Multivariate Analysis for the Behavioral Sciences Kimmo Vehkalahti 2024 Scribd Download

Download the Full Version of textbook for Fast Typing at textbookfull.
com
Multivariate Analysis for the Behavioral Sciences

Kimmo Vehkalahti
https://textbookfull.com/product/multivariate-analysis-for-
the-behavioral-sciences-kimmo-vehkalahti/
OR CLICK BUTTON
DOWNLOAD NOW
Download More textbook Instantly Today - Get Yours Now at textbookfull.com

Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.
Multivariate analysis for the behavioral sciences 2nd

Edition Brian S. Everitt
https://textbookfull.com/product/multivariate-analysis-for-the-
behavioral-sciences-2nd-edition-brian-s-everitt/
textboxfull.com
Applied Power Analysis for the Behavioral Sciences 2nd

Edition Christopher L. Aberson
https://textbookfull.com/product/applied-power-analysis-for-the-
behavioral-sciences-2nd-edition-christopher-l-aberson/
textboxfull.com
Statistics for the behavioral sciences Gravetter
https://textbookfull.com/product/statistics-for-the-behavioral-
sciences-gravetter/
textboxfull.com
Statistics for the Behavioral Sciences Frederick J.

Gravetter
sciences-frederick-j-gravetter/
textboxfull.com
Essential Statistics for the Behavioral Sciences Gregory
J. Privitera
https://textbookfull.com/product/essential-statistics-for-the-
behavioral-sciences-gregory-j-privitera/
textboxfull.com
Statistical Concepts for the Behavioral Sciences Harold O.

Kiess
https://textbookfull.com/product/statistical-concepts-for-the-
behavioral-sciences-harold-o-kiess/
textboxfull.com
Essentials of Statistics for the Behavioral Sciences

Frederick J. Gravetter
https://textbookfull.com/product/essentials-of-statistics-for-the-
behavioral-sciences-frederick-j-gravetter/
textboxfull.com
Statistics for the Behavioral Sciences 4th Edition Susan

A. Nolan
sciences-4th-edition-susan-a-nolan/
textboxfull.com
Introduction to Quantitative Data Analysis in the

Behavioral and Social Sciences Michael J. Albers
https://textbookfull.com/product/introduction-to-quantitative-data-
analysis-in-the-behavioral-and-social-sciences-michael-j-albers/
textboxfull.com
Multivariate Analysis
for the
Behavioral Sciences
Second Edition
Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)

Chapman & Hall/CRC
Statistics in the Social and Behavioral Sciences Series
Series Editors
Jeff Gill Steven Heeringa
Washington University, USA University of Michigan, USA
Wim J. van der Linden Tom Snijders

Pacific Metrics, USA University of Groningen, The Netherlands
Aims and scope
Large and complex datasets are becoming prevalent in the social and behavioral sciences and
statistical methods are crucial for the analysis and interpretation of such data. This series aims
to capture new developments in statistical methodology with particular relevance to
applications in the social and behavioral sciences. It seeks to promote appropriate use of
statistical, econometric and psychometric methods in these applied sciences by publishing a
broad range of reference works, textbooks and handbooks.
The scope of the series is wide, including applications of statistical methodology in sociology,
psychology, economics, education, marketing research, political science, criminology, public
policy, demography, survey methodology and official statistics. The titles included in the series
are designed to appeal to applied statisticians, as well as students, researchers and
practitioners from the above disciplines. The inclusion of real examples and case studies is
therefore essential.
Recently Published Titles
Handbook of International Large-Scale Assessment: Background, Technical Issues,

and Methods of Data Analysis
Leslie Rutkowski, Matthias von Davier, and David Rutkowski
Generalized Linear Models for Categorical and Continuous Limited Dependent Variables
Michael Smithson and Edgar C. Merkle
Incomplete Categorical Data Design: Non-Randomized Response Techniques for
Sensitive Questions in Surveys
Guo-Liang Tian and Man-Lai Tang
Handbook of Item Response Theory, Volume One: Models
Wim J. van der Linden
Handbook of Item Response Theory, Volume Two: Statistical Tools
Handbook of Item Response Theory, Volume Three: Applications
Computerized Multistage Testing: Theory and Applications
Duanli Yan, Alina A. von Davier, and Charles Lewis
Multivariate Analysis for the Behavioral Sciences, Second Edition
Kimmo Vehkalahti and Brian S. Everitt
For more information about this series, please visit: https://www.crcpress.com/go/ssbs

Multivariate Analysis
for the
Behavioral Sciences
Second Edition
Kimmo Vehkalahti
Brian S. Everitt

First edition published as Multivariable Modeling and Multivariate Analysis for the Behavioral
Sciences.
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
c 2019 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
International Standard Book Number-13: 978-0-8153-8515-8 (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, re-
produced, 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.
Library of Congress Cataloging-in-Publication Data
Names: Everitt, Brian, author. | Vehkalahti, Kimmo, author.

Title: Multivariate analysis for the behavioral sciences / Kimmo Vehkalahti &
Brian S. Everitt
Other titles: Multivariable modeling and multivariate analysis for the
behavioral sciences
Description: Second edition. | Boca Raton, Florida : CRC Press [2019] |
Earlier edition published as: Multivariable modeling and multivariate
analysis for the behavioral sciences / [by] Brian S. Everitt. | Includes
bibliographical references and index.
Identifiers: LCCN 2018041904| ISBN 9780815385158 (hardback : alk. paper) | ISBN
9781351202275 (e-book)
Subjects: LCSH: Social sciences—Statistical methods. | Multivariate analysis.
Classification: LCC HA31.35 .E94 2019 | DDC 519.5/35—dc23
LC record available at https://lccn.loc.gov/2018041904
Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com
and the CRC Press Web site at

http://www.crcpress.com

Dedication
Brian dedicates the book to the memory of his parents,

Emily Lucy Everitt and Sidney William Everitt.
Kimmo dedicates the book to Sirpa, the love of his life.

Libros de Estadística-Ciencia de Datos|Statistics-Data Science
Contents
Preface xiii
Preface to Multivariable Modeling and Multivariate Analysis
for the Behavioral Sciences xv
Authors xix
Acknowledgments xxi
1 Data, Measurement, and Models 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Types of Study . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Surveys . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.2 Experiments . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.3 Observational Studies . . . . . . . . . . . . . . . . . . 5
1.2.4 Quasi-Experiments . . . . . . . . . . . . . . . . . . . 6
1.3 Types of Measurement . . . . . . . . . . . . . . . . . . . . . 7
1.3.1 Nominal or Categorical Measurements . . . . . . . . 7
1.3.2 Ordinal Scale Measurements . . . . . . . . . . . . . . 8
1.3.3 Interval Scales . . . . . . . . . . . . . . . . . . . . . . 8
1.3.4 Ratio Scales . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.5 Response and Explanatory Variables . . . . . . . . . 10
1.4 Missing Values . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.5 The Role of Models in the Analysis of Data . . . . . . . . . 11
1.6 Determining Sample Size . . . . . . . . . . . . . . . . . . . 14
1.7 Significance Tests, p-Values, and Confidence Intervals . . . 16
1.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2 Looking at Data 23
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2 Simple Graphics—Pie Charts, Bar Charts, Histograms,
and Boxplots . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2.1 Categorical Data . . . . . . . . . . . . . . . . . . . . 24
2.2.2 Interval/Quasi-Interval Data . . . . . . . . . . . . . . 32
2.3 The Scatterplot and beyond . . . . . . . . . . . . . . . . . . 37
2.3.1 The Bubbleplot . . . . . . . . . . . . . . . . . . . . . 40
2.3.2 The Bivariate Boxplot . . . . . . . . . . . . . . . . . 42
2.4 Scatterplot Matrices . . . . . . . . . . . . . . . . . . . . . . 45
vii
viii Contents
2.5 Conditioning Plots and Trellis Graphics . . . . . . . . . . . 48

2.6 Graphical Deception . . . . . . . . . . . . . . . . . . . . . . 55
2.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3 Simple Linear and Locally Weighted Regression 63

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.2 Simple Linear Regression . . . . . . . . . . . . . . . . . . . 64
3.2.1 Fitting the Simple Linear Regression Model to the
Pulse Rates and Heights Data . . . . . . . . . . . . . 66
3.2.2 An Example from Kinesiology . . . . . . . . . . . . . 67
3.3 Regression Diagnostics . . . . . . . . . . . . . . . . . . . . . 69
3.4 Locally Weighted Regression . . . . . . . . . . . . . . . . . 73
3.4.1 Scatterplot Smoothers . . . . . . . . . . . . . . . . . 75
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4 Multiple Linear Regression 83

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.2 An Example of Multiple Linear Regression . . . . . . . . . 85
4.3 Choosing the Most Parsimonious Model When Applying
Multiple Linear Regression . . . . . . . . . . . . . . . . . . 90
4.3.1 Automatic Model Selection . . . . . . . . . . . . . . . 95
4.3.2 Example of Application of the Backward Elimination 96
4.4 Regression Diagnostics . . . . . . . . . . . . . . . . . . . . . 98
4.5 Multiple Linear Regression and Analysis of Variance . . . . 102
4.5.1 Analyzing the Fecundity of Fruit Flies by Regression 102
4.5.2 Multiple Linear Regression for Experimental Designs 104
4.5.3 Analyzing a Balanced Design . . . . . . . . . . . . . 105
4.5.4 Analyzing an Unbalanced Design . . . . . . . . . . . 106
4.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5 Generalized Linear Models 113

