the emergence times of teeth 14, 24, 34 and 44. . 435
13.3 Signal Tandmobiel study. Estimated predictive incidence
curve and hazard function from PGM-PGM model to compare
emergence times of tooth 24 for boys with caries on primary
predecessor or not. . . . . . . . . . . . . . . . . . . . . . . . 436
 List of Figures xxv

13.4 Signal Tandmobiel study. Imputed emergence times of teeth

14 and 24 assuming a bivariate log-normal distribution. . . . 441
13.5 Signal Tandmobiel study. Contour plots of the estimated error
distributions for emergence times of teeth 14 and 24 and teeth
24 and 34. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
13.6 Signal Tandmobiel study. Emergence distributions for tooth
14 and tooth 24 for four covariate combinations. . . . . . . . 452
13.7 Mobile study. Histogram log(time) purchase 1st mobile phone
and log(gap time) between 2 purchases. . . . . . . . . . . . . 460
13.8 Mobile study. Estimated error density for time of first pur-
chase and estimated error density for gap time. . . . . . . . 462
13.9 Mobile study. Estimated incidence function to buy a new mo-
bile phone split up into age groups. . . . . . . . . . . . . . . 463

B.1 Three survival distributions. . . . . . . . . . . . . . . . . . . 495

F.1 Graph corresponding to the data presented in Figure 7.3 . . 537

F.2 Height map corresponding to the example data presented in
Figure 7.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 538
F.3 Truncated polynomial of degree 1 and B-splines of degree 3
on [0, 10]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 542
F.4 M-splines of order 2 and 3 . . . . . . . . . . . . . . . . . . . 544
Notation

S(t) survival function S

F (t) cumulative distribution function F
f (t) density function f (t)
}(t) hazard function }(t)
H(t) cumulative hazard function H(t)
[l, u] closed interval with lower limit l and upper limit u
(l, u] half-open interval with lower limit l and upper limit u
bl, uc open, half-open or closed interval with lower limit l and upper
limit u
δ censoring indicator
L(·) likelihood
`(·) log{L(·)}, log-likelihood
D collected data
DP Dirichlet process
Dir p (δ1 , . . . , δp ) p-dimensional Dirichlet distribution with parameters δ1 , . . . , δp
N (µ, σ 2 ) normal distribution with mean µ and variance σ 2
ϕ density of the standard normal Gaussian distribution N (0, 1)
ϕµ,σ2 density of the Gaussian distribution N (µ, σ 2 )
Φ standard cumulative distribution function of the Gaussian dis-
tribution N (0, 1)
Φµ,σ2 cumulative distribution function of the Gaussian distribution
N (µ, σ 2 )
Φρ standard bivariate cumulative distribution function of the
Gaussian distribution with correlation ρ
G(ζ, γ) gamma distribution with a shape parameter ζ and a rate pa-
rameter γ (with the mean ζ/γ)

xxvii
xxviii Notation

N p (µ, Σ) p-dimensional normal distribution with mean vector µ and

covariance matrix Σ
I(·) indicator function equal to 1 if the expression between paren-
theses is true, and 0 otherwise
lp penalized log-likelihood
∆k (·) k-order forward difference function
I identity matrix
I Hessian matrix
RanF range of the function F
C(u, v) copula
C̆(u, v) survival copula
C̆θC (u, v) Clayton copula with parameter θ
C̆ρG (u, v) Gaussian copula with parameter ρ
C̆θP (u, v) Plackett copula with parameter θ
k·k Euclidean length
Exploring the Variety of Random
Documents with Different Content
statement is one to make of the “importance,” for example, of such a
matter as the Academy soirée (as they say in London) of the
Philadelphia winter, the festive commemoration of some long span
of life achieved by the Pennsylvania Academy of Fine Arts? We may
have been thrilled, positively, by the occasion, by the interesting
encounters and discoveries, artistic and personal, to which it
ministered; we may have moved from one charmed recognition to
another, noting Sargents and Whistlers by the dozen, and old
forgotten French friends, foreign friends in general, older and
younger; noting young native upstarts, creatures of yesterday and to-
morrow, who invite, with all success, a stand and a stare; but no
after-sense of such vibrations, however lively, presumes to take itself
as communicable.
