Primer of Applied Regression & Analysis of Variance 3E - 3rd

Edition

Visit the link below to download the full version of this book:

https://medipdf.com/product/primer-of-applied-regression-analysis-of-variance-3e
-3rd-edition/

Click Download Now

PRIMER OF APPLIED REGRESSION
AND ANALYSIS OF VARIANCE, THIRD EDITION

Copyright © 2016 by McGraw-Hill Education. All rights reserved. Printed in the United States of
America. Except as permitted under the United States Copyright Act of 1976, no part of this
publication may be reproduced or distributed in any form or by any means, or stored in a data base or
retrieval system, without the prior written permission of the publisher.

Previous edition copyright © 2001 by The McGraw-Hill Companies, Inc.

1 2 3 4 5 6 7 8 9 0 DOC/DOC 20 19 18 17 16
ISBN 978-0-07-182411-8
MHID 0-07-182411-1

e-ISBN 978-0-07-182244-2
e-MHID 0-07-182244-5

This book was set in Palatino LT Std-Roman 10/12 by MPS Limited.

The editors were Kathryn Schell and Robert Pancotti.
The production supervisor was Richard Ruzycka.
Project management was provided by Vivek Khandelwal, MPS Limited.
The text designer was José Fonfrias; the cover designer was Anthony Landi.
RR Donnelley was the printer and binder.

This book is printed on acid-free paper.

Library of Congress Cataloging-in-Publication Data

Glantz, Stanton A., author.
Primer of applied regression & analysis of variance / Stanton A. Glantz, Bryan K. Slinker, Torsten
B. Neilands. — Third edition.
p. ; cm.
Primer of applied regression and analysis of variance
Includes bibliographical references and index.
ISBN 978-0-07-182411-8 (hardcover : alk. paper) — ISBN 0-07-182411-1 (hardcover : alk. paper)
I. Slinker, Bryan K., author. II. Neilands, Torsten B., author. III. Title. IV. Title: Primer of applied
regression and analysis of variance.
[DNLM: 1. Biometry—methods. 2. Analysis of Variance. 3. Data Interpretation, Statistical. 4.
Regression Analysis. WA 950]
QH323.5
570.1’5195—dc23
2015033598

McGraw-Hill Education books are available at special quantity discounts to use as premiums and sales
promotions or for use in corporate training programs. To contact a representative, please visit the
Contact Us pages at www.mhprofessional.com.
 To Marsha, Aaron, and Frieda Glantz for coming to
Vermont so that we
could originally write this book and for their continuing
support on all the adventures since then

To Kathy, Kyle, and Sarah Slinker, who make it all

worthwhile

To J.B. and Juanita Neilands, Dianne Seaborg, and Penny

Seaborg
for your love and essential support, past and present
Contents

Preface

ONE Why Do Multivariate Analysis?

OUR FIRST VISIT TO MARS
DUMMIES ON MARS
SUMMARY
PROBLEM

TWO The First Step: Understanding Simple Linear Regression

MORE ON MARS
The Population Parameters
HOW TO ESTIMATE THE LINE OF MEANS FROM A SAMPLE
The Best Straight Line Through the Data
Variability About the Regression Line
Standard Errors of the Regression Coefficients
HOW CONVINCING IS THE TREND?
Testing the Slope of the Regression Line
Comparing Slopes and Intercepts of Two Regression Lines
Testing the Regression as a Whole
Cell Phone Radiation, Reactive Oxygen Species, and DNA Damage
in Human Sperm
CONFIDENCE INTERVALS FOR REGRESSION
Confidence Interval for the Line of Means
Confidence Interval for an Observation
CORRELATION AND CORRELATION COEFFICIENTS
The Relationship Between Regression and Correlation
DOING REGRESSION AND CORRELATION ANALYSIS WITH A
COMPUTER
 HEAT EXCHANGE IN GRAY SEALS
SUMMARY
PROBLEMS