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.2 Binary Response Variables . . . . . . . . . . . . . . . . . . . 115
5.3 Response Variables That Are Counts . . . . . . . . . . . . . 117
5.3.1 Overdispersion and Quasi-Likelihood . . . . . . . . . 119
5.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6 Applying Logistic Regression 123

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
6.2 Odds and Odds Ratios . . . . . . . . . . . . . . . . . . . . . 123
6.3 Applying Logistic Regression to the GHQ Data . . . . . . . 125
6.4 Selecting the Most Parsimonious Logistic Regression Model 130

Contents ix
6.5 Driving and Back Pain: A Matched Case–Control Study . . 134

6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
6.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
7 Survival Analysis 139

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
7.2 The Survival Function . . . . . . . . . . . . . . . . . . . . . 140
7.2.1 Age at First Sexual Intercourse for Women . . . . . . 142
7.3 The Hazard Function . . . . . . . . . . . . . . . . . . . . . 144
7.4 Cox’s Proportional Hazards Model . . . . . . . . . . . . . . 146
7.4.1 Retention of Heroin Addicts in Methadone Treatment 149
7.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
7.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
8 Analysis of Longitudinal Data I: Graphical Displays and

Summary Measure Approach 155
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
8.2 Graphical Displays of Longitudinal Data . . . . . . . . . . . 157
8.3 Summary Measure Analysis of Longitudinal Data . . . . . . 159
8.3.1 Choosing Summary Measures . . . . . . . . . . . . . 160
8.3.2 Applying the Summary Measure Approach . . . . . . 162
8.3.3 Incorporating Pre-Treatment Outcome Values into the
Summary Measure Approach . . . . . . . . . . . . . 164
8.3.4 Dealing with Missing Values When Using the Sum-
mary Measure Approach . . . . . . . . . . . . . . . . 164
8.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
8.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
9 Analysis of Longitudinal Data II: Linear Mixed Effects

Models for Normal Response Variables 169
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
9.2 Linear Mixed Effects Models for Repeated Measures Data . 170
9.3 How Do Rats Grow? . . . . . . . . . . . . . . . . . . . . . . 174
9.3.1 Fitting the Independence Model to the Rat Data . . 174
9.3.2 Fitting Linear Mixed Models to the Rat Data . . . . 176
9.4 Computerized Delivery of Cognitive Behavioral
Therapy—Beat the Blues . . . . . . . . . . . . . . . . . . . 181
9.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
9.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
10 Analysis of Longitudinal Data III: Non-Normal Responses 189

10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
10.2 Marginal Models and Conditional Models . . . . . . . . . . 190
10.2.1 Marginal Models . . . . . . . . . . . . . . . . . . . . 190
10.2.2 Conditional Models . . . . . . . . . . . . . . . . . . . 194

x Contents
10.3 Using Generalized Estimating Equations

to Fit Marginal Models . . . . . . . . . . . . . . . . . . . . 196
10.3.1 Beat the Blues Revisited . . . . . . . . . . . . . . . . 196
10.3.2 Respiratory Illness . . . . . . . . . . . . . . . . . . . 197
10.3.3 Epilepsy . . . . . . . . . . . . . . . . . . . . . . . . . 201
10.4 Using Generalized Linear Mixed Effects Models
to Fit Conditional Models . . . . . . . . . . . . . . . . . . . 203
10.4.1 Respiratory Illness . . . . . . . . . . . . . . . . . . . 203
10.4.2 Epilepsy . . . . . . . . . . . . . . . . . . . . . . . . . 204
10.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
10.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
11 Missing Values 209

11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
11.2 Missing Data Mechanisms . . . . . . . . . . . . . . . . . . . 210
11.3 Dealing with Missing Values . . . . . . . . . . . . . . . . . . 212
11.4 Imputing Missing Values . . . . . . . . . . . . . . . . . . . . 213
11.5 Analyzing Multiply Imputed Data . . . . . . . . . . . . . . 215
11.6 Example of the Application of Multiple Imputation . . . . . 216
11.6.1 Complete-Case Analysis . . . . . . . . . . . . . . . . 217
11.6.2 Mean Imputation . . . . . . . . . . . . . . . . . . . . 218
11.6.3 Multiple Imputation . . . . . . . . . . . . . . . . . . 219
11.7 Beat the Blues Revisited (Again) . . . . . . . . . . . . . . . 219
11.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
11.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
12 Multivariate Data and Multivariate Analysis 225

12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
12.2 The Initial Analysis of Multivariate Data . . . . . . . . . . 226
12.2.1 Summary Statistics for Multivariate Data . . . . . . 226
12.2.2 Graphical Descriptions of the Body
Measurement Data . . . . . . . . . . . . . . . . . . . 229
12.3 The Multivariate Normal Probability Density Function . . . 230
12.3.1 Assessing Multivariate Data for Normality . . . . . . 233
12.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
12.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
13 Principal Components Analysis 239

13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
13.2 Principal Components Analysis (PCA) . . . . . . . . . . . . 239
13.3 Finding the Sample Principal Components . . . . . . . . . . 241
13.4 Should Principal Components be Extracted from
the Covariance or the Correlation Matrix? . . . . . . . . . . 244
13.5 Principal Components of Bivariate Data
with Correlation Coefficient r . . . . . . . . . . . . . . . . . 246

Contents xi
13.6 Rescaling the Principal Components . . . . . . . . . . . . . 248

13.7 How the Principal Components Predict
the Observed Covariance Matrix . . . . . . . . . . . . . . . 248
13.8 Choosing the Number of Components . . . . . . . . . . . . 249
13.9 Calculating Principal Component Scores . . . . . . . . . . . 251
13.10 Some Examples of the Application of PCA . . . . . . . . . 252
13.10.1 Head Size of Brothers . . . . . . . . . . . . . . . . . 252
13.10.2 Crime Rates in the United States . . . . . . . . . . . 255
13.10.3 Drug Usage by American College Students . . . . . 260
13.11 Using PCA to Select a Subset of the Variables . . . . . . . 264
13.12 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
13.13 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
14 Multidimensional Scaling and Correspondence Analysis 267

14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
14.2 Multidimensional Scaling . . . . . . . . . . . . . . . . . . . 269
14.2.1 Classical Multidimensional Scaling . . . . . . . . . . 270
14.2.2 Connection to Principal Components . . . . . . . . . 273
14.2.3 Road Distances in Finland . . . . . . . . . . . . . . . 274
14.2.4 Mapping Composers of Classical Music . . . . . . . . 278
14.2.5 Nonmetric Multidimensional Scaling . . . . . . . . . 280
14.2.6 Re-mapping Composers of Classical Music . . . . . . 281
14.3 Correspondence Analysis . . . . . . . . . . . . . . . . . . . 284
14.3.1 Simple Example of the Application
of Correspondence Analysis . . . . . . . . . . . . . . 286
14.3.2 Connections of Work Activities and Job Advantages 288
14.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
14.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
15 Exploratory Factor Analysis 295

15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
15.2 The Factor Analysis Model . . . . . . . . . . . . . . . . . . 296
15.3 Estimating the Parameters in the Factor Analysis Model . . 299
15.4 Determining the Number of Factors . . . . . . . . . . . . . 301
15.5 Fitting the Factor Analysis Model: An Example . . . . . . . 302
15.6 Rotation of Factors . . . . . . . . . . . . . . . . . . . . . . . 304
15.6.1 A Simple Example of Graphical Rotation . . . . . . . 306
15.6.2 Numerical Rotation Methods . . . . . . . . . . . . . 309
15.6.3 A Simple Example of Numerical Rotation . . . . . . 311
15.7 Estimating Factor Scores . . . . . . . . . . . . . . . . . . . 311
15.7.1 Analyzing the Crime Rates by Factor Analysis . . . . 312
15.8 Exploratory Factor Analysis and Principal Component
Analysis Compared . . . . . . . . . . . . . . . . . . . . . . . 315
15.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
15.10 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317

xii Contents
16 Confirmatory Factor Analysis and Structural Equation

Models 319
16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
16.2 Estimation, Identification, and Assessing the Fit for Confir-
matory Factor Analysis and Structural Equation Models . . 320
16.2.1 Estimation . . . . . . . . . . . . . . . . . . . . . . . . 320
16.2.2 Identification . . . . . . . . . . . . . . . . . . . . . . 321
16.2.3 Assessing the Fit . . . . . . . . . . . . . . . . . . . . 322
16.3 Examples of Confirmatory Factor Analysis . . . . . . . . . . 324
16.3.1 Ability and Aspiration . . . . . . . . . . . . . . . . . 325
16.3.2 Drug Usage among Students . . . . . . . . . . . . . . 327
16.4 Eight Factors of Systems Intelligence . . . . . . . . . . . . . 331
16.4.1 Testing the Factorial Validity of the SI Inventory . . 333
16.5 Structural Equation Models . . . . . . . . . . . . . . . . . . 335
16.5.1 Example of a Structural Equation Model . . . . . . . 335
16.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
16.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
17 Cluster Analysis 341