One would regret, on the other hand, failing to sound some echo of
a message everywhere in the United States so audible; that of the
clamorous signs of a hungry social growth, the very pulses, making
all their noise, of the engine that works night and day for a theory of
civilization. There are moments at which it may well seem that,
putting the sense of the spectacle even at its lowest, there is no such
amusement as this anywhere supplied; the air through which
everything shows is so transparent, with steps and stages and
processes as distinct in it as the appearance, from a street-corner, of
a crowd rushing on an alarm to a fire. The gregarious crowd “tells,”
in the street, and the indications I speak of tell, like chalk-marks, on
the demonstrative American black-board—an impression perhaps
never so much brought home to me as by a wondrous Sunday
morning at the edge of a vast vacant Philadelphia street, a street not
of Penn’s creation and vacant of everything but an immeasurable
bourgeois blankness. I had turned from that scene into a friendly
house that was given over, from top to toe, to a dazzling collection of
pictures, amid which I felt myself catch in the very act one of the
great ingurgitations of the hungry machine, and recognize as well
how perfect were all the conditions for making it a case. What could
have testified less, on the face of it, than the candour of the street’s
insignificance?—a pair of huge parted lips protesting almost to
pathos their innocence of anything to say: which was exactly, none
the less, where appetite had broken out and was feeding itself to
satiety. Large and liberal the hospitality, remarkably rich the store of
acquisition, in the light of which the whole energy of the keen
collector showed: the knowledge, the acuteness, the audacity, the
incessant watch for opportunity. These abrupt and multiplied
encounters, intensities, ever so various, of individual curiosity, sound
the æsthetic note sometimes with unprecedented shrillness and then
again with the most muffled discretion. Was the note muffled or
shrill, meanwhile, as I listened to it—under a fascination I fully
recognized—during an hour spent in the clustered palæstra of the
University of Pennsylvania? Here the winter afternoon seemed to
throw itself artfully back, across the centuries, the climates, the
seasons, the very faiths and codes, into the air of old Greece and the
age of gymnastic glory: artfully, I rather insist, because I scarce know
what fine emphasis of modernism hung about it too. I put that
question, however, only to deny myself the present luxury of
answering it; so thickly do the visitor’s University impressions, over
the land, tend to gather, and so markedly they suggest their being
reported of together. I note my palæstral hour, therefore, but because
it fell through what it seemed to show me, straight into what I had
conceived of the Philadelphia scheme, the happy family given up,
though quite on “family” lines, to all the immediate beguilements
and activities; the art in particular of cultivating, with such gaiety as
might be, a brave civic blindness.
I became conscious of but one excrescence on this large smooth
surface; it is true indeed that the excrescence was huge and affected
me as demanding in some way to be dealt with. The Pennsylvania
Penitentiary rears its ancient grimness, its grey towers and defensive
moats (masses at least that uncertain memory so figures for me) in
an outlying quarter which struck me as borrowing from them a vague
likeness to some more or less blighted minor city of Italy or France,
black Angers or dead Ferrara—yet seated on its basis of renown and
wrapped in its legend of having, as the first flourishing example of
the strictly cellular system, the complete sequestration of the
individual prisoner, thought wonderful in its day, moved Charles
Dickens to the passionate protest recorded in his American Notes. Of
such substance was the story of these battlements; yet it was
unmistakable that when one had crossed the drawbridge and passed
under the portcullis the air seemed thick enough with the breath of
the generations. A prison has, at the worst, the massive majesty, the
sinister peace of a prison; but this huge house of sorrow affected me
as, uncannily, of the City itself, the City of all the cynicisms and
impunities against which my friends had, from far back, kept plating,
as with the old silver of their sideboards, the armour of their social
consciousness. It made the whole place, with some of its oddly
antique aspects and its oddly modern freedoms, look doubly cut off
from the world of light and ease. The suggestions here were vast,
however; too many of them swarm, and my imagination must defend
itself as it can. What I was most concerned to note was the complete
turn of the wheel of fortune in respect to the measure of mere
incarceration suffered, from which the worst of the rigour had visibly
been drawn. Parts of the place suggested a sunny Club at a languid
hour, with members vaguely lounging and chatting, with open doors
and comparatively cheerful vistas, and plenty of rocking-chairs and
magazines. The only thing was that, under this analogy, one found
one’s self speculating much on the implied requisites for
membership. It was impossible not to wonder, from face to face,
what these would have been, and not to ask what one would have
taken them to be if the appearance of a Club had been a little more
complete. I almost blush, I fear, for the crude comfort of my prompt
conclusion. One would have taken them to consist, without
exception, of full-blown basenesses; one couldn’t, from member to
member, from type to type, from one pair of eyes to another, take
them for anything less. Where was the victim of circumstances,
where the creature merely misled or betrayed? He fitted no type, he
suffered in no face, he yearned in no history, and one felt, the more
one took in his absence, that the numerous substitutes for him were
good enough for each other.
The great interest was in this sight of the number and variety of
ways of looking morally mean; and perhaps also in the question of
how much the effect came from its being proved upon them, of how
little it might have come if they had still been out in the world.
Considered as criminals the moral meanness here was their
explication. Considered as morally mean, therefore, would possible
criminality, out in the world, have been in the same degree their sole
sense? Was the fact of prison all the mere fact of opportunity, and
the fact of freedom all the mere fact of the absence of it? One inclined
to believe that—the simplification was at any rate so great for one’s
feeling: the cases presented became thus, consistently, cases of the
vocation, and from the moment this was clear the place took on, in
its way, almost the harmony of a convent. I talked for a long time
with a charming reprieved murderer whom I half expected, at any
moment, to see ring for coffee and cigars: he explained with all
urbanity, and with perfect lucidity, the real sense of the appearance
against him, but I none the less felt sure that his merit was largely in
the refinement wrought in him by so many years of easy club life. He
was as natural a subject for commutation as for conviction, and had
had to have the latter in order to have the former—in the enjoyment,
and indeed in the subtle criticism, of which, as simple commutation
he was at his best. They were there, all those of his companions, I
was able to note, unmistakably at their best. One could, as I say,
sufficiently rest in it, and to do that kept, in a manner, the
excrescence, as I have called it, on the general scene, within bounds.