THREE Regression with Two or More Independent Variables

WHAT WE REALLY DID ON MARS
HOW TO FIT THE BEST PLANE THROUGH A SET OF DATA
Computing the Regression Coefficients
Variability About the Regression Plane
Standard Errors of the Regression Coefficients
Muddying the Water: Multicollinearity
DOES THE REGRESSION EQUATION DESCRIBE THE DATA?
Incremental Sums of Squares and the Order of Entry
Relationship to t Tests of Individual Regression Coefficients
THE COEFFICIENT OF DETERMINATION AND THE MULTIPLE
CORRELATION COEFFICIENT
MORE DUMMIES ON MARS
MECHANISMS OF TOXIC SHOCK
PROTEIN SYNTHESIS IN NEWBORNS AND ADULTS
GENERAL MULTIPLE LINEAR REGRESSION
Multiple Regression in Matrix Notation
DIABETES, CHOLESTEROL, AND THE TREATMENT OF HIGH
BLOOD PRESSURE
BABY BIRDS BREATHING IN BURROWS
POLYNOMIAL (AND SOME OTHER NONLINEAR)
REGRESSIONS
Heat Exchange in Gray Seals Revisited
Other Nonlinear Regressions
INTERACTIONS BETWEEN THE INDEPENDENT VARIABLES
How Bacteria Adjust to Living in Salty Environments
The Response of Smooth Muscle to Stretching
SUMMARY
PROBLEMS
FOUR Do the Data Fit the Assumptions?
ANOTHER TRIP TO MARS
LOOKING AT THE RESIDUALS
A QUANTITATIVE APPROACH TO RESIDUAL ANALYSIS
Standardized Residuals
Using Residuals to Test for Normality of the Residuals
Leverage
Studentized Residuals
Cook’s Distance
WHAT DO YOU DO WITH AN INFLUENTIAL OBSERVATION
ONCE YOU HAVE FOUND IT?
Problems with the Data
Problems with the Model
Data Transformations
WATER MOVEMENT ACROSS THE PLACENTA
CHEAPER CHICKEN FEED
HOW THE BODY PROTECTS ITSELF FROM EXCESS ZINC
AND COPPER
Back to Square One
CHEAPER CHICKEN FEED REVISITED: BOOTSTRAPPING
AND ROBUST STANDARD ERRORS
Bootstrap Standard Errors
Robust Standard Errors
CLUSTERED DATA
AIDS ORPHANS IN UGANDA
SUMMARY
PROBLEMS

FIVE Multicollinearity and What to Do About It

WHERE MULTICOLLINEARITY COMES FROM
BACK TO MARS
DETECTING AND EVALUATING MULTICOLLINEARITY
Qualitative Suggestions of Harmful Multicollinearity
 Correlations Among the Independent Variables
The Variance Inflation Factor
Auxiliary Regressions
The Correlations of the Regression Coefficients
THE CONSEQUENCES OF HAVING TWO PUMPS IN ONE
HEART
FIXING THE REGRESSION MODEL
Centering the Independent Variables
Deleting Predictor Variables
More on Two Pumps in One Heart
FIXING THE DATA
Getting More Data on the Heart
USING PRINCIPAL COMPONENTS TO DIAGNOSE AND TREAT
MULTICOLLINEARITY
Standardized Variables, Standardized Regression, and the
Correlation Matrix
Principal Components of the Correlation Matrix
Principal Components to Diagnose Multicollinearity on Mars
Principal Components and the Heart
Principal Components Regression
More Principal Components on Mars
The Catch
Recapitulation
SUMMARY
PROBLEMS

SIX Selecting the “Best” Regression Model

SO WHAT DO YOU DO?
WHAT HAPPENS WHEN THE REGRESSION EQUATION
CONTAINS THE WRONG VARIABLES?
WHAT DOES “BEST” MEAN?
The Coefficient of Determination R2
The Adjusted R2
 The Standard Error of the Estimate sy|x
Independent Validations of the Model with New Data
The Predicted Residual Error Sum of Squares, PRESS
Bias Due to Model Underspecification and Cp
But What Is “Best”?
SELECTING VARIABLES WITH ALL POSSIBLE SUBSETS
REGRESSION
What Determines an Athlete’s Time in a Triathlon?
SEQUENTIAL VARIABLE SELECTION TECHNIQUES
Forward Selection
Backward Elimination
Stepwise Regression
Interpreting the Results of Sequential Variable Selection
Another Look at the Triathlon
PREDICTIVE OPTIMISM
SUMMARY
PROBLEMS

SEVEN Missing Data

PREVENTION IS KEY
MISSING DATA MECHANISMS
AD HOC MISSING DATA HANDLING METHODS
Listwise Deletion
Single Imputation: Mean and Regression Imputation
MAXIMUM LIKELIHOOD ESTIMATION WITH COMPLETE
DATA
Using Maximum Likelihood to Estimate Population Mean and
Standard Deviation
Maximum Likelihood Regression
Putting It All Together: Martian Weights Predicted by Heights and
Water Consumption via Maximum Likelihood Estimation
REGRESSION ANALYSIS VIA MEANS, VARIANCES, AND
COVARIANCES
The Multivariate Normal Distribution and Covariance
 Estimating the Regression Coefficients Based on the Mean and
Covariance Matrices
Back to Mars
MAXIMUM LIKELIHOOD REGRESSION ESTIMATION WITH
INCOMPLETE DATA
Missing Martians
Excess Zinc and Copper and Missing Data
Missing Data Mechanisms Revisited: Three Mechanisms for
Missing Martians
NON-NORMAL DATA AND MAXIMUM LIKELIHOOD
ESTIMATION
Smoking, Social Networks, and Personality
The Multivariate Normality Assumption
MULTIPLE IMPUTATION
Generate the Multiply-Imputed Data Sets
Analyzing the Multiply-Imputed Data Sets
Combining Results from Multiply-Imputed Data Sets and
Quantifying the Uncertainty Due to Multiple Imputation
MULTIPLE IMPUTATION EXTENSIONS AND
COMPLICATIONS
Number of Imputations
How Many and Which Independent Variables Should be Included
in the Imputation Process?
Should a Dependent Variable with Complete Data be Included in
Generating Multiple Imputations?
Small Samples
Non-Normal Data
Clustered Data
Data Not Missing at Random
Excess Zinc and Copper and Missing Data Revisited
SUMMARY
PROBLEMS