17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
17.2 Cluster Analysis . . . . . . . . . . . . . . . . . . . . . . . . 343
17.3 Agglomerative Hierarchical Clustering . . . . . . . . . . . . 344
17.3.1 Clustering Individuals Based on Body Measurements 347
17.3.2 Clustering Countries on the Basis of Life Expectancy 348
17.4 k-Means Clustering . . . . . . . . . . . . . . . . . . . . . . . 352
17.4.1 Clustering Crime Rates . . . . . . . . . . . . . . . . . 355
17.5 Model-Based Clustering . . . . . . . . . . . . . . . . . . . . 356
17.5.1 Clustering European Countries . . . . . . . . . . . . 359
17.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
17.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
18 Grouped Multivariate Data 365

18.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
18.2 Two-Group Multivariate Data . . . . . . . . . . . . . . . . . 366
18.2.1 Hotelling’s T 2 Test . . . . . . . . . . . . . . . . . . . 366
18.2.2 Fisher’s Linear Discriminant Function . . . . . . . . 369
18.3 More Than Two Groups . . . . . . . . . . . . . . . . . . . . 374
18.3.1 Multivariate Analysis of Variance (MANOVA) . . . . 374
18.3.2 Classification Functions . . . . . . . . . . . . . . . . . 378
18.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382
18.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382
References 385
Index 401

Preface
In some respects this book is a second edition of Multivariable Modeling and

Multivariate Analysis for the Behavioral Sciences but in others it is largely
a new book with new chapters on missing values and the analysis of longi-
tudinal data where the response variable can not be assumed to be normally
distributed. The book also includes a wider account of generalized linear mod-
els as well as a new chapter on multidimensional scaling and correspondence
analysis and separate chapters on exploratory and confirmatory factor analy-
sis; in the latter there is also new coverage of structural equation models. A
number of interesting, new examples and exercises have been added in several
chapters.
The original lengthy title tried to explain that the book covered situations
where the variables of interest consisted of a response variable and explana-
tory variables and where interest lies in finding suitable models relating the
response to the explanatory variables, hence multivariable modeling, in addi-
tion to techniques that can be applied to what has historically been termed
multivariate data where there is no division of the variables and the aim is to
find a parsimonious description of the structure of the data. The current book
contains extended coverage of both situations but has been given a somewhat
shorter title, Multivariate Analysis for the Behavioral Sciences, because both
types of data mentioned can be represented symbolically by an n × q matrix
X containing the q variable values for n number of units (often subjects) in
the data set. Chapters 3–11 will describe methods for dealing with data when
one of the q variables is a response and the others explanatory, while the re-
maining chapters deal with methods for the analysis of data sets where there
is no such division of the variables. We hope this makes it clear that this book
covers accounts of a wider range of statistical methodology than covered in
the conventional ‘multivariate analysis’ textbook.
The ‘we’ opening the last sentence above brings us to the most important
change between this book and that mentioned in the first line of this Preface,
namely the arrival of a co-author Dr. Kimmo Vehkalahti. It is Kimmo who
is responsible for many of the changes and most of the new material in this
book compared with the original on which it is based.
Most chapters include ‘technical detail’ sections which briefly describe the
theory behind the methods of concern in the chapter. These sections often
require a familiarity with some relatively advanced mathematics, for exam-
ple, matrix algebra, for their understanding (see Puntanen et al., 2011, 2013).
Readers without the necessary grounding in maths can largely ignore such sec-
xiii
xiv Preface
tions and instead concentrate on the examples given to help them understand
the practical implications of the methods.
There are exercises at the end of each chapter, some of which are ‘starred’
(*) to indicate that they are more challenging and could perhaps be used as
the basis of student projects. Data sets for most exercises are not given in the
text but are available on the associated web sites (see later) where they are
identified by the relevant exercise number. For the starred exercises the web
sites also contain pointers to the appropriate analysis.
The web site for the book is www.crcpress.com and in addition the book
has a GitHub repository (https://github.com/KimmoVehkalahti/MABS) for
distributing the complete data sets and the R code for reproducing the exam-
ples and for answering the exercises; consequently this allows us to abbreviate
the listings of most data sets in the text.
We hope that this book will be found useful in a number of different ways,
including:
• As the main part of a formal statistics course for advanced undergraduates

and postgraduates in all areas of the behavioral sciences,
• As a supplement to an existing course,
• For self-study,
• For researchers in the behavioral sciences undertaking statistical analyses
on their data,
• For statisticians teaching statistics to psychologists and others,
• For statisticians using R when teaching intermediate statistics courses
both in the behavioral sciences and in other areas.
Brian S. Everitt
Dulwich, London
Kimmo Vehkalahti
Vuosaari, Helsinki

Preface to Multivariable Modeling
and Multivariate Analysis for the
Behavioral Sciences
The Encyclopedia of Statistics in Behavioral Science (Everitt and Howell,

2005) opens with the following paragraph:
Forty years ago there was hardly a field called “behavioral science.” In
fact, psychology largely was the behavioral sciences, with some help
from group theory in sociology and decision making in economics. Now,
of course, psychology has expanded and developed in a myriad of ways,
to the point where behavioral science is often the most useful term.
Physiological psychology has become neuroscience, covering areas not
previously part of psychology. Decision-making has become decision
science, involving people from economics, marketing, and other disci-
plines. Learning theory has become cognitive science, again exploring
problems that were not even considered 40 years ago. And develop-
ments in computing have brought forth a host of new techniques that
were not possible in the days of manual and electronic calculators.
With all these changes, there have been corresponding changes in the
appropriate statistical methodologies.
Despite the changes mentioned in the last sentence of this quotation, many
statistical books aimed at psychologists and others working in the behavioral
sciences continue to cover primarily simple hypothesis testing, using a variety
of parametric and nonparametric significance tests, simple linear regression,
and analysis of variance. Such statistical methodology remains important in
introductory courses, but represents only the first step in equipping behavioral
science students with enough statistical tools to help them on their way to
success in their later careers. The aim of this book is to encourage students
and others to learn a little more about statistics and, equally important, how
to apply statistical methods in a sensible fashion. It is hoped that the following
features of the text will help it reach its target:
• The central theme is that statistics is about solving problems; data relevant
to these problems are collected and analyzed to provide useful answers. To
this end, the book contains a large number of real data sets arising from
real problems. Numerical examples of the type that involve the skiing
activities of belly dancers and politicians are avoided as far as possible.
xv
xvi Preface to Multivariable Modeling and Multivariate Analysis
• Mathematical details of methods are confined to numbered and separated

Technical Sections. For the mathematically challenged, the most difficult
of these displays can, at least as a last resort, be ignored. But the study of
the relevant mathematical material (which on occasion will include the use
of vectors and matrices) will undoubtedly help in the reader’s appreciation
of the corresponding technique.
• Although many statistical methods require considerable amounts of arith-
metic for their application, the burden of actually performing the neces-
sary calculations has been almost entirely removed by the development
and wide availability of powerful and relatively cheap personal computers
and associated statistical software packages. It is assumed, therefore, that
all students will be using such tools when undertaking their own analyses.
Consequently, arithmetic details are noticeable largely by their absence,
although a little arithmetic is included where it is considered helpful in
explaining a technique.
• There are many challenging data sets both in the text and in the exercises
provided at the end of each chapter. All data sets, both in the body of
the text and in the exercises, are given on the Web site associated with
the book, as are the answers to all the exercises. (Because the major-
ity of data sets used in the book are available on the book’s Web site
(http://www.crcpress.com/product/isbn/9781439807699), tables of data
in the text only give a small subset of each data set.)
As mentioned in the penultimate bullet point above, the text assumes that
readers will be using one or another of the many available statistical software
packages for data analysis. This raises the thorny question for the author of
what information should be provided in the text about software. Would, for
example, screen dumps from SPSS be useful, or listings of STATA code? Per-
haps, but neither are included here. Instead, all the computer code used to
analyze the many examples to be found in the text is given on the book’s
Web site, and this code is in the R language, where R is a software system for
statistical computing, data analysis, and graphics. This may appear a strange
choice for a book aimed at behavioral scientists, but the rationale behind the
choice is first that the author uses R in preference to other statistical soft-
ware, second that R can be used to produce many interesting and informative
graphics that are difficult if not impossible to produce with other software,
third that R is free and can be easily downloaded by students, and fourth, R
has a very active user community and recently developed statistical methods
become available far more quickly than they do with other packages. The only
downside with R is that it takes a little more time to learn than say using
“point-and-click” SPSS. The initial extra effort, however, is rapidly rewarded.
A useful book for learning more about R is Everitt and Hothorn (2009).
The material covered in the book assumes the reader is familiar with the
topics covered in introductory statistics courses, for example, population, sam-

Preface to Multivariable Modeling and Multivariate Analysis xvii
ple, variable, parameter, significance test, p-value, confidence interval, correla-

tion, simple regression, and analysis of variance. The book is primarily about
methods for analyzing data but some comments are made in Chapter 1 about
the various types of study that behavioral researchers may use and their de-
sign. And it is in Chapter 1 that the distinction between multivariable and
multivariate—both of which appear in the book’s title—will be explained.
It is hoped that the text will be useful in a number of different ways,
including:
• As the main part of a formal statistics course for advanced undergraduates

and postgraduates in all areas of the behavioral sciences.
• As a supplement to an existing course.
• For self-study.
• For researchers in the behavioral sciences undertaking statistical analyses
on their data.
• For statisticians teaching statistics to psychologists and others.
• For statisticians using R when teaching intermediate statistics courses
both in the behavioral sciences and in other areas.
B. S. Everitt
Dulwich, U.K.