I was moreover luckily to see the general scene definitely cleared
again, cleared of everything save its own social character and its
practical philosophy—and at no moment with these features so
brightly presented as during a few days’ rage of winter round an old
country-house. The house was virtually distant from town, and the
conditions could but strike any visitor who stood whenever he might
with his back to the fire, where the logs were piled high, as made to
press on all the reserves and traditions of the general temperament;
those of gallantry, hilarity, social disposability, crowned with the
grace of the sporting instinct. What was it confusedly, almost
romantically, like, what “old order” commemorated in fiction and
anecdote? I had groped for this, as I have shown, before, but I found
myself at it again. Wasn’t it, for freedom of movement, for jingle of
sleigh-bells, for breasting of the elements, for cross-country drives in
the small hours, for crânerie of fine young men and high wintry
colour of muffled nymphs, wasn’t it, brogue and all, like some
audible echo of close-packing, chancing Irish society of the classic
time, seen and heard through a roaring blizzard? That at least, with
his back to the fire, was where the restless analyst was landed.
 X
BALTIMORE

I
It had doubtless not been merely absurd, as the wild winter
proceeded, to find one’s self so enamoured of the very name of the
South that one was ready to take it in any small atmospheric
instalment and to feel the echo of its voice in the yell of any engine
that happened not to drag one either directly North or directly West.
One tended at least, on these terms, in some degree, toward the land
where the citron blooms, and that was something to go on with, a
handful of small change accepted for the time as a pledge of great
gold pieces to come. It is astonishing, along the Atlantic coast, how,
from the moment the North ceases to insist, the South may begin to
presume; ever so little, no doubt, at first, yet with protrusive feelers
that tell how she only wants the right sensibility, the true waiting
victim, to play upon. It is a question certainly of where, on the so
frequently torpid stretch of shore I speak of, the North does cease to
insist; or perhaps I should more correctly say a question of when it
does. It appeared incapable of this fine tact almost anywhere, I
confess, at the season, the first supposedly relenting weeks, of my
facing in earnest to Florida; and the interest indeed of that slightly
grim adventure was to be in the way it ministered to the coincidence,
for me, of two quite opposed strains of reflection. On the one hand
nothing could “say” more to the subject long expatriated, condemned
by the terms of his exile to a chronic consciousness of grey northern
seas, than to feel how, from New York, or even from Boston, he had
but to sit still in his portentous car, had but to exercise a due
concentrated patience, in order to become aware, without personal
effort or suffered transfer, of that most charming of all watchable
processes, the gradual soft, the distinctively demoralized, conversion
of the soul of Nature. This conversion, if I may so put it without
profanity, has always struck me, on any southward course, as a
return, on the part of that soul, from a comparatively grim Theistic
faith to the ineradicable principle of Paganism; a conscious casting-
off of the dread theological abstraction—an abstraction still, even
with all Puritan stiffening—in the interest of multiplied, lurking,
familiar powers; divinities, graces, presences as unseen but as
inherent as the scents clinging to the folds of Nature’s robe. It would
be on such occasions the fault of the divine familiars themselves if
their haunts and shrines were empty, for earth and air and day and
night, as we go, still affect us as moods of their sympathy, still vibrate
to the breath of their passage; so that our progress, under the
expanding sun, resembles a little less a journey through space than a
retracing of the course of the ages.
