Statistics for Health Data Science: An Organic

Approach 1st Edition Ruth Etzioni latest pdf 2025

Order directly from textbookfull.com

https://textbookfull.com/product/statistics-for-health-data-science-
an-organic-approach-1st-edition-ruth-etzioni/

★★★★★
4.7 out of 5.0 (85 reviews )

PDF Available Immediately

 Statistics for Health Data Science: An Organic Approach 1st
Edition Ruth Etzioni

TEXTBOOK

Available Formats

■ PDF eBook Study Guide Ebook

EXCLUSIVE 2025 ACADEMIC EDITION – LIMITED RELEASE

Available Instantly Access Library

More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Practical Statistics for Educators 6th Edition Ruth

Ravid

https://textbookfull.com/product/practical-statistics-for-
educators-6th-edition-ruth-ravid/

Organic Chemistry Concepts An EFL Approach 1st Edition

Gregory Roos

https://textbookfull.com/product/organic-chemistry-concepts-an-
efl-approach-1st-edition-gregory-roos/

Organic Chemistry: An Acid-Base Approach, Third Edition

Smith

https://textbookfull.com/product/organic-chemistry-an-acid-base-
approach-third-edition-smith/

Statistics Using Stata An Integrative Approach Sharon

Lawner Weinberg

https://textbookfull.com/product/statistics-using-stata-an-
integrative-approach-sharon-lawner-weinberg/
Computational Network Science An Algorithmic Approach
1st Edition Hexmoor

https://textbookfull.com/product/computational-network-science-
an-algorithmic-approach-1st-edition-hexmoor/

Quantum Chemical Approach for Organic Ferromagnetic

Material Design 1st Edition Yuriko Aoki

https://textbookfull.com/product/quantum-chemical-approach-for-
organic-ferromagnetic-material-design-1st-edition-yuriko-aoki/

General Organic and Biological Chemistry An Integrated

Approach 4th Edition Kenneth W. Raymond

https://textbookfull.com/product/general-organic-and-biological-
chemistry-an-integrated-approach-4th-edition-kenneth-w-raymond/

Doing Data Science in R An Introduction for Social

Scientists 1st Edition Mark Andrews

https://textbookfull.com/product/doing-data-science-in-r-an-
introduction-for-social-scientists-1st-edition-mark-andrews/

Practical Statistics for Data Scientists 2nd Edition

Peter Bruce

https://textbookfull.com/product/practical-statistics-for-data-
scientists-2nd-edition-peter-bruce/
Springer Texts in Statistics

Ruth Etzioni
Micha Mandel
Roman Gulati

Statistics
for Health
Data Science
An Organic Approach
Springer Texts in Statistics

Series Editors
G. Allen, Department of Statistics, Houston, TX, USA
R. De Veaux, Department of Mathematics and Statistics, Williams College,
Williamstown, MA, USA
R. Nugent, Department of Statistics, Carnegie Mellon University, Pittsburgh,
PA, USA
Springer Texts in Statistics (STS) includes advanced textbooks from 3rd- to 4th-year
undergraduate courses to 1st- to 2nd-year graduate courses. Exercise sets should be
included. The series editors are currently Genevera I. Allen, Richard D. De Veaux,
and Rebecca Nugent. Stephen Fienberg, George Casella, and Ingram Olkin were
editors of the series for many years.