Authors
Kimmo Vehkalahti is a fellow of the Teachers’ Academy, University of

Helsinki, Finland. He has been a part of the faculty of Social Sciences for
over 25 years, currently as senior lecturer of the Social Data Science in the
Centre for Research Methods. He is author of a Finnish textbook on measure-
ment and survey methods. The present book is his first international textbook
on statistics. His research and teaching activities are related to open data
science, multivariate analysis, and introductory statistics. His spare time is
divided (unequally) between jogging and trail running, reading, watching ice
hockey, holidays with his wife, and singing tenor in choir.
Brian S. Everitt is professor emeritus, King’s College, London, UK. He

worked at the Institute of Psychiatry, University of London for over 35 years,
finally as head of the Biostatistics and Computing Department and profes-
sor of behavioural statistics. He is author or co-author of over 70 books on
statistics and approximately 100 papers and other articles, and was a section
editor for the Encyclopedia of Biostatistics, published by Wiley. In retirement,
he divides his time between working as editor-in-chief of Statistical Methods
in Medical Research, playing tennis, watching cricket, long walking holidays
with his wife, and playing classical guitar in private.
xix
Acknowledgments
Kimmo would like to thank the faculty of Social Sciences at the University
of Helsinki for granting him a six-month sabbatical without which this book
could not have been written. Three months of this sabbatical were spent at the
Universitat Pompeu Fabra in Barcelona at the kind invitation and hospitality
of Professor Michael Greenacre. The hard work needed to get to grips with
R and LATEX was greatly aided by the sharing of an office (and coffee breaks)
with Michael at Campus de la Ciutadella. Gràcies!
Kimmo’s thanks are also given to Maria Anna Donati, Ulla Palotie, Juha
Törmänen, Raimo Hämäläinen, Esa Saarinen, and Seppo Mustonen, all of
whom gave their permission to use their interesting data sets in this book;
he has the warmest memories of staying in the beautiful home of the lovely
Firenze family. Grazie Caterina Primi, Michele e Luisa!
Thanks are given to Dr. Deepayan Sarkar, the author of Lattice: Multi-
variate Data Visualization with R, and Springer, the publishers of the book,
for permission to use Figures 2.8, 2.9, 4.5, 4.6, 4.7, and 5.16 in Chapter 2 of
this book.
Brian and Kimmo thank each other for an extremely smooth collaboration
that allowed this book to be completed on time whilst allowing them to become
and remain good friends; they also agree that sincere thanks are given to
Rob Calver of Taylor & Francis, who has been, as always, supportive and
encouraging during the writing of the book.
xxi
1
Data, Measurement, and Models
1.1 Introduction
Statistics is a general intellectual method that applies wherever data,
variation, and chance appear. It is a fundamental method because data,
variation and chance are omnipresent in modern life. It is an indepen-
dent discipline with its own core ideas, rather than, for example, a
branch of mathematics . . . Statistics offers general, fundamental and
independent ways of thinking.
Journal of the American Statistical Association
Quintessentially, statistics is about solving problems; data (measurements or

observations) relevant to these problems are collected, and statistical analyses
are used to provide useful answers. But the path from data collection to anal-
ysis and interpretation is often not straightforward. Most real-life applications
of statistical methodology have one or more nonstandard features, meaning
in practice that there are few routine statistical questions, although there are
questionable statistical routines. Many statistical pitfalls lie in wait for the
unwary. Indeed, statistics is perhaps more open to misuse than most other
subjects, particularly by the nonstatistician with access to powerful statistical
software. The misleading average, the graph with “fiddled axes,” the inappro-
priate p-value, and the linear regression fitted to nonlinear data are just four
examples of horror stories that are part of statistical folklore.
Statisticians often complain that many of those working in the behavioral
sciences put undue faith in significance tests, use complex methods of analysis
when the data merit only a relatively simple approach, and sometimes abuse
the statistical techniques they are employing. Statisticians become upset (and
perhaps feel a little insecure) when their advice to, say, “plot a few simple
graphs,” is ignored in favor of a multivariate analysis of covariance or similar
statistical extravagance.
However, if statisticians are at times horrified by the way in which be-
havioral scientists apply statistical techniques, behavioral scientists may be
no less horrified by many statisticians’ apparent lack of awareness of what
stresses behavioral research can place on an investigator. A statistician may,
for example, demand a balanced design with 30 subjects in each cell so as to
1
2 Multivariate Analysis for the Behavioral Sciences
achieve some appropriate power for the analysis. But it is not the statistician
who is faced with the frustration caused by a last-minute phone call from a
subject who cannot take part in an experiment that has taken several hours
to arrange. Again, the statistician advising on a longitudinal study may call
for more effort in carrying out follow-up interviews so that the study avoids
statistical problems produced by the presence of missing data. It is, however,
the behavioral researcher who must continue to persuade people to talk about
potentially distressing aspects of their lives, who must confront possibly dan-
gerous respondents, or who arrives at a given (and often remote) address to
conduct an interview, only to find that the person is not at home. Many statis-
ticians often do not appear to appreciate the complex stories behind each data
point in many behavioral studies. One way of improving the possible commu-
nication problems between behavioral scientist and statistician is for each
to learn more about the language of the other. There is already available a
plethora of, for example, “Statistics for Psychologists” books, but sadly, (as
far as we know) no “Psychology for Statisticians” equivalent. Perhaps there
should be?
Having outlined briefly a few caveats about the possible misuse of statistics
and the equally possible conflict between statistician and behavioral scientist,
it is time to move on to consider some of the basics of behavioral science
studies and their implications for statistical analysis.
1.2 Types of Study

It is said that, when Gertrude Stein lay dying, she roused briefly and asked her
assembled friends, “Well, what’s the answer?” They remained uncomfortably
quiet, at which she sighed, “In that case, what’s the question?”
Research in the behavioral science, as in science in general, is about search-
ing for the answers to particular questions of interest. Do politicians have
higher IQs than university lecturers? Do men have faster reaction times than
women? Should phobic patients be treated by psychotherapy or by a behav-
ioral treatment such as flooding? Do children who are abused have more prob-
lems later in life than children who are not abused? Do children of divorced
parents suffer more marital breakdowns themselves than children from more
stable family backgrounds?
In more general terms, scientific research involves a sequence of asking and
answering questions about the nature of relationships among variables (e.g.,
How does A affect B? Do A and B vary together? Is A significantly different
from B? and so on). Scientific research is carried out at many levels that
differ in the types of question asked and therefore in the procedures used to
answer them. Thus, the choice of which methods to use in research is largely
determined by the kinds of questions that are asked.

Data, Measurement, and Models 3
Of the many types of investigation used in behavioral research, the most

common are perhaps the following:
• Surveys
• Experiments
• Observational studies
• Quasi-experiments
Some brief comments about each of these four types are given below; a more
detailed account is available in the papers by Stretch, Raulin, and Graziano,
and by Dane, all of which appear in the second volume of the excellent Com-
panion Encyclopedia of Psychology (see Colman, 1994).
1.2.1 Surveys
Survey methods are based on the simple discovery that “asking questions is a
remarkably efficient way to obtain information from and about people” (Schu-
man and Kalton, 1985, p. 635). Surveys involve an exchange of information
between researcher and respondent; the researcher identifies topics of interest,
and the respondent provides knowledge or opinion about these topics. Depend-
ing upon the length and content of the survey as well as the facilities available,
this exchange can be accomplished via written questionnaires, in-person in-
terviews, or telephone conversations; and, in the 21st century, surveys via the
Internet are increasingly common.
Surveys conducted by behavioral scientists are usually designed to elicit
information about the respondents’ opinions, beliefs, attitudes, and values.
Perhaps one of the most famous surveys of the 20th century was that con-
ducted by Alfred Charles Kinsey, a student of human sexual behavior in the
1940s and 1950s. The first Kinsey report, Sexual Behavior in the Human Male,
appeared in 1948 (see Kinsey et al., 1948), and the second, Sexual Behavior
in the Human Female, in 1953 (see Kinsey et al., 1953). It is no exaggeration
to say that both reports caused a sensation, and the first quickly became a
bestseller.
Surveys are often a flexible and powerful approach to gathering information
of interest, but careful consideration needs to be given to several aspects of
the survey if the information is to be accurate, particularly when dealing
with a sensitive topic. Having a representative sample, having a large-enough
sample, minimizing nonresponse, and ensuring that the questions asked elicit
accurate responses are just a few of the issues that the researcher thinking of
carrying out a survey needs to consider. Readers are referred to Bradburn et
al. (2004) and Tourangeau et al. (2000) for a broad coverage of practical advice
for questionnaire construction and Laaksonen (2018), Groves et al. (2009), de
Leeuw et al. (2008) and Lehtonen and Pahkinen (2004) for detailed accounts
of survey sampling and survey methodology.