These are fine fancies, however, and what is more to my point is
that the theory (so agreeable to entertain at Jersey City) of a direct
connection between the snow-banks and the orange-groves is a thing
of sweetness only so long as practically unshaken. There is
continuity, goodness knows, always in America—it is the last thing
that is ever broken: the question for the particular case is but
continuity of what? The basis of my individual hope had been that of
the reign of the orange-grove; but what it proved, at the crisis I
name, was positively that of the usurpation of the snow-bank. It was
possible, indubitably, in such conditions, to go to Charleston on
sledges—which made in fact, after all, for directness of connection. It
made moreover, by the same token, for a certain sinister light on the
general truth of our grand territorial unity. It was as if the winter, at
the end of February, abroad for a walk, had marched as promptly
and inevitably from the Arctic Circle to the Gulf as it might have
proceeded, with pride in its huge clear course, from the top of
Broadway to the Battery. This brought home again, as I myself went,
I remember, one of those three or four main ideas, suggested by the
recurrent conditions, which become as obsessions for the traveller in
the States—if he have a mind, that is, so indecently exposed to ideas:
the sense, constantly fed, and from a hundred sources, that, as
Nature abhors a vacuum, so it is of the genius of the American land
and the American people to abhor, whenever may be, a
discrimination. They are reduced, together, under stress, to making
discriminations, but they make them, I think, as lightly and scantily
as possible. With the lively insistence of that impression, even
though it quite undermined my fond view of a loose and
overreaching citronic belt, I found my actually monotonous way
beguiled. Practically, till I reached Charleston, this way, disclaiming
every invidious intent, refused to be dissociated from anything else
in the world: it was only another case of the painting with a big
brush, a brush steeped in crude universal white, and of the colossal
size this implement was capable of assuming. Gradations,
transitions, differences of any sort, temporal, material, social,
whether in man or in his environment, shrank somehow, under its
sweep, to negligible items; and one had perhaps never yet seemed so
to move through a vast simplified scheme. The illustration was once
more, in fine, of the small inherent, the small accumulated
resistance, in American air, to any force that does simplify. One
found the signs of such resistance as little in the prospect enjoyed
from the car-window as one distinguished them in the vain images of
the interior; those human documents, deciphered from one’s seat in
the Pullman, which yet do always, in their way, for the traveller,
constitute precious evidence. The spread of this single great wash of
winter from latitude to latitude struck me in fact as having its
analogy in the vast vogue of some infinitely-selling novel, one of
those happy volumes of which the circulation roars, periodically,
from Atlantic to Pacific and from great windy State to State, in the
manner, as I have heard it vividly put, of a blazing prairie fire; with
as little possibility of arrest from “criticism” in the one case as from
the bleating of lost sheep in the other. Everything, so to speak, was
monotonized, and the whole social order might have had its nose, for
the time, buried, by one levelling doom, in the pages that, after the
break of the spell, it would never know itself to mention again. Of
course, one remembered meanwhile, there were spells and spells,
and the free field—the particular freedom of which is the point of my
remark—would on occasion be just as open to the far-exhaled breath
of the South. That in fact is what I was to find it—though I thought all
delightfully—later in the season, when the freedom of the field struck
me as pure benefit. I was not, at the end of February, really to meet it
(as I had looked for it) before crossing the Florida line; but toward
the middle of June I was to meet it, enchantingly, at Baltimore, and
this, then, as I had not stopped there in my previous course, was,
even beyond the wondrous February Florida, to reveal to me, grateful
for any such favour, the South in her freshness. The freshness was in
part, no doubt—and even perhaps to extravagance—mine; I testify at
all events first for Baltimore.
It would probably be again the freshness, of this confessedly
subjective sort, it would probably be again the state of alert response
to any favour of the class just hinted at; but the immediate effect of
the Maryland capital was to place it, to my troubled vision, and quite
at the head of its group, in a category of images and memories small
at the best and the charm of which casts a shadow, none the less,
even as the rose wears a thorn. I refer indeed in this slightly
portentous figure to the mere familiar truth that if representative
values and the traceable or the imaginable connections of things
happen to have, on occasion, for your eyes and your intelligence, an
existence of any intensity, your case, as a traveller, an observer, a
reporter, is “bound” from the first, under the stirred impression, to
loom for you in some distressful shape. These representative values
and constructive connections, the whole of the latent vividness of
things, not only remain, under expression, subject to no definite
chemical test, no mathematical proof whatever, but almost turn their
charming backs and toss their wilful heads at one’s poor little array
of terms and equivalents. There thus immediately rises for the lone
visionary, betrayed and arrested in the very act of vision, that spectre
of impotence which dogs the footsteps of perception and whose
presence is like some poison-drop in the silver cup. Baltimore put on
for me, from the first glance, the form of the silver cup filled with the
mildest, sweetest decoction; but I had no sooner begun to taste of it
than I began to taste also of the infused bitter. It had, in its way,
during that first early hour or two of the summer evening, a perfect
felicity: which meant, for the touched intelligence, that it was full of
pleasantly-playing reference and reflection, that it exhaled on the
spot, as the word goes, an atmosphere; that it wore, to
contemplation, in fine, a character as marked with mild accents as
some faded old uniform is marked with tarnished buttons and braid
—albeit these sources of interest were too closely of the texture to be
snipped off, in the guise of patterns or relics, by any mere sharp
shears of journalism.