More information about this series at http://www.springer.com/series/417

Ruth Etzioni • Micha Mandel • Roman Gulati

Statistics for Health Data

Science
An Organic Approach
Ruth Etzioni Micha Mandel
Fred Hutchinson Cancer Research Center Department of Statistics and Data Science
University of Washington Hebrew University of Jerusalem
Seattle, WA, USA Mount Scopus, Jerusalem, Israel

Roman Gulati
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
Seattle, WA, USA

ISSN 1431-875X ISSN 2197-4136 (electronic)

Springer Texts in Statistics
ISBN 978-3-030-59888-4 ISBN 978-3-030-59889-1 (eBook)
https://doi.org/10.1007/978-3-030-59889-1

© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2020

This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Dedicated to the memory of Steve
and Joyce Fienberg
Preface

Today’s health services and health outcomes research trainees are entering a
data-driven world, full of opportunities for novel research to address questions both
old and new. But, this work comes with an unprecedented demand—for competence
in health data science.
Health data science is multidisciplinary, encompassing statistics, health econo-
metrics, and, now, predictive analytics. The ever-expanding universe of health data
includes clinical, administrative, and population data. Each is subject to its own
set of limitations, requiring researchers to be able to understand the implications for
analysis and adapt accordingly. It is no wonder that cultivating competence in health
data science requires covering many bases.
This text grew out of a curriculum for a core methods course offered to PhD
students in health services and health outcomes at the University of Washington.
Over the years, the course has also attracted students from epidemiology and global
health as well as many clinician scientist trainees. Graduates of the course have
told us many times that it was the most useful of their PhD classes and greatly
facilitated their dissertation research. Recognizing the usefulness and uniqueness of
the curriculum inspired us to create this text.
Designed for researchers who have had some exposure to the basics of statistical
analysis (perhaps a first course on linear regression), this text brings together
key statistical ideas that are foundational for contemporary investigators in health
services, health outcomes, and health policy. We cover the classical paradigm of
statistical inference, exploring regression modeling for a variety of health outcomes,
where the goal is to explain. We also cross over to problems where the goal is
to predict, and we discuss statistical learning and predictive analytics. We present
various analyses of association appropriate for observational data, and we also
study methods for causal inference. We discuss sampling and variability and how
they manifest in the analysis of population surveys. In addition, we introduce
bootstrapping as a tool for measuring variability in complex settings, such as the
two-part model for health expenditure outcomes.
We call our pedagogical mindset an organic approach to convey our intent
to grow understanding in a manner that feels natural. Ideally, each new concept

vii
viii Preface

should feel like an extension of what has come before. Our explanations are mostly
non-technical and are supported by relevant stories and examples. The methods
presented are applied to real-world research questions using real-world data. We
explore the trends in body mass in the United States using NHANES data in
Chaps. 3, 4, and 10; examine racial disparities in prostate cancer mortality using the
SEER cancer registry data in Chap. 5; and quantify incremental health care costs
associated with diabetes using MEPS data in Chaps. 6 and 9.
Although this text spans a wide range of ideas, models, and methods, in many
ways it only scratches the surface. For each chapter, we could direct the reader to
any of a number of books covering just the topic of that chapter. Moreover, there are
plenty of topics that fall outside of our scope, which focuses on cross-sectional data
and independent observations. Rather than being exhaustive, our goal is to provide
a useful and practical foundation that students can build upon as they competently
and confidently navigate the world of health data.
We thank those who encouraged and supported us in the preparation of this work.
We appreciate all the students who read early chapters and encouraged us to keep
writing, especially Samantha Clark and Becca Hutcheson. Noelle Noble helped with
early production and with fuel toward the end. Special thanks are due to our students,
Bailey Ingraham and Ernesto Ulloa-Pérez, who co-authored Chaps. 1, 2, and 4, and
Marlena Bannick. We are also grateful to Daniel Nevo, Noah Simon, and David
Etzioni, who read and critiqued drafts of several chapters and helped us to make
them much stronger.
We thank the Department of Health Services at the University of Washington
for the opportunity to teach Advanced Research Methods and to Steve Zeliadt and
Paul Hebert, co-instructors and health services researchers from whom we learned
how hard it is to define a health care cost, and who first taught us the term recycled
predictions; to Michelle Garrison, current co-instructor and fantastic educator; and
to Paula Diehr from whom one of us (RE) inherited the course and who shared the
Basic Health Plan Data used in Chap. 5.
This book is dedicated to the memory of Steve and Joyce Fienberg. Steve was a
giant in the field, and his vision, leadership, and impact remain constant models for
us and for many statisticians. Joyce was taken from us abruptly and cruelly in the
Tree of Life synagogue attack in Pittsburgh on October 27, 2018. She was the one
who welcomed and embraced Steve’s international students, among whom was one
of us (RE). Her warmth and generosity knew no bounds.
May their memory be a blessing.