Examples of data collected in surveys and their analysis are given in several
later chapters.
1.2.2 Experiments
According to Sir Ronald Fisher, perhaps the greatest statistician of the 20th
century, “experiments are only experience carefully planned in advance and
designed to form a secure basis of new knowledge.” The essential feature of an
experiment is the large degree of control in the hands of the experimenters,
and in designed experiments the goal is to allow inferences to be drawn about
the effects of an intervention of interest that are logically compelled by the
data and hence allow assessment of a causal relationship. In many cases the
“intervention” will be some form of therapy in which case the experiment is
usually called a clinical trial.
In an experiment, the researcher controls the manner in which subjects are
allocated to the different levels of the experimental factors. In a comparison of
a new treatment with one used previously, for example, the researcher would
have control over the scheme for allocating subjects to the two treatments.
The manner in which this control is exercised is of vital importance if the
results of the experiment are to lead to a largely unambiguous assessment of
the effect of treatment. The objective in allocation is that the groups to be
compared should be alike in all respects except the intervention (treatment)
received. Comparable groups prior to the intervention ensure that differences
in outcomes after the intervention reflect effects of the intervention in an
unbiased fashion. Let us begin by considering two flawed allocation procedures
that are unlikely to achieve the desired degree of similarity of the two groups.
• Perhaps the first subjects to volunteer to take part in the experiment

should all be given the new treatment, for example, and the later ones
the old treatment? The two groups formed in this way may differ in level
of motivation and so subsequently in performance. Observed treatment
differences would be confounded with differences produced by the alloca-
tion procedure. Alternatively, early volunteers might be more seriously ill,
those desperate to find a new remedy that works, and again, this might
lead to a bias in the measured difference between the two treatments.
• So what about putting alternate subjects into each group? The objection to
this is that the experimenter will know who is receiving what treatment
and may be tempted to “tinker” with the scheme to ensure that those
patients who are most ill receive the new treatment.
So, how should we form the groups that will be used to assess an experi-
mental intervention? The answer is deceptively simple—use randomization.
The group to which a participant in the experiment is allocated is decided
by chance. It could be arranged by flipping a coin each time a new eligi-
ble patient arrives, and allocating the patient to the new treatment if the

Data, Measurement, and Models 5
result is a head, or to the old treatment if a tail appears. In practice, of

course, a more sophisticated randomization procedure will be used. The es-
sential feature, however, is randomization, rather than the mechanism used to
achieve it. Randomization was introduced into scientific experiments far more
recently, when in 1926 Fisher randomly assigned individual blocks or plots of
land in agricultural experiments to receive particular types of “treatment”—
different amounts of fertilizer. The primary benefit that randomization has
is the chance (and therefore impartial) assignment of extraneous influences
among the groups to be compared, and it offers this control over such influ-
ences whether or not they are known by the experimenter to exist. Note that
randomization does not claim to render the two samples equal with regard to
these influences; if, however, the same procedure was applied to repeated sam-
ples from the population, equality would be achieved in the long run. Thus,
randomization ensures a lack of bias, whereas other methods of assignment
may not. In a properly conducted, randomized, experiment the interpretation
of an observed group difference is largely unambiguous; its cause is very likely
to be the different treatments or conditions received by the groups.
Several of the data sets introduced and analyzed in later chapters arise
from experimental studies, often clinical trials.
1.2.3 Observational Studies

Suppose a researcher is interested in investigating how smoking cigarettes
affects a person’s systolic blood pressure. Using the experimental approach
described earlier, people would have to be allocated at random to two groups,
the members of one group being asked to smoke some quantity of cigarettes
per day, and the members of the other group required not to smoke at all.
Clearly, no ethical committee would approve of such a study. So, what can be
done? An approach that would get ethical approval is to measure the systolic
blood pressure of naturally occurring groups of individuals who smoke, and
those who do not, and then compare the results. This would then be what is
known as an observational study, defined by Cochran (1965) as follows:
An empiric comparison of “treated” and “control” groups in which the
objective is to elucidate cause-and-effect relationships but where it is
not possible to use controlled experimentation, in the sense of being
able to impose the procedures or treatments whose effects it is desired
to discover, or to assign patients at random to different procedures.
Many observational studies involve recording data on the members of naturally
occurring groups, generally over a period of time, and comparing the rate
at which a particular event of interest occurs in the different groups (such
studies are often referred to as prospective). If, for example, an investigator
was interested in the health effects of a natural disaster such as an earthquake,
those who experienced the earthquake could be compared, on some outcome
variable of interest, with a group of people who did not.

Another commonly used type of observational study is the case-control

investigation. Here, a group of people (the cases) all having a particular char-
acteristic (a certain disease perhaps) are compared with a group of people
who do not have the characteristic (the controls), in terms of their past expo-
sure to some event or risk factor. The cases and controls are usually matched
one-to-one for possible confounding variables. An example of such a study is
reported in Lehman et al. (1987). Here the researchers collected data follow-
ing the sudden death of a spouse or a child in a car crash. They matched 80
bereaved spouses and parents to 80 controls drawn from 7582 individuals who
came to renew their driver’s license. Specifically, they matched for gender, age,
family income before crash, education level, and number and ages of children.
The types of analyses suitable for observational studies are often the same
as those used for experimental studies. Unlike experiments, however, the lack
of control over the groups to be compared in an observational study makes
the interpretation of any difference between the groups detected in the study
open to a variety of interpretations. In the smoking and systolic blood pressure
study, for example, any difference found between the blood pressures of the
two groups would be open to three possible interpretations:
• Smoking causes a change in systolic blood pressure.
• Level of blood pressure has a tendency to encourage or discourage smoking.
• Some unidentified factors play a part in determining both the level of
blood pressure and whether or not a person smokes.
In the design of an observational study, an attempt is made to reconstruct
some of the structure and strengths of an experiment. But the possible am-
biguity in interpretation of the results from an observational study, however
well designed, means that the observational approach is not as powerful as a
designed experiment. A detailed account of observational studies is given in
Rosenbaum (2002).
1.2.4 Quasi-Experiments
Quasi-experimental designs resemble experiments proper but are weak on
some of the characteristics. In particular (and as in the observational study),
the ability to manipulate the groups to be compared is not under the inves-
tigator’s control. But, unlike the observational study, the quasi-experiment
involves the intervention of the investigator in the sense that he or she applies
a variety of different “treatments” to naturally occurring groups. In investi-
gating the effectiveness of three different methods of teaching mathematics to
15 year olds, for example, a method might be given to all the members of a
particular class in a school. The three classes that receive the different teach-
ing methods would be selected to be similar to each other on most relevant
variables, and the methods would be assigned to classes on a chance basis.
For more details of quasi-experiments see Shadish et al. (2002).