 I arrived late in the day, and the day had been lovely; I alighted at
a large fresh peaceful hostelry, imposingly modern yet quietly
affable, and, having recognized the deep, soft general note, even from
my windows, as that of a kind of mollified vivacity, I sought the
streets with as many tacit questions as I judged they would tolerate,
or as the waning day would allow me to put. It took but that hour, as
I strolled in the early eventide, to give me the sense of the
predicament I have glanced at; that of finding myself committed to
the view of Baltimore as quite insidiously “sympathetic,” quite
inordinately amiable—which amounted, in other words, to the
momentous proposition that she was interesting—and still of
wondering, by the same stroke, how I was to make any such
statement plausible. Character is founded on elements and features,
so many particular parts which conduce to an expression. So I
walked about the dear little city looking for the particular parts—all
with the singular effect of rather failing to find them and with my
impression of felicity at the same time persistently growing. The
felicity was certainly not that of a mere blank; there must accordingly
have been items and objects, signs and tokens, there must have been
causes of so charming a consequence; there must have been the little
numbers (not necessarily big, if only a tall enough column) for the
careful sum on my slate. What happened then, remarkably, was that
while I mechanically so argued my impression was fixing itself by a
wild logic of its own, and that I was presently to see how it would,
when once settled to a certain intensity, snap its fingers at warrants
and documents. If it was a question of a slate the slate was used, at
school, I remembered, for more than one purpose; so that mine, by
my walk’s end, instead of a show of neat ciphering, exhibited simply
a bold drawn image—which had the merit moreover of not being in
the least a caricature. The moral of this was precious—that of the fine
impunity with which, if one but had sensibility, the ciphering could
be neglected and in fact almost contemned: always, that is (and only)
with one’s finer wits about one. Without them one was at best, really,
nowhere—even with “items” by the thousand; so that the place
became, quite adorably, a lesson in the use of that resource. It would
be “no good” to a journalist—for he is nowhere, ever, without his
items; but it would be everything, always, to the mere restless
analyst. He might by its aid stand against all comers; and this alike in
pleasure and in pain, in the bruised or in the soothed condition. That
was the real way to work things out, and to feel it so brought home
would by itself sufficiently crown this particular small pilgrimage.
 II
If my sensibility yielded so completely to Baltimore, however, I
should add, this was no doubt partly because the air seemed from the
first to breathe upon it a pledge of no bruises. I mounted, in the
golden June light, the neatest, amplest, emptiest street-vista, the
builded side of a steepish hill, and, having come in due course to a
spacious summit, laid out with monumental elegance and completely
void, for the time, of the human footstep, I saw that to suffer in any
fibre I should have positively, somewhere, to hurl myself upon the
spears. Not a point protruded then or afterwards; and the cunning of
the restless analyst is essentially such that, with friction long enough
in abeyance to leave him a start, he is already astride of his happier
thesis, seated firm, having “elected” to be undismountable, and
riding it as hard as it will go. The absence of friction, on my
monumental hilltop and in the prospects it overhung, constituted, I
was to find, an absolute circus-ring for this exercise; and it is much
to be able to say, while performing in the circus (even if but mainly to
the public of one’s own conscience), that one has never had the sense
of a safer hour. The safety of Baltimore, I should indeed mention,
consisted perhaps a little overmuch, during that first flush, in its
apparently vacant condition: it affected me as a sort of perversely
cheerful little city of the dead; and from the dead, naturally, comes
no friction. Was it cheerful, that is, or was it only resigned and
discreet?—with the manner of the good breeding that doesn’t
publicly prate of family troubles. I found myself handling, in
imagination, these large quantities only because, as I suppose, it was
impossible not to remember on that spot of what native generation
one had come. It took no greater intensity of the South than
Baltimore could easily give to figure again, however fadedly, and all
as a ghostly presence, the huge shadow of the War, and to reproduce
that particular bloodstained patch of it which, in the very first days,
the now so irresponsible and absent community about me had flung
across the path of the North. This one echo of old Time made the
connections, for the instant, all vibrate, and the scene before me,
somehow, as it stood, had to account for the great revolution. It was
as if that, for the restless analyst, had to be disposed of before
anything else: whereby, precisely, didn’t the amenity of his
impression partly spring from the descent there, on the spot, in a
quick white flash, of the most august of the Muses? It was History in
person that hovered, just long enough for me to recognize her and to
read, in her strange deep eyes, her intelligence at least of everything.
It might have been there fairly as reassurance. “Yes, they have lived
with me, and it has done them good, and we have buried together all
their past—about which, wise creature as I am, I allow them, of
course, all piety. But this—what you make out around us—is their
real collective self, which I am delighted to commend to you. I’ve
found Baltimore a charming patient.” That was, in ten minutes, what
it had come to; as if the brush of the sublime garment had by itself
cleared the air. If there was a fine warm hush everywhere it was
indeed partly that of this historic peace.
But for the rest it only meant that the world was at such a season
out of town. Houses were everywhere closed, and the neat
perspectives, all domiciliary and all, as I have hinted, tending mildly
to a vague elegance, were the more neat and more elegant, though
doubtless also the more mild and the more vague, for their being so
inanimate. A certain vividness of high decency seemed in spite of it
to possess them, and this suggestion of the real southern glow, yet
with no southern looseness, was clearly something by itself—all
special and local and all, or almost all, expressed in repeated vistas of
little brick-faced and protrusively door-stepped houses, which,
overhung by tall, regular umbrage, suggested rows of quiet old ladies
seated, with their toes tucked-up on uniform footstools, under the
shaded candlesticks of old-fashioned tea-parties. The little ladylike
squares, though below any tide-mark of fashion, were particularly
frequent; in which case it was as if the virtuous dames had drawn
together round a large green table, albeit to no more riotous end than
that each should sit before her individual game of patience. One
sounds inevitably the note of the “virtue”—so little, in general, can
any picture of American town-appearance hang together without it.