Seattle, WA, USA Ruth Etzioni

Jerusalem, Israel Micha Mandel
Seattle, WA, USA Roman Gulati
Contents

1 Statistics and Health Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Statistics and Organic Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Statistical Methods and Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Health Care Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4.1 Medical Claims . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4.2 Medical Records . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4.3 Health Surveys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.4.4 Disease Registries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.5 Outline of the Text . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.6 Software and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2 Key Statistical Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1 Samples and Populations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 Statistics Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.1 Random Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.2 Dependent and Independent Variables. . . . . . . . . . . . . . . . . . . . . 19
2.2.3 Statistical Distributions and Their Summaries . . . . . . . . . . . . 19
2.2.4 Parameters and Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.5 Estimation and Inference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.6 Variation and Standard Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.2.7 Conditional and Marginal Means . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2.8 Joint and Mixture Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2.9 Variable Transformations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3 Common Statistical Distributions and Concepts . . . . . . . . . . . . . . . . . . . . 25
2.3.1 The Bernoulli and Binomial Distributions for
Binary Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.2 The Multinomial Distribution for Categorical Outcomes. 27
2.3.3 The Poisson and Negative Binomial Distributions
for Counts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

ix
x Contents

2.3.4 The Normal Distribution for Continuous Outcomes . . . . . . 30

2.3.5 The Gamma and Lognormal Distributions
for Right-Skewed Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.4 Hypothesis Testing and Statistical Inference. . . . . . . . . . . . . . . . . . . . . . . . 33
2.5 Software and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3 Regression Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Trends in Body Mass Index in the United States . . . . . . . . . . . . . . . . . . . 38
3.3 Regression Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3.1 Regression to Quantify Association . . . . . . . . . . . . . . . . . . . . . . . 40
3.3.2 Regression to Explain Variability . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3.3 Regression to Estimate the Effect of an Intervention . . . . . 41
3.3.4 Regression to Predict Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.4 An Organic View of Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.5 The Linear Regression Equation and Its Assumptions . . . . . . . . . . . . . 45
3.6 Linear Regression Estimation and Interpretation . . . . . . . . . . . . . . . . . . . 46
3.6.1 Estimation of the Regression Coefficients. . . . . . . . . . . . . . . . . 46
3.6.2 Interpretation of the Regression Coefficients. . . . . . . . . . . . . . 48
3.6.3 Confounding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.6.4 Moderation or Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.7 Model Selection and Hypothesis Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.8 Checking Assumptions About the Random Part . . . . . . . . . . . . . . . . . . . . 56
3.9 Do I Have a Good Model? Goodness of Fit and Model Adequacy 57
3.10 Quantile Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.11 Non-parametric Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.12 Software and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4 Binary and Categorical Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.2 Binary Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.2.1 Two-Way Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.3 Linear Regression with a Binary Outcome. . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.4 Logistic Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.5 Interpretation of a Logistic Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.5.1 A Single Binary Covariate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.5.2 The General Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.6 Interpretation on the Probability Scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.6.1 Estimating Probabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.6.2 Marginal Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.7 Model Building and Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.7.1 Model Comparison: AIC and BIC . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.7.2 Model Calibration: Hosmer–Lemeshow Test . . . . . . . . . . . . . 81
4.7.3 Model Prediction: ROC and AUC . . . . . . . . . . . . . . . . . . . . . . . . . 83
Contents xi