Exploring the Variety of Random
Documents with Different Content
only grossly unfair to the woman worker, but it does not tally with
fact. A fine example of the kind of defence for the practice of
sweating women workers that can be based on this assumption is
quoted by the Women’s Bureau from an unnamed commercial
magazine. “Eighty-six per cent of women workers,” runs this
masterpiece of sophistry, “live at home or with relatives. [So, in all
likelihood, do eighty-six per cent of male workers.] It is immaterial in
these cases whether the earnings of each measure up to the cost of
living scheduled for a single woman living alone, so that the theory of
the need of a sufficient wage to support a single woman living alone
does not apply to eighty-six per cent of the entire population [sic].”
This quotation, says the Bureau, is typical of the attitude of the
employer who pays his women employees less than a living wage on
the plea that they live at home and therefore have few expenses. It is
equally remarkable in its ruthless disregard of the just claim of the
woman worker to the same share in the product of her toil that the
male worker is allowed; and in its disregard of the fact that so long
as eighty-six per cent of women workers are forced to accept a
starvation-wage because they live at home, the other fourteen per
cent who do not live at home will be forced by the pressure of
competition to accept the same starvation-wage. The question how
this fourteen per cent will eke out a living—whether through
overwork, begging or prostitution—does not of course concern the
employer; for it is one of the striking differences between chattel-
slavery and wage-slavery that the owner of the wage-slave is under
no obligation to keep his workers from starving. That is, presumably,
their own lookout.
If employers are not given to concerning themselves with this
question, however, communities are. Thirteen States have enacted
laws fixing a minimum wage for women, three have fixed minimum
wages in specified occupations, one has fixed a minimum wage
which its industrial welfare commission has power to change, and
nine have created boards or commissions with power to fix minimum
wage-rates. It may be noted that in those States where the rate is
fixed by law, it has not responded to the rising cost of living. In Utah
and Arkansas, for example, the minimum wage for an experienced
woman is $7.50 a week. There is constant effort by interested
individuals and organizations to get similar laws enacted in other
States, in spite of the fact that in 1923 the Supreme Court of the
United States declared unconstitutional the minimum wage-law of
the District of Columbia. Such efforts, of course, are in reality efforts
to secure class-legislation, as are all attempts to secure special
enactments designed to benefit or protect women.
Of such enactments there is an ever increasing number. So rapidly
do they increase, indeed, that women may be said to be in a fair way
to exchange the tyranny of men for that of organized uplift. They are
sponsored by those well-meaning individuals who deplore social
injustice enough to yearn to mitigate its evil results, but do not
understand it well enough to attack its causes; by women’s
organizations whose intelligence is hardly commensurate with their
zeal to uplift their sex; and by men’s labour-organizations which are
quite frankly in favour of any legislation that will lessen the chances
of women to compete with men in the labour-market.[27] Given the
combined suasion of these forces, and the inveterate sentimentalism
which makes it hard for legislators to resist any plea on behalf of “the
women and children,” almost anything in the way of rash and ill-
considered legislation is possible, and even probable. There is on
the statute-books of the various States an imposing array of laws
designed to “protect” women workers. There are only four States
which do not in some way limit the hours of work for women; there
are eleven which limit the number of successive days that they may
work; fourteen have fixed the amount of time that shall be allowed
them for their midday meal; twelve have ruled that a woman may
work only a given number of hours without a rest-period. Sixteen
States prohibit night-work in certain industries or occupations; two
limit her hours of night-work to eight. There is also a tendency to
extend to women special protection against the hazards of industry.
In seventeen States the employment of women in mines is
prohibited. Two States prohibit their employment in any industry
using abrasives. In four States they are not allowed to oil moving
machinery. Three regulate their employment in core-making; and
four regulate the amount of the weight that they may be required to
lift—the maximum ranging, oddly enough, from fifteen pounds in
Ohio and Pennsylvania to seventy-four pounds in Massachusetts. In
addition to those regulations which prohibit women from working in
certain occupations or under certain conditions, “each State,” says
the Women’s Bureau, “has many laws and rulings which prescribe
the conditions under which women should work, covering such
matters as the lifting of weights, provision of seats, and proper
provision for sanitation and comfort.” In six States, industrial
commissions have power to make regulations for the health and
welfare of workers. In three, the commissions have power to make
regulations for women and minors only, and in one, for women,
minors, learners, and apprentices.
Perhaps the most striking thing about all these multiform regulations
governing the employment of women is the amount of misplaced
zeal that they denote. “In most cases,” says the Women’s Bureau,
“the laws which prohibit their employment have little bearing on the
real hazards to which they are exposed.... Prohibiting the
employment of women on certain dusty processes does not solve
the problem of any industrial disease in a community. Men are also
liable to contract pulmonary diseases from exposure to dusts.... It is
very possible that under the guise of ‘protection’ women may be shut
out from occupations which are really less harmful to them than
much of the tedious, heavy work both in the home and in the factory
which has long been considered their special province. Safe
standards of work for women must come to be safe standards for
men also if women are to have an equal chance in industry.” The
italics are mine. It is worth mentioning here that only two States
prohibit the employment of women in the lead-industry, which so far
is the only one that has been proved more harmful to women than to
men. The mass of legislation and regulation designed to protect
women from the fatigues and hazards of industry would seem, then,
to have been animated more by chivalry than by scientific
knowledge; and while chivalry may be all very well in its place, it can
hardly be expected to solve the industrial problem of women.
In connexion with so-called welfare-legislation, it is interesting to
observe that women and children are customarily grouped together
as classes requiring protection; and that various laws affecting their
position in industry have been sanctioned by the courts as being for
the good of the race and therefore not to be regarded as class-
legislation. Such decisions certainly would appear to be reasonable
in so far as they apply to children, who are the rising generation of
men and women, and should be protected during their immaturity.
But they can be held valid as they affect women only if woman is
regarded as primarily a reproductive function. This view, apparently,
is held by most legislators, courts, and uplifters; and they have an
unquestionable right to hold it. Whether, however, they are just in
attempting to add to the burdens of the working woman by imposing
it upon her in the form of rules that restrict her opportunities, is
another question. One thing is certain: if discriminative laws and
customs are to continue to restrict the opportunities of women and
hamper them in their undertakings, it makes little difference for
whose benefit those laws and customs are supposed to operate,
whether for the benefit of men, of the home, of the race, or of women
themselves; their effect on the mind of woman and her opportunities,
will be the same. While society discriminates against her sex, for
whatever reason, she can not be free as an individual.
Should nothing, then, be done to protect women from the disabilities
and hazards to which they are subject in the industrial world? Better
nothing, perhaps, than protection which creates new disabilities.
Laws which fix fewer hours of work for women than for men may
result in shortening men’s hours also in factories where many
women are employed; but they may result in the substitution of men
—or children—for women in factories where but few have been
employed. Laws prohibiting night-work may reduce the chances of
women to get much-needed employment, and may sometimes shut
them out of work which would offer higher returns on their labour
than anything they might get to do during the day—as, for example,
night-work in restaurants, where the generous tips of after-theatre
patrons add considerably to the earnings of waiters. Moreover, it is
hard to see on what ground night-work could be held to be more
harmful for women than for men. Minimum-wage laws may fix a legal
limit to the greed of employers, but they can not prevent the
underpayment of women workers, for they are based on theoretical
notions of a living wage, and have no relation to the actual value of
the individual’s labour. Where they are fixed by law, as I have
remarked, a rise in the cost of living may render them ineffectual. As
for those laws which undertake to protect women against the
hazards of industry, they have usually, as the Women’s Bureau has
shown, very little relation to the hazards to which women are actually
exposed; but they constitute a real barrier to industrial opportunity.
On the whole, the vast and unwieldy array of laws and rules
designed either to protect the woman worker, or to safeguard the
future of the race at her expense, are a pretty lame result of a great
deal of humanitarian sound and fury. Parturiunt montes.
It is quite natural that the result should be lame; for these protections
and safeguards represent so many attempts to mind some one
else’s business; and the great difficulty about minding some one
else’s business is that however good one’s intentions may be, one
can never really know just where that some one’s real interests lie,
or perfectly understand the circumstances under which he may be
most advantageously placed in the way to advance them, for the
circumstances are too intimately bound up with his peculiar
temperament and situation. As Mill has remarked in a passage which
I have already quoted, the world has learned by long experience that
affairs in which the individual is the person directly interested go right
only when they are left to his own discretion, and that any
interference by authority, save to protect the rights of others, is
mischievous. The tendency of modern welfare-legislation is to make
a complete sacrifice of individual rights not to the rights but to the
hypothetical interests of others; and for every individual who
happens to benefit by the sacrifice, there is another who suffers by it.
If it is hard to regulate one human being for his own good, it is
impossible to regulate people en masse for their own good; for there
is no way of making a general rule affect all individuals in the same
way, since no two individuals are to be found who are of precisely
the same temperament and in precisely the same situation.
There is in all this bungling effort to ameliorate the ills of working
women and to safeguard through them the future of the race, a tacit
recognition of economic injustice and a strange incuriousness about
its causes. One would naturally expect that the conditions which
move people to seek protective legislation would move them to
question the nature of an economic system which permits such
rapacity that any class of employees requires to be protected from it.
Surely the forces of righteousness must know that there are reasons
for the existence of the conditions which move them to pity and
alarm; yet they seem quite willing to go on indefinitely battling
against the conditions, and winning with great effort legislative
victories which are constantly being rendered ineffectual through lax
administration of laws, through the reluctance of employees to
jeopardize their positions by testifying against employers, or through
unforeseen changes in economic conditions. During all this waste of
time and effort, this building and crumbling and rebuilding of
protective walls around the labourer, the causes of economic
injustice continue their incessant operation, producing continuously a
new crop of effects which are like so many windmills inviting attack
by the Don Quixotes of reform.
Let us consider the effects of economic injustice on women, side by
side with the reformer’s work upon those effects. Women in industry