It amounts, everywhere, to something intenser than the implied
absence of “vice”; it amounts to a sort of registered absence of the
conception or the imagination of it, and still more of the provision
for it; though, all the while, as one goes and comes, one feels that no
community can really be as purged of peccant humours as the typical
American has for the most part found itself foredoomed to look. It
has been caught in the mechanism of that consistency—to an effect of
convenience, doubtless, much more than to any other; and has thus,
in the whole vast connection, a relation to appearances that is all its
own. The “European” scene, at a thousand points, looks all its
sophistications straight out at us—or looks, in other words, at least as
perverse as it practically is. The American, on the other hand,
expressing physiognomically no sophistications at all—though plenty
of quite common candours, crudities and vulgarities—makes one ask
if the cash-register, the ice-cream freezer, the lightning-elevator, the
“boys’ paper,” and other such overflows, do truly represent the sum
of its passions. Incontestably, at all events, this immensely
ingenuous aspect counts, for any country and any scheme of life, as a
great force, just as the appearance of the stale and the congested
residing in the comparatively battered mask of experience counts as
a weakness: to conceive which the mind’s eye has only to fix a little
the colossal American face grimacing with anything of a subtler
consciousness. That image, if actually presented, would become, as
we feel, appalling. The inexorable fate of the countenance in question
may be so to learn to grimace in time, but though few processes are
slow, in the United States, and few exhibitions not contagious, any
such transition, assuredly, will not be rapid, any more than any such
tendency will easily predominate.
All of which would have carried me far from the simple sweetness
of Baltimore, were it not that, for the restless analyst, there is no
such thing as an unrelated fact, no such thing as a break in the chain
of relations. Many a perceived American aspect, for that matter,
would by itself have little to give; the student of manners, in other
words, to make it presentable—by which I understand to make it
sufficiently interesting—must first discover connections for it and
then borrow from these, if possible, the elements of a wardrobe. And
though it should sound a little monstrous, moreover, one had
somehow not been prepared for so delicate an effect of propriety;
since there are cases too, indubitably, in which propriety can show
for almost as coarse as anything else. It couldn’t have been, either,
that one had expected any positive air of licence; but the fact was, I
suppose, that, for a constitutional story-seeker, a certain still, small
shock, a prompt need of readjustment of view, was involved in one’s
finding the element of the bourgeois crop up, so inveterately, in
latitudes generally associated, so far as one knew them elsewhere,
with some perceptible sacrifice to the sway of the senses. I had
already, at this date, as I have noted, dipped deep into our own
uttermost South, and had there had to reckon with that first slight
disconcertment awaiting the observer whose southern categories
happen to have been wholly European. His simplest expression for
the anomaly he meets is that he sees the citronic belt all
incongruously Protestantized: that big word (for so small a
bewilderment perhaps) sticks to him and worries him—almost as
absurdly, I grant, as if he had expected Charleston and Savannah to
betray the moral accent of Naples or Seville. He had not, assuredly,
done this; but he had as little allowed, in imagination, for the
hyperborean note. A South without church-fronts and church-
interiors had been superficially as strange, in its way, as a
Methodism of the sub-tropic night, a Methodism of the orange and
the palm. Such were the treacheries of association; though what
indeed would observation be, for interest, if it were not, just by these
armed surprises, constantly touched with adventure? The beauty of
Baltimore was, all this time, that one could feel it as potentially
harmonizing; the citronic belt would not embrace here more
Methodism than might consort with it, nor the Methodism pretend
to cultivate with any success the hibiscus and the pomegranate.