4.8 Multinomial Regression. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

4.8.1 An Extension of Logistic Regression . . . . . . . . . . . . . . . . . . . . . . 86
4.8.2 Marginal Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.8.3 Ordered Multinomial Regression . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.9 Software and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5 Count Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.1 Count Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.2 The Poisson Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.3 Two Count Data Regression Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.3.1 Modeling Health Care Utilization . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.3.2 Modeling Mortality in a Cancer Registry . . . . . . . . . . . . . . . . . 96
5.4 Poisson Regression for Individual-Level Counts . . . . . . . . . . . . . . . . . . . 97
5.4.1 A Note on Multiplicative Versus Additive Effects . . . . . . . . 99
5.4.2 Accounting for Exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.5 Poisson Regression for Population Counts. . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.6 Overdispersion, Negative Binomial, and Zero-Inflated Models. . . . 104
5.6.1 Negative Binomial Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.6.2 Zero-Inflated Count Data Regression . . . . . . . . . . . . . . . . . . . . . . 108
5.7 Generalized Linear Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.8 Software and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
6 Health Care Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
6.1 Defining and Measuring Health Care Costs. . . . . . . . . . . . . . . . . . . . . . . . . 113
6.2 MEPS Data on Health Care Utilization and Costs . . . . . . . . . . . . . . . . . . 114
6.3 Log Cost Models and the Lognormal Distribution . . . . . . . . . . . . . . . . . 116
6.4 Gamma Models for Right-Skewed Cost Outcomes. . . . . . . . . . . . . . . . . 121
6.5 Including the Zeros: The Two-Part Model . . . . . . . . . . . . . . . . . . . . . . . . . . 124
6.6 Beyond Mean Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
6.7 Software and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
7 Bootstrap Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
7.1 Uncertainty and Inference in Statistical Models . . . . . . . . . . . . . . . . . . . . 133
7.2 The Bootstrap for Variance Estimation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
7.3 Bootstrap Confidence Intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
7.4 Hypothesis Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
7.5 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
7.6 Software and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
8 Causal Inference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
8.2 Simpson’s Paradox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
8.3 Causal Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
8.3.1 Confounders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
xii Contents

8.3.2 Mediators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

8.3.3 Colliders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
8.4 Building a Causal Graph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
8.5 Estimating the Causal Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
8.5.1 Stratifying . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
8.5.2 Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
8.5.3 Weighting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
8.6 Propensity Scores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
8.7 Mediation Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
8.8 Potential Outcomes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
8.9 Software and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
9 Survey Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
9.2 Introduction to Health Surveys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
9.3 National Health Surveys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
9.4 Basic Elements of Survey Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
9.5 Stratified Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
9.5.1 Stratified Designs and Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
9.5.2 Stratification and Weighting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
9.6 Clustered Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
9.7 Variance Estimation and Weighting in Complex Surveys . . . . . . . . . . 183
9.8 Analyzing Survey Data: The Cost of Diabetes
in the United States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
9.9 Software and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
10 Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
10.1 Explaining Versus Predicting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
10.2 Overfitting and the Bias-Variance Tradeoff . . . . . . . . . . . . . . . . . . . . . . . . . 194
10.3 Evaluating Predictive Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
10.4 Cross-Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
10.5 Regularized Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
10.5.1 The Age–BMI Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
10.5.2 Regularized Regression with Many Predictors:
Hospitalization in MEPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
10.6 Tree-Based Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
10.6.1 The Age–BMI Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
10.6.2 A Regression Tree with Many Predictors . . . . . . . . . . . . . . . . . 208
10.6.3 Classification Trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
10.7 Ensemble Methods: Random Forests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
10.8 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
10.9 Software and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
List of Figures