suffer, as I have shown, the injustice of inequality with men as
regards wages, opportunities, training, and tenure of employment.
The reformer attacks the problem of wages, and secures minimum-
wage laws based on some one’s theory of what constitutes a living
wage. No allowance is made for dependents because women,
theoretically, have none. The amount allowed may from the first be
inadequate, even for one person, or it may be rendered inadequate
by a rise in the cost of living. In either case, it is purely arbitrary, and
bears no relation whatever to the value of the worker’s services. Still,
such legislation might be better than nothing if there were nothing
better to be done. The reformer is less zealous in his attempt to
provide women with opportunities; his showing in this field is less
impressive than in that of wages. Still, he has done something. If he
has not been entirely responsible for the opening to women of many
positions in government service, he has at least greatly assisted in
securing them these opportunities. Farther than this, it must be
admitted, it is difficult for him to go. He might, indeed, exert himself
to see that women are provided by one means or another with equal
opportunities to get training, but he can do little to affect the policies
of private employers of labour, who can hardly be dictated to
concerning whom they shall hire and whom they shall retain. Nor can
he prevent employers from laying off women workers first when
there is a slowing down in production. In three, then, out of four of
the disadvantages which bear more heavily on women in industry
than on men, the reformer, with all his excellent intentions, is unable
to be very helpful; while in his zeal to safeguard the race, whose
future appears to him to depend entirely on the health of the female
sex, he has multiplied their disadvantages in the manner I have
already described, without, however, having made any noteworthy
advance toward the accomplishment of his purpose.
Now, had he chosen to inquire into the causes of the artificial
disabilities by which women workers are handicapped, he might
have discovered that these and the industrial hazards which cause
him such grave concern may be traced to the same fundamental
source; and that the just and only effective way of removing these
disabilities and hazards is to eradicate the source. Women in
industry are the victims of traditional prejudices: I have shown what
those prejudices are—the idea that woman’s place is the home, that
women workers have no dependents, that they work for pin-money
and therefore do not need a living wage, that upon them alone
depends the future health of the race. But as I remarked at the
beginning of this chapter, these prejudices could not be turned to the
disadvantage of the woman worker if it were not for the overcrowding
of the labour-market. So long as there are more people looking for
work than there are jobs to be had, the advantage in fixing terms and
conditions of labour is on the side of the employer. If men are obliged
by their need to put up with underpayment, women will be forced to
accept an even worse rate; if the tenure of men is uncertain, that of
women will be even more so. If the conditions of industry are
hazardous, the alternative of starvation will force the workers to risk
injury or death unless the employer be required by law to maintain
the proper safeguards. Suppose, however, that labour were scarce,
that for every worker looking for employment there were a dozen
employers looking for workers. Under such circumstances, the
employer would be glad enough to hire the worker who could fill his
particular requirements, without regard to sex, as employers did
during the war when labour was scarce; and he would pay the
worker a wage determined not by theory or prejudice, but by the
amount of competition for the worker’s services. If the employment
he offered were hazardous, he would be obliged to maintain proper
safeguards in order to retain his employees, and in addition would
probably be forced to pay them a higher wage than they could earn
in some safer employment. If he did not do these things, his workers
would simply leave him for more satisfactory positions. Nor would he
be able to overwork his employees, for if he attempted to do so,
some rival employer would outbid him for their services by offering
better hours and easier conditions of labour. Thus the peculiar
disabilities of women workers would disappear with the disabilities of
labourers in general, and not a stroke of legislation would be
required to make industry both safe and profitable for the woman
worker.
This condition is not unnatural or impossible. It is the present
condition of chronic unemployment, of expensive and ineffectual
“welfare” legislation, of wasteful and futile struggles between
organized capital and organized labour—it is this condition that is
entirely unnatural. I have mentioned its cause in Chapter III, and I
shall discuss it further in my next chapter. Upon its removal, and not
upon regulations which hamper the woman worker and reduce her to
the status of a function, the future of the race depends. The
ancestors of coming generations are men as well as women, and
posterity will derive its heritage of health from its ancestors of both
sexes. Its prospect of health will not be improved by legislation
calculated to safeguard the health of women workers, so long as the
children they bear continue to be exposed to an involuntary poverty
which breeds ignorance, imbecility, disease and crime. The
happiness as well as the health of future generations will depend in
great measure upon the extent to which both men and women can
release themselves from the deteriorating conditions of economic
exploitation.
II
It is in business and in professional pursuits that the occupational
progress of women, and their emancipation from traditional
prejudices, are most marked. Although in the lower ranks of labour in
these pursuits there is a mass of women who, impelled by necessity,
work for low wages at mechanical tasks which offer no chance of
advancement, there is, nearer the top, a large group of women who
have been more fortunate in worldly position and education, and who
are spurred as much either by interest in their work or a desire to be
self-supporting, as by actual need to earn; who share, in other
words, the attitude that leads young men to strike out for themselves
even though their fathers may be able to support them. It is the
woman animated by these motives who is doing most for the
advancement of her sex; for it is she, and not the woman who works
through necessity, who really challenges the traditional prejudices
concerning the proper place of women. The woman labourer proves
the need of women to earn; the business woman or professional
woman who works because she wants to work, is establishing the
right of women to earn. More than this, as she makes her way into
one after another of the occupations that have been held to belong
to men by prescriptive right, she is establishing her claim, as a
human being, to choose her work from the whole wide field of human
activity. It is owing to the attitude towards life adopted by such
women, to their preference of independence and action over the
dependence and passivity in vogue not so many years ago, that it is
coming to be quite the expected thing that young women of the well-
to-do classes shall set out to earn their living, as young men do,
instead of stopping under the parental roof, with a watchful eye out
for men who will marry and support them. Need I remark that nothing
is more likely than this new attitude to bring about the substitution of
the “union by affection” for the union by interest? The woman who is
economically independent is under much less temptation to marry
from economic motives than the woman for whom marriage
represents the only prospect of security.
There is still a goodly number of prejudices and discriminations to be
overcome before women in business and the professions shall stand
on an equal footing with men as regards opportunity and
remuneration. Except where she is in business for herself, the
woman in these pursuits must generally be content with a lower rate
of pay than men; and if observation may be taken to count for
anything, she is expected to work somewhat harder for what she
gets—less loafing on the job is tolerated in her than in the male
employee. She is also more likely to find herself pocketed; that is to
say, in a position from which, because of her sex, there is no
possibility of further advance because the higher positions are
reserved for men. It is so universally the rule that women must
content themselves with reaching the lower rungs of the
occupational ladder, that the instances where they manage to attain
to places of responsibility and authority are still rare enough to be
found worthy of remark in the press. The same thing is true of
political positions; women are not yet represented in politics in
anything like a just proportion to their numbers, nor are they often
able to get themselves either elected or appointed to responsible
positions. None the less, considering the comparatively short time
since their emergence into the business world and the world of
public affairs, they are already making an excellent showing.
The world of business and the professions, like the world of industry,
has its occupations which are considered peculiarly suitable for
women. Strictly subordinate positions are thought to suit them very
well; hence there is quite an army of women stenographers,
bookkeepers, clerks and secretaries to be found in the business
section of any modern city. The personnel of the nursing profession
is made up almost exclusively of women; and the work of teaching in
our public schools, especially where it is most conspicuously
underpaid, is largely in their hands. There is, to be sure, an
impression current among members of school boards that marriage
disqualifies a woman for the teaching profession; but the single
woman is fairly secure in her position, possibly because it does not
pay well enough to be very attractive to men. Occupations
connected with the arts are also held, in this country, to be
particularly well adapted for women, although it must be noted that
the prejudice of male musicians is effective enough to exclude them
from the personnel of our important orchestras. It is in the creative
arts that their work is most welcomed; more especially in the field of
literature; and this may seem strange, in view of the fact that so
many eminent authorities believe that their sex renders them
incapable of attaining any significance in creative work. It is, I
apprehend, rather to the low opinion in which aesthetic pursuits are
held in this country than to a high opinion of female ability, that this
peculiar condition must be ascribed.
But if certain occupations are considered peculiarly appropriate for
women, there is none the less a great deal of prejudice against them
in others. The idea that woman’s place is the home has no more
disappeared from the world of business and the professions than it
has disappeared from the world of industry, even though it is the
business woman and the professional woman who are doing most to
dislodge it. And here it may be well to remark a fact that has already
been noted, with some pointed comment, by Ethel Snowden,
namely: that woman’s invasion of the gainful occupations appears to
be found unwomanly in proportion to the importance of the position
to which she aspires.
It is the married woman in business or in professional work, as it is in
industry, who suffers most from the surviving prejudices concerning
her sex. When there are economies to be effected through the
discharge of workers, the idea that the married woman is normally a
dependent comes immediately to the fore, and she is the first
employee to be discharged. For example, Equal Rights of 8 August,
1925, noted in an editorial that the city of St. Louis had begun a
campaign for economy by discharging twelve married women; that
there was a movement on in Germany to reduce governmental
expenses by a wholesale discharge of women employees; and that,
according to rumour, Mr. Coolidge’s campaign of economy was
being made to bear most heavily on married women. The comment
of Equal Rights on the action of the city of St. Louis is worth quoting:
St. Louis employed twenty-seven married women. It
investigated the economic condition of all these, retained nine,
discharged twelve, and was, at last report, still considering the
case of the other six. St. Louis did not investigate the economic
condition of the men employees, to see whether or not these
might continue to live if they were discharged. St. Louis did not
try to find out whether or not these men had fathers, brothers,
mothers, or wives who might support them while they were
looking for other jobs. St. Louis assumed that men have a right
to economic independence and the increased happiness and
opportunity that it brings. St. Louis assumed that women have
no such right.
In other words, St. Louis assumed, as the German and American