EIGHT One-Way Analysis of Variance

 USING A t TEST TO COMPARE TWO GROUPS
Does Secondhand Tobacco Smoke Nauseate Martians?
Using Linear Regression to Compare Two Groups
THE BASICS OF ONE-WAY ANALYSIS OF VARIANCE
Traditional Analysis-of-Variance Notation
Accounting for All the Variability in the Observations
Expected Mean Squares
Using Linear Regression to Do Analysis of Variance with Two
Groups
USING LINEAR REGRESSION TO DO ONE-WAY ANALYSIS OF
VARIANCE WITH ANY NUMBER OF TREATMENTS
Hormones and Depression
MULTIPLE COMPARISON TESTING
The Bonferroni t Test
More on Hormones and Depression
Holm t Test
Holm–Sidak t Test
What Is a Family?
Diet, Drugs, and Atherosclerosis
TESTING THE ASSUMPTIONS IN ANALYSIS OF VARIANCE
Formal Tests of Homogeneity of Variance
More on Diet, Drugs, and Atherosclerosis
Alternatives to the Usual F Statistic When Variances Are Unequal
Alternatives to the t Test Statistic When Variances Are Unequal
Maturing Rat Lungs
SUMMARY
PROBLEMS

NINE Two-Way Analysis of Variance

TRADITIONAL TWO-WAY ANALYSIS OF VARIANCE
PERSONALITY ASSESSMENT AND FAKING HIGH GENDER
IDENTIFICATION
Traditional Analysis of Variance
 Using Regression to Perform Two-Way Analysis of Variance
AN ALTERNATIVE APPROACH FOR CODING DUMMY
VARIABLES
An Alternative Approach to Personality
Why Does It Matter How We Code the Dummy Variables?
The Kidney, Sodium, and High Blood Pressure
What Do Interactions Tell Us?
Multiple Comparisons in Two-Way Analysis of Variance
More on the Kidney, Sodium, and High Blood Pressure
UNBALANCED DATA
ALL CELLS FILLED, BUT SOME CELLS HAVE MISSING
OBSERVATIONS
The Case of the Missing Kidneys
Summary of the Procedure
What If You Use the Wrong Sum of Squares?
Multiple Comparisons with Missing Data
ONE OR MORE CELLS EMPTY
Multiple Comparisons with Empty Cells
More on the Missing Kidney
Multiple Comparisons for the Missing Kidney
Recapitulation
RANDOMIZED BLOCK DESIGNS
A More Thorough Study of Martian Nausea
What Do You Gain by Blocking?
Regression Implementation of Randomized Blocks
Recapitulation
SUMMARY
PROBLEMS