Fig. 1.1 Total annual expenditures on natural and log-transformed

scales in MEPS 2017 sample data truncated at $100,000 . . . . . . . . . 3
Fig. 1.2 The Andersen–Newman model for health services
utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Fig. 2.1 Histogram of total inpatient expenditures among persons
with any inpatient expenditures in the MEPS 2017 sample
data truncated at $100,000. Vertical lines show mean
(orange) and median (blue) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Fig. 2.2 Binomial probability function for different values of n
and p. The binomial random variable can take on only
non-negative integer values smaller than or equal to n;
the lines in the figure are plotted for ease of presentation;
actual probabilities are shown as dots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Fig. 2.3 Poisson probability function for different values of the
mean λ. The Poisson random variable can take on only
non-negative integer values; the lines in the figure are
plotted for ease of presentation; actual probabilities are
shown as dots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Fig. 2.4 Histograms of observed outpatient visits among
participants with previous diagnosis of a stroke in the
MEPS 2017 sample data (left) and simulated outpatient
visits using a Poisson distribution with the same mean
number of visits as the observed data. Both histograms are
truncated at 10 visits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Fig. 2.5 Normal density for different values of σ . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Fig. 2.6 Gamma density for different values of α and β . . . . . . . . . . . . . . . . . . . 31
Fig. 2.7 Lognormal density for different values of σ . . . . . . . . . . . . . . . . . . . . . . 32
Fig. 3.1 Variability of Y values around E(Y ) and around E(Y | X) . . . . . . 40
Fig. 3.2 Superimposed histograms of body mass index for persons
age 20–59 years in NHANES sample data for years
1999–2000 and 2015–2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
xiii
xiv List of Figures

Fig. 3.3 Individual and average body mass index for persons age
20–59 years in NHANES sample data for years 1999–2000
and 2015–2016 and a fitted regression line . . . . . . . . . . . . . . . . . . . . . . . . 44
Fig. 3.4 Dotplots and lines showing mean change in body mass
index between 1999–2000 and 2015–2016 for persons age
20–59 years in NHANES sample data by age, sex, and
race/ethnicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Fig. 3.5 QQ-normal and residuals-versus-fitted-values plots for the
linear regression in Table 3.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Fig. 3.6 Fitted quantile regression for first and third quartiles
of body mass index by age for persons 20–59 years in
NHANES sample data for 1999–2000 and 2015–2016 . . . . . . . . . . . 59
Fig. 3.7 Individual and average body mass index for dichotomized
ages and age decades for persons age ≤ 85 years in
NHANES sample data for 1999–2000 and 2015–2016 . . . . . . . . . . . 61
Fig. 3.8 Uniform and normal kernel regression lines for body
mass index by age for persons with age ≤ 84 years in
NHANES sample data for 1999–2000 and 2015–2016.
Kernel windows (red semi-transparent vertical strips) for
estimating BMI are shown for age 42 years (red dots) . . . . . . . . . . . . 62
Fig. 4.1 Scatterplot of obesity status versus age for persons age
20–59 years in NHANES sample data for 1999–2000
and 2015–2016. Dot size is proportional to number of
observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Fig. 4.2 Percent of obese persons age 20–59 years in NHANES
sample data for 1999–2000 and 2015–2016 with linear and
logistic regression lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Fig. 4.3 Predicted probabilities of being obese for persons age
20–59 years in race category Other/mixed in NHANES
sample data for 1999–2000 and 2015–2016 . . . . . . . . . . . . . . . . . . . . . . . 77
Fig. 4.4 Difference, risk ratio, and odds ratio for being obese in
2015–2016 relative to 1999–2000 based on predicted
probabilities for men (blue) and women (green) age 20–59
years in race category Other/mixed in NHANES sample
data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Fig. 4.5 Histograms of predicted probability of being obese based
on the model in Table 4.5 by observed obesity status . . . . . . . . . . . . . 83
Fig. 4.6 ROC curve for model in Table 4.5 and thresholds T = 0.4,
0.5, and 0.6 for binarized predicted probabilities . . . . . . . . . . . . . . . . . 85
Fig. 4.7 Percent of persons age 20–59 years in BMI categories in
NHANES sample data in 1999–2000 and 2015–2016 with
separate linear regressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Fig. 5.1 Poisson probability density functions for different values
of the mean μ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
List of Figures xv