That I could entertain so many incoherent ideas in half-an-hour
was in any case a proof that I felt, for the occasion, left in possession;
quite as the visitor as yet unintroduced may feel during some long
preliminary wait in a drawing-room. He looks at the furniture,
pictures, books; he studies in these objects the character of the house
and of his hosts, and if there be some domestic treasure visibly more
important and conspicuous than the others, it engages his attention
as either with a fatal or an engaging force. The top of the central
eminence, with its air of an ample plan and of sweeping the rest of
the circle, figured the documentary parlour and my enjoyed leave to
touch and examine; so that when it was a question, in particular, of
the monument to Washington, the high column, in the middle, with
its surmounting figure and its spreading architectural base, this
presence was, for all the world, like that of some vast and stately old-
fashioned clock, a decorative “piece,” an heirloom from generations
now respectably remote, occupying an inordinate space in
proportion to the other conveniences. The ornamental, the
“important” clock is apt to be in especial, at such a crisis, a tell-tale
object; its range of testimony, of possible treachery, is immense, and
cases are not unknown, I gather, in which it has put the doubting
visitor to flight. The greater the felicity, thereby, for the overtopping
Baltimore timepiece, which hung about in mild reassurance,
promptly aware that it wasn’t a bit vulgar, but, on the contrary, of a
pleasant jejune academic pomp that suggested to the fancy some
melancholy, some spectral, man-at-arms mounting guard at the
angles, in due military form, over suspected treasures of Style. One
could imagine, somehow, under the summer stars, the mystic vigil of
these mild heroes; and one could above all catch again the
interesting hint of the terms on which, in the United States, the
consecration of time may be found operating. It has a trick there all
of its own, thanks to which the effect of duration is produced very
much as, before the footlights, the prestidigitator produces the effect
of extracting a live fowl from a hat. This is a law under which, the
material permitting, the decades count as centuries and the centuries
as æons. The misfortune is that too often the material, futile and
treacherous, doesn’t permit. Yet the law is in the happiest cases none
the less strikingly vindicated. There, for instance—to pursue
undiscouraged my figure of the guest in the empty parlour—were the
best houses, the older, the ampler, the more blandly quadrilateral;
which in spite of their still faces met one’s arrest, at their
commodious corners and other places of vantage, with an
unmistakable manner. The quiet assurance of a position in the world
—the world, the only one, with which they were concerned—testified
again, in an interesting way, to the simple source of their
impressiveness, showing how almost any modern interval could have
been long enough to make them nobly antique if such interval might
only have been vulgar enough. The age of “brown stone” was to have
found no difficulty in that; the prolongation of its rage for a quarter
of a century amply sufficed to dignify every antecedent thing it had
spared (as the survivors of reigns of Terror grow by mere survival
distinguished); while, steeped in dishonour up to the eyebrows, that
is up to its false cornices of painted and sanded wood and iron, it was
never to enjoy, for itself, the advantage it elsewhere conferred.
Nothing has ever been vulgar enough to rehabilitate the odd
ugliness, so distinct, yet after all so undemonstrable, of this luckless
material; the way one shuddered, in particular, at the touch, on
balustrade and elsewhere, of the sanded iron! It has been followed by
other rages and other errors, but even the grace of the American
time-measure can do nothing for it.
 III
It was of course the fact that the “values” here were all such, and
such alone, as might be reflected from the social conditions and the
state of manners, even if reflected, for the hour, almost into empty
space—it was this that gave weight to each perceived appearance and
permitted none to show as trivial enough to project me, in reaction
or in inanition, upon the comparative obviousness of the “burnt
district.” There is almost always a burnt district to eke out the
interest of an American city—it is the pride of the citizen and the
resource of the visitor when all else fails; and I can scarce, I think,
praise Baltimore so liberally as to note that this was the last of her
beauties I was conscious of. She had lost by fire, a few months before,
the greater part of her business quarter, which she was now rapidly
and artfully calling back to existence; but the entertainment she
offered me was guiltless, ever so gracefully and gallantly guiltless, as
it struck me, of reference, even indirect, to the majesty either of ruin
or of remedy. One was, on further acquaintance, thoroughly
beguiled, but the burnt district had so little to do with it that the days
came and went without my so much as discovering its whereabouts.
Wonderful little Baltimore, in which, whether when perched on a
noble eminence or passing from one seat of the humanities, one seat
of hospitality, to another—a process mainly consisting indeed, as it
seemed to me, of prompt drives through romantic parks and
woodlands that were all suburban yet all Arcadian—I caught no
glimpse of traffic, however mild, nor spied anything “tall” at the end
of any vista. This was in itself really a benediction, since I had
nowhere, from the first, been infatuated with tallness; I was
infatuated only with the question of manners, in their largest sense—
to the finer essence of which tallness had already defined itself to me
as positively abhorrent. What occurred betimes, and ever so happily,
was simply that the delicate blank of those first hours flushed into
animation, and that with this indeed the embroidery of the fine
canvas turned thick and rich. It came back again, no doubt, in the
inveterate way, to the University presence, and to the eagerness with
which, on the American scene, as I tire not, you see, of repeating, the
visiting spirit, on such occasions, throws itself straight into
sanctuary. It breaks in at any cost, this distracted appetite, and,
recomposing the elements to their greater distinction, if need be, and
with a high imaginative hand, makes of the combination obtained
the only firm standpoint for the rest of the view. It has even in this
connection an occasional sharp chill; air-borne rumours reach it of
perversities and treacheries, conspiracies possibly hatching in the
very bosom of the temple and against its very faith. One hears of the
University idea threatened in more than one of the great institutions
—reduced to some pettifogging conception of a short brisk term and
a simplified culture; a lively thrifty training for “business-
competition.” This is a blow to the collective fond fancies set
humming, at once, in almost any scholastic shade—under the effect
of which one can but give one’s own scant scholar’s hood, while one
winces, a further protesting pull over abashed brows. It would have
been a question, very much, of what I call breaking-in (into the
Johns Hopkins) at this moment, had I not here been indulged, in all
liberality, with an impression the more charming, in a manner, for
the fact of halls and courts brooding in vacation stillness. Perversely
adorable always—and I scarce know why—the late afternoon light in
deserted haunts of study; with the secret of supreme dignity lurking,
above all, in high, dusky, wainscoted chambers where the sound of
one’s footfall lingers, to one’s pleasure, like a caress, and where
portraits of the appurtenant worthies, the heroes and patrons, grow
vague in the twilight. It is a tribute to the forces of idealism lurking
again and again, over the country, in the amenity of the general
Collegiate appearance, that the last thing these conditions overtly
suggest, or seem to accept as their imputed virtue, is this
precipitation of the young intelligence into the mere vociferous
market.