TEN Repeated Measures

ONE-WAY REPEATED-MEASURES ANALYSIS OF VARIANCE
Hormones and Food
Comparison with Simple Analysis of Variance
 Multiple Comparisons in Repeated-Measures Analysis of Variance
Recapitulation
TWO-FACTOR ANALYSIS OF VARIANCE WITH REPEATED
MEASURES ON ONE FACTOR
Partitioning the Variability
Testing the Non–Repeated-Measures Factor
Testing the Repeated-Measures Factor
Is Alcoholism Associated with a History of Childhood Aggression?
THE GENERAL LINEAR MODEL
Traditional Regression Model Structure for Mixed Models
Estimated by Maximum Likelihood
Maximum Likelihood Estimation for Linear Mixed Models
HYPOTHESIS TESTING IN MAXIMUM LIKELIHOOD
ESTIMATION
Testing the ANOVA Effects Using the Regression Equation
Testing the Individual Coefficients
The Wald χ2 and F Tests
Maximum Likelihood Mixed Models Analysis of Drinking and
Antisocial Personality Using an Overspecified Model
Maximum Likelihood Mixed Models Analyses of the Study of
Drinking and Antisocial Personality: Including Subjects with
Missing Data on the Dependent Variable
BETTER ESTIMATES OF THE COVARIANCE STRUCTURE FOR
REPEATED MEASURES
Treating Between-Subjects Effects as Random Effects
Estimating the Covariance in the Residuals
This Is Your Rat’s Brain on Drugs
TWO-WAY ANALYSIS OF VARIANCE WITH REPEATED
MEASURES ON BOTH FACTORS
Candy, Chewing Gum, and Tooth Decay
MISSING DATA IN REPEATED MEASURES ON BOTH OF TWO
FACTORS
More on Chewing Gum
RESTRICTED MAXIMUM LIKELIHOOD ESTIMATION
 Secondhand Smoke and the Cardiovascular System
MISSING DATA IN TWO-FACTOR OLS ANALYSIS OF
VARIANCE WITH REPEATED MEASURES ON ONE FACTOR
Assumptions Underlying Repeated-Measures Analysis of Variance
Expected Mean Squares, Fixed Effects, Random Effects, and Mixed
Models
What Happens to the Expected Mean Squares in Mixed Models
When There Are Missing Data?
More on Drinking and Antisocial Personality
OLS METHODS FOR TWO-WAY ANALYSIS OF VARIANCE
WITH REPEATED MEASURES ON BOTH FACTORS
Partitioning the Variability
Candy, Chewing Gum, and Tooth Decay Revisited
What Happens to the Expected Mean Squares When There Are
Missing Data?
More on Chewing Gum
Compound Symmetry Revisited: Rat Brains and Cocaine
ACCOUNTING FOR BETWEEN-SUBJECTS VARIABILITY IN
LINEAR REGRESSION
Random-Effects Regression: Reduced Nausea Cigarettes
SUMMARY
PROBLEMS

ELEVEN Mixing Continuous and Categorical Variables:

Analysis of Covariance
HIGH-PRESSURE PREGNANCIES
From the Regression Perspective
From the ANOVA Perspective
Confounding Variables
HOW DOES ANALYSIS OF COVARIANCE WORK?
The Relationship of ANCOVA to Regression
Adjusted Means
Testing the Assumption of Homogeneous Slopes
COOL HEARTS
 Multiple Comparisons After Traditional ANCOVA
Did We Gain Anything by Doing ANCOVA Instead of ANOVA?
WHAT HAPPENS WHEN THE RANGE OF VALUES OF THE
COVARIATES IS VERY DIFFERENT AMONG THE GROUPS?
Fat-Free Exercising
WHAT HAPPENS WHEN THE SLOPES ARE NOT
HOMOGENEOUS?
MORE ASSUMPTION CHECKING
Ridding Your Body of Drugs
MORE COMPLICATED ANALYSES OF COVARIANCE
SUMMARY
PROBLEMS

TWELVE Regression with a Qualitative Dependent Variable:

Logistic Regression
LOGISTIC REGRESSION
Our Last Visit to Mars
Odds
The Multiple Logistic Equation
ESTIMATING THE COEFFICIENTS IN A LOGISTIC
REGRESSION
Maximum Likelihood Estimation
HYPOTHESIS TESTING IN LOGISTIC REGRESSION
Testing the Logistic Equation
Testing the Individual Coefficients
Confidence Intervals for Individual Coefficients
Back to Mars
IS THE LOGISTIC REGRESSION EQUATION AN APPROPRIATE
DESCRIPTION OF THE DATA?
Regression Diagnostics for Logistic Regression
Goodness-of-Fit Testing
ARE BONE CANCER PATIENTS RESPONDING TO
CHEMOTHERAPY?
STEPWISE LOGISTIC REGRESSION
 Nuking the Heart
CONVERGENCE PROBLEMS IN LOGISTIC REGRESSION
LOGISTIC REGRESSION FOR LONGITUDINAL AND
CLUSTERED DATA
Robust Standard Errors in Logistic Regression
Generalized Estimating Equations
Generalized Linear Mixed Models Analysis of Clustered Binary
Data
Selecting an Approach for Clustered Data
SUMMARY
PROBLEMS

THIRTEEN Regression Modeling of Time-to-Event Data:

Survival Analysis
SURVIVING ON PLUTO
Censoring on Pluto
Estimating the Survival Curve
The Hazard Function
THE PROPORTIONAL HAZARDS MODEL
Bone Marrow Transplantation to Treat Adult Leukemia
Testing the Individual Coefficients
Survival Following Surgery for Pancreatic Cancer
ADDITIONAL TOPICS IN PROPORTIONAL HAZARDS
REGRESSION
Testing the Proportionality Assumption
Variable Selection and Stepwise Techniques
Recurrent Events and Clustered Data
Missing Data
SUMMARY
PROBLEMS

FOURTEEN Nonlinear Regression

EXPONENTIAL MODELS