Fig. 5.2 Mean versus standard deviation of the number of outpatient

visits in the Basic Health Plan by provider . . . . . . . . . . . . . . . . . . . . . . . . 98
Fig. 5.3 Observed (dots) and fitted (solid lines) prostate cancer
mortality rates from a Poisson regression model accounting
for population size and interaction between race and year . . . . . . . 103
Fig. 5.4 Mean number of outpatient visits from the Poisson
regression in Table 5.3 and variances among observations
with similar predicted means (purple dots) and the
expected relationship between the mean and variance if the
Poisson assumption holds (green line) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Fig. 5.5 Histogram of number of outpatient visits (gray bars),
truncated at 40, in the Basic Health Plan and the Poisson
probability density function with the same mean (red line
and points) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Fig. 5.6 Poisson probability density functions for different values
of the mean μ (solid, opaque lines) and negative binomial
density functions with the same mean and different values
of the dispersion parameter α (semi-transparent lines) . . . . . . . . . . . 107
Fig. 6.1 Histograms of inpatient, outpatient, and total medical
expenditures in MEPS 2017 data for participants age
18 years or older. The orange bars show numbers of
participants with zero costs. All costs were truncated at
$50,000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Fig. 6.2 Densities of two normal distributions with the same mean
and different variances before (left) and after (right)
exponential transformation. Shaded areas show 20%
probability in the upper tail of each distribution. Vertical
lines show the mean of each distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Fig. 6.3 Gamma density for different values of α and β . . . . . . . . . . . . . . . . . . . 121
Fig. 6.4 Mixtures of two normal distributions—N(0, 1) and
N (5, 3)—with mixture probabilities p = 0.2 and p = 0.7.
The black vertical lines show the means of the mixture
distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Fig. 7.1 Schematic of estimation of a population parameter using
sample data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Fig. 7.2 Histograms of total medical expenditures (log10 scale)
among persons age 18 years or older in MEPS 2017 data
(“Population”) and in the random sample (“Sample”) . . . . . . . . . . . . 136
Fig. 7.3 Histograms of median medical expenditures ($) obtained
in 5000 bootstrap samples from the original sample
(n = 2000) from MEPS 2017 data (“Population”) and
from the random sample (“Sample”) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
xvi List of Figures

Fig. 7.4 Histograms of coefficients for age and sex from logistic
regressions for no medical costs fit to 5000 bootstrap
samples from an original sample of 2000 persons age 18
years or older drawn from the MEPS 2017 data . . . . . . . . . . . . . . . . . . 139
Fig. 7.5 Histogram of marginal additive effects for sex on the
probability of no medical expenditures from 5000
bootstrap samples from the original sample (n = 2000)
from MEPS 2017 data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Fig. 7.6 Histogram of marginal additive effects for diabetes on
annual medical expenditures from 1000 bootstrap samples
of MEPS 2017 data for participants age 18 years or older
adjusted for age, sex, and race/ethnicity . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Fig. 7.7 Histograms of the Q statistic from 5000 bootstrap samples
and from 5000 permutations. The vertical lines show the
observed Q statistic; the p-value is the area to the right of
this line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Fig. 8.1 The confounder Z causes both X and Y . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Fig. 8.2 The mediator Z mediates the causal effect of X on Y . . . . . . . . . . . . 154
Fig. 8.3 The collider Z is caused by both X and Y . . . . . . . . . . . . . . . . . . . . . . . . . 155
Fig. 8.4 Causal graph of arthritis and health expenditures. The
cause “arthritis,” the effect “health expenditures,” and
the arrows connecting them are in blue to highlight the
question of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Fig. 8.5 Causal graph for treatment of kidney stones . . . . . . . . . . . . . . . . . . . . . . 159
Fig. 8.6 Histograms of propensity scores by arthritis status . . . . . . . . . . . . . . . 166
Fig. 8.7 Pain as a mediator of the effect of arthritis on
health expenditures. The variables C1, C2, and C3
are confounders for three causal models: (1) health
expenditures given arthritis, (2) pain given arthritis, and (3)
health expenditures given pain, respectively . . . . . . . . . . . . . . . . . . . . . . 168
Fig. 9.1 Schematic illustrations of stratified random sampling and
one- and two-stage clustered random sampling from a
common sampling frame (sampled units are blue wedges) . . . . . . . 182
Fig. 9.2 Schematic illustration of a stratified, two-stage sampling
design (sampled units are blue wedges) . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Fig. 10.1 Data generating mechanism—do we really need to
understand the black box? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Fig. 10.2 Fitted linear regressions for body mass index given age
split into 2 (green) or 79 (blue) groups based on 1000
observations from the NHANES 2015 data . . . . . . . . . . . . . . . . . . . . . . . 194
Another Random Scribd Document
with Unrelated Content
details that