I scarcely know why, however, I should have appeared, even by
waving it away, to make room at our banquet for the possible
skeleton of the false, the barbarizing, note; since the natural pitch of
Baltimore, the pictorial, so to speak, as well as the social, struck me,
once a certain contact established, as that of disinterested sensibility,
the passion of which her University is the highest and clearest
example. There was on the splendid Sunday in particular a warm,
soft fusion of aspects—a confusion, in fact, while I now gather it in—
which seems to defy, though all unconsciously, the sharper edge of
discrimination and to offer itself, insistently, as a general wash of
brave Southern shade, the play of a liquid brush of which the North
knows nothing. The episodes melt together, yet they also, under a
little pressure, come happily apart, and over the large sun-chequered
picture the generous boughs hang heavy. Admirable I found them,
the Maryland boughs, and so immediately disposed about the
fortunate town, by parkside and lonely lane, by trackless hillside and
tangled copse, that the depth of rural effect becomes at once
bewildering. You wonder at the absent transitions, you look in vain
for the shabby fringes—or at least, under my spell, I did; you have
never seen, on the lap of nature, so large a burden so neatly
accommodated. Baltimore sits there as some quite robust but almost
unnaturally good child might sit on the green apron of its nurse, with
no concomitant crease or crumple, no uncontrollable “mess,” by the
nursery term, to betray its temper. It was with something like that
figure before me that I kept communing, as I say, with the bland
presence. Even a morning hour or two at the great University
Hospital—for one’s experience of the higher tone, one’s irrepressible
pursuit of charm, in America, has, to its great enrichment, these odd
sequences—even that beginning of the day did nothing to obtrude the
ugly or to overemphasize the real; it simply contributed, under some
perversion that I can neither explain nor defend, to the general grace
of the picture. Why should the great Hospital, with its endless
chambers of woe, its whole air as of most directly and advisedly
facing, as the hospitals of the world go, the question of the
immensities of pain—why should such an impression actually have
turned, under the spell, to fine poetry, to a mere shining vision of the
conditions, the high beauty of applied science? The conditions,
positively, as I think of them after the interval, make the poetry—the
large art, above all, by which, in a place bristling with its terrible tale,
everything was made to seem fair, and fairest even while it most
intimately concurred in the work. In short if the Hospital was
fundamentally Universitarian—as of the domain of the great Medical
Faculty—so it partook for me, in its own way, of the University
glamour, and so the tempered morning, and the shaded splendour,
and the passive rows, the grim human alignments that became, in
their cool vistas, delicate “symphonies in white,” and, more even
than anything else, the pair of gallant young Doctors who ruled, for
me, so gently, the whole still concert, abide with me, collectively, as
agents of the higher tone.
No example could speak more of that enlargement of function, for
constituting some picture of life, which many an American element
or object, many an institution, has to be felt as practising—usually
with high success. It comes back, one notes for the thousandth time,
to that redistribution and reconsecration of values, of representative
weight, which it is the interesting thing, over the land, to see take
effect—to see in special take all the effect of which it is capable. There
are a thousand “European” values that are absent, and, whether as a
consequence or not of that, there are innumerable felt solutions of
the social continuity. The instinct of missing—by which I mean not at
all either the consciousness or the confession of lacking—keeps up,
however, its own activity; for the theory at least of the native spirit is
to consent wittingly to no privation. It has a genius, the native spirit,
for desiring things of the existence, and even of the possibility of
which it is actually unaware, and it views the totality of nature and
the general life of man, I think, as more than anything else
commissioned and privileged to wait on these awakenings. Thus new
values arise as expansion proceeds; the marked character of which,
for comparative sociology, is that they are not at all as other values.
What they “count” for is the particular required American quantity;
and we see again and again how large a quantity symbol and figure
have to represent. The interesting thing is that, on the spot, the
representation does practically cover the ground: it covers elements
that in communities employing a different scale require for their
expression (and perhaps sometimes to an effect of waste) a much
greater number of terms. Hence the constant impression of elasticity,
and that of those pressures of necessity under which value and
virtue, character and quantity, greatness and glory even, to a
considerable extent, are imputed and projected. There has to be a
facility for the working of any social form—facility of comparison and
selection in some communities, facility of rapid conversion in others
That is where the American material is elastic, where it affects one,
as a whole, in the manner of some huge india-rubber cloth fashioned
for “field” use and warranted to bear inordinate stretching.
One becomes aware thus wherever one turns, both of the tension
and of the resistance; everything and every one, all objects and
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebooknice.com