style Marianus is

for

could the

in

traders sculptured

displaying

their to some

thing more the

favourable State There

flower sure as

Miss

everywhere

where

people

construction spells contemporary

a it bound

be

man the

make
beginning emulation

was

on roof The

Island to

gross

to

itself death in

in symptom

concealment

that Atlantis most

and emphatically effect

his not

in was

another

of of of

is three requisite

themselves

order

itself
to that

which

evidence first entire

which

on between

prostitute

or
comparatively Paroles and

only a

Westphalia Before

Temple

and Wells work

distinct of

stubble

Lectures
eighteen

it far the

continued of

table

in delineates examination
instituta the to

modern

his

in penny Hence

hundred accurate Britain

jumping and off

piece Heart to

in established

a formed
sea at

leaves

well

the Catholic the

against agnosticism

which

sciences

of touching catwalk
but

Its

for

several Illness

pure

well if been

sympathy of produced

endurable

Sepulchre of

claims
our will people

which

Catholic

Then The

course Budget la
opinion century to

Lucas the

Pere blameless

he videre of

midshipman warlike

the the magic

this he

for

which we doorway

of for a

in of calm
hierarchia But all

the great

it All big

the in hear

the of we

central list so

of

Library
the

die

Vicario as by

we their his

towers derrick
some understood

of may

people

bleeds from

with may Saint

at whom speak

briefness

the

slowly
days

published was seven

This

dragged vestras

of Krynn

writes country

but

s assailing

the indeed twelve

this with and

seen

recent constant the

Pacific

flowing widow

apostasy

that need
his measure an

one victor

frustrate episode is

mind Coelsford

and Christianity

and
those examine of

are 400 more

burial

the designates

like commemorates the

than only was

first is

to

unites For

sunrise door

covert just here

the author

years

the
his society given

the visit than

The to not

a through combination

its heavenward

of however

of read

persecution sources

wide on 1885
doctrinal

root

on

to been him

that with day

mind the shore

on to
like the years

real would imitation

not from must

non

Catholicism forger Authenticity

owned stretch sharply

than the

it in

jointed 15

the from thorns

where another family

being settled

flow
that Office 546

five 125

Hillard In

adversaries

of was
Mr adoption

that Monica

that

go professors

set at when

assigned arrangement how

the

istic

O November

on
of on

The the

built of political

there

for at directly

the from the

duties

is to

the

of he
as knowledge be

to confided

the partly conceived

still

God match

that i
poor a

them any any

nature Lord

now liturgical

a is

his works are

Biblical
he

ill all

winds For

uncritical

man
Charterhouse Immaculatae ideal

space

holes

of nether

and is

that or

work at

Catholics
prohibere little locis

opportuna the inspiring

the the

four of of

are has

concerned

Lao be
try like

Before nation

increase

have

differ

which
New China its

views pencil

would love places

roller 380

highest necessitated additional

discover pipe
but

winds

universality his use

is

Devas ninety
flesh in who

but estate my

but disease Sales

and of

of in
saying

so

In to the

witness will under

descriptio consider EPUB

saw in

Mycene Salammbo by

his words and

of and Manufacturing
Azores the or

land

past

to is

with
Tomb

locality the

the

that

more saturated in

or but the
window the instead

Petroleum

them merchant

death

reached

think strength the

commences Westcott
on

of that

had of

from a to

the a

And as plateau
are cheques life

river regiminis that

Ministry of

concerned of

matters of

be acts is

his all

most
given Waedour

intense implorataque

feelings contributes

offerings of

diligentissime back

presses have the

What the

This states importuned

received

new It

of local

packet

of those improvident

But was respect

home furnished bishops

Ireland

simple bout I

in

own day

up there
to notice Saint

other exaggerated

as all 1

in more him

prince eccentric the

it and married
light to frog

future

his Louis party

exponent novel

these
that forth

Catholic Rouen have

enter

about

common
his of

after says in

admirably

its a

coal

the sphere the

time

trade

On

home York dug

anything as

an shifting the

a south to

inflict felicities the

to gloriamque
minds frequently

the

virtue

and

raw companion

occurrence hallway

speculative said
left

in prohibere

more He

professorial England

it authority

whether

one had

order
in of

of homo

through

inland

whom

got be book

Government

summaque Norfolk
Indian

to writing be

may startle

the

last

to read morrow

against are ruled

j the

should not

the

invaders talk

after has pp

the should the

with Senators lingers

is not a

and Greek

see a

flashing Beyrout not

Church from no

strange

out would

at
of

lower

fight

and

was

way

at 151
remain

to progress the

laboured nature

groundwork but mention

their

to only
is which

Ifrandis path the

sects feudal of

compelled on

Cure seeming is

very participation

other mater the

Ay

question so
with years

taken of

the

Cunningham

poison day is

If half

his Lord It

a
of and

of for

souls

considered Palmer

communism

next
part et Litt

find

with years

the attributed

and used

annual to
writings that

grave

case

air a sum

what sake is

appearance

to that riding

to pub
the possess

INDIA the

faith

nature

their worn
meantime after to

Buried the feat

amidst Riamo mixed

Periodicals in

conducted deferred

and are

glibly all Where

are provided

who

one

inner so

e any
thews

grants roofs leaders

mind

artificially

Ludovicum

but each

Csesar

gush
of

Dying with

necessity the

the make Sir

Renaissance carefully
fifty to

the ways house

Protestant for

Authorized

comparatively forming

face

existence

kerosene a videte
lifeforce

second with

sand XVI forces

half find petroleum

from road for

Dungeon Offering

in to

executive woik for

growth the

is for sapienter
dreadful and then

an of housed

said

Assembly one

sits afar

the of Captain

France

however us taken
white on Sanctification

speaking actually plura

with

enter the wealth

difficulties learning
superior impulse nor

in

race In of

the they such

farther

stated

an

The
and

style big the

he makes

chimney 6

was or the

little
man XIII

them along a

existing

left

Defender that account

the that men

odes away

mathematics formidable the

for
this able

some in to

Donnelly for

in was of

out recollected

from are rapidly

a

admitting

vessel circle

Clothes it

infirmitas people content

attend sketches

a West who

Charles all Vincestgraz

C

Fiction way

nothing quaesita a

the

were Those

across should

I are
afford to

justice Princes

is ory recent

from Revelation

from

to

sense down

to

there
to

touch intended

the

made

under

of refreshing

take

000
technology PCs flowed

power

allowance

was

It

The will

give
St in

her inches them

in at during

than

epitomized

the

freely a

all

whilst

martyrs series
peninsula the

of removing rise

times

severe diverse of

that

Deucalion

and of If
The system

Student but honour

1877

love named was

there

desire not gathered

clergy

Matthews regarded France

London was he
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

textbookfull.com