Governments apparently assume, and as most private employers
assume, that women are employed on sufferance; especially married
women. Of course it should be remembered that the position of the
married woman in this respect is only worse than that of single
women, and that the position of women is only worse than that of
men; for, as I have already remarked, under a monopolistic
economic system the opportunity to earn a living by one’s labour
comes to be regarded as a privilege instead of a natural right.
Women are simply held to be less entitled to this privilege than men.
That marriage should so often assume the nature of a disability for
the woman who either wishes or is obliged to earn, whereas it often
operates in favour of the male worker, may be attributed to the
traditional assumption that married women are dependent on, and
subject to, their husbands. I remarked in the preceding chapter that
the married woman who wishes to engage in business finds herself,
in many communities, hampered by legal disabilities arising from her
marital status, whereas her husband is under no corresponding
disabilities. Her position as an industrial and salaried worker is
rendered insecure if not by law, at least by the same psychology that
keeps legal disabilities in force. This psychology may be defined as
the expectation that a woman when she marries shall surrender a
much greater degree of personal freedom than the man she marries.
The man who does not object to his wife’s having a career is
considered generous and long-suffering. His insistence on her
abandoning it and contenting herself with looking out for his
domestic comfort is thought to be quite natural.[28] On the other
hand, the woman who interferes in any way with a husband’s career
is regarded as an extremely selfish person; while any sacrifice of
herself and her ambitions to her husband and his, is thought of
merely as a matter of wifely duty. How often does one hear that such
and such a woman has given up her position because “her husband
didn’t want her to work.” There is, too, a very general assumption
that every married woman has children and should stay at home and
take care of them. Now, perhaps every married woman should have
children; perhaps in a future state of society men and women will
marry only when they wish to bring up a family. But at present it is
not so; therefore at present the assumption that a married woman
should stay at home and take care of her children leaves out of
account the fact that a large and increasing number of married
women are childless. It may be contended that these women should
stay at home and take care of their husbands; but even if we assume
that the unremitting personal attention of his wife is essential to the
comfort and happiness of a married man, there would still remain the
question of his title to this attention at the cost of her own interests.
We are dealing here with an attitude which, general though it be, has
been outmoded by the conditions of modern life. The sexual division
of interests and labour which has been insisted upon so long among
European peoples does not very well fit in with the organization of
industrial and social life in the twentieth century. Our social ideology,
like our political ideology, is of the eighteenth century; and its
especial effectiveness at present is by way of obscuring our vision of
the changed world that has emerged from the great economic
revolution of the last century. A division of interests and labour which
was convenient if not just under the conditions of economic and
social life which preceded the industrial revolution, is neither
convenient nor just under the conditions which prevail today. The
care of young children and the management of a household may
result in an unequal division of labour in families where the
husband’s inability to provide for the needs of his family forces the
wife to assume the burdens of a breadwinner. When one reads
through the literature on the question of hours of labour for women in
industry, one is struck by the persistent stressing of the married
woman’s double burden of breadwinning and housekeeping. These
women, it seems, must not only earn money to contribute to their
families’ support, but they must, before setting out for work and after
returning from it, prepare the family meals, get the children ready for
school or the day-nursery, take them there and call for them, wash,
sew, and perform a hundred other household tasks. This double
burden is often made an argument for establishing shorter hours of
work for women in industry, but never for expecting the husband to
share the wife’s traditional burden as she has been forced to share
his. I have no doubt that innumerable husbands are doing this; but
there is no expectation put upon them to do it, and those who do not
are in no wise thought to shirk their duty to their families, as their
wives would be thought to do if they neglected to perform the labour
of the household.
Quite analogous to this attitude of the advocates of special
legislation for working women is that of the people who concern
themselves with the so-called problem of the educated woman,
which is supposed to be that of reconciling domesticity with
intellectual pursuits. A timely illustration of this attitude is the
establishment by Smith College of an institute for the “co-ordination
of women’s interests.” The purpose of this institute, in the words of
President Neilson, is “to find a solution of the problem which
confronts almost every educated woman today—how to reconcile a
normal life of marriage and motherhood with a life of intellectual
activity, professional or otherwise.” Here again is the tacit
assumption that marriage is the special concern of woman, and one
whose claims must take precedence over her other interests,
whatever they may be; that marriage and motherhood constitute her
normal life, and her other interests something extra-normal which
must somehow be made to fit in if possible. I have heard of no
institute intended to find a way to reconcile the normal life of
marriage and fatherhood with a life of intellectual activity,
professional or otherwise; although when one considers how many
educated men of today are obliged to compromise with their
consciences in order to secure themselves in positions which will
enable them to provide for their families, one is persuaded that some
such institute might be at least equally appropriate and equally
helpful with that which Smith College has established.
Let us forget for a moment the sophisticated traditional attitude
toward this question of marriage and parenthood, and go back, as it
were, to the beginning—to a fact recognized in the animal world and
not entirely overlooked by primitive man, namely: that every offspring
has two parents who are equally responsible for its care and
protection. In the animal kingdom one finds a widely varied division
of the labour connected with the care of the young. For example, the
male of certain species is found to perform functions which our own
usage has led us to regard as maternal. Among the viviparous
animals the heavier share of responsibility rests with the female
during the gestation, birth and extreme youth of the offspring; and
among primitive human beings the actual physical dependence of
the offspring on the mother is likely to be prolonged over a period of
several years. It was, perhaps, this necessity of a close physical
association between mother and child that led to a sexual division of
labour under which the mother undertook the physical care of
children while the father undertook the task of providing food. It must
be remarked, however, that this division of labour by no means
excludes productive labour on the part of the woman. Among most
tribes she augments the food-supply through agriculture, grubbing,
or sometimes through fishing or hunting; and there are tribes,
notably in Africa, where she is the sole provider for the family. The
Vaertings have remarked that the drudgery connected with the care
of children is invariably imposed by the dominant upon the subject
sex; a view which is in perfect consonance with what we know of the
general human willingness to transfer to other shoulders the burden
of uninteresting though necessary labour. Since women have most
often been subject, they have most often been forced to undertake
this drudgery, either in lieu of or in addition to the labour of providing
food and shelter for their families.
This is to say that their subject position has added considerably to
what newspaper editors and other commentators are fond of calling
the burden of Eve. Since woman is the childbearing sex, it has
seemed natural to a great many peoples to increase the
disadvantage at which her share in reproduction naturally places her,
by making her confinement at home permanent instead of
occasional, and by permitting her few, if any, interests save those
connected with reproduction; in short, by prolonging and enhancing
her subjection to the demands of the race. This is why the term
married woman is still taken to imply the term housekeeper; an
implication which, as the Freeman remarked editorially some years
ago, modern civilization must renounce “if it wants such of its women
as are editors and bank-presidents to be mothers as well.”
Civilization shortens the period of the child’s physical dependence on
the mother by shortening the period of lactation. On the other hand,
it increases fecundity to such an extent that where religious
superstition or ignorance prevents the use of contraceptives, the
burden of childbearing is greatly increased. This result of civilization
is not, however, commonly found among the educated classes; and
even among those classes where children are most numerous, I
have already shown that women are not restrained by motherhood
from engaging in gainful occupations outside the home. On the
contrary, the number of their offspring is more often their chief
incentive to this course. Among well-to-do families, prepared foods
and wet-nursing have for a long time been rather generally employed
to relieve mothers even of the responsibility of lactation, while the
custom of assigning the physical care of children to hired substitutes
has reduced their actual work to that of bringing the child into the
world. That this mode of caring for children is approved by all
classes is evident from their readiness to adopt it when fortune
favours them with an opportunity. It is occasionally inveighed against
by moralists, but on the whole it is coveted and approved, especially
while women devote to frivolous pursuits the leisure that it leaves
them. When a woman adopts this mode in order to reconcile
motherhood with a serious interest outside the home, it is a different
matter, and lays her open to the charge of neglecting her family,
though in fact she may spend no more hours away from home than
the woman who gives her morning to shopping and her afternoon to
playing bridge. Why this should be the case I am at a loss to know,
unless it be that a serious interest outside the home appears to
smack too much of an assertion of her right to live her life for her
own sake rather than for the sake of the race or that of her husband
—a self-assertion not readily to be accepted without such
reservations as find expression in institutes designed to “co-ordinate
women’s interests.”
It appears, then, that the care of the young is the concern of both
sexes, and is so recognized in the animal world and among human
beings; and that among the latter such differences in usage as exist
touching this matter are differences in the apportioning of the
burden. Even in our own day, when there is observable a tendency
to forget that the child has more than one parent—that parent being
the mother—the father’s claim to his children is still recognized in
law, often to the prejudice of the mother’s; and so, likewise, is his
obligation to provide for them. Indeed, the child may be said to be
regarded as exclusively the mother’s only while it is young; for it is a
general custom among us to speak of Mrs. So-and-So’s baby, but of
Mr. So-and-So’s son or daughter. Let us, then, recognize the claim
and interest of both parents. Let us also remember that the
economic organization has so extensively altered that the traditional
division of labour—this division is always profoundly affected by
consideration of the young—has been outmoded as far as
thousands of families are concerned. Let us also assume that
woman has established her right to be considered as a human being
rather than a function or a chattel. Then it must seem reasonable to
assume that the co-ordination of interests to be brought about
concerns both sexes equally; that the problem to be confronted is
that of reconciling a normal life of marriage and parenthood not only
with the freest possible development of intellectual interest but with
the utmost devotion to any chosen profession.
I can not pretend to foretell how this problem will be settled; for its
solution will depend upon the general solution of the labour-problem.
It may be that the necessary collectivism of modern industry will
result in a collectivist system of caring for children. Such a system
would by no means be an innovation; it would simply constitute an
extension and adaptation of means which already exist—of nurseries
for very small children and schools for older ones. Whatever its
demerits might be, such a system would certainly represent an
enormous economy of effort. The average home is adapted less to
the needs of children than to those of adults; hence a mother of
young children must spend a great deal of her time in preventing her
young charges from injuring themselves with dangerous household
implements, from falling downstairs or off of furniture too high for
them, and from touching objects which would not be safe in their
hands. In a properly equipped nursery, on the other hand, the
furniture and all the objects are adapted to the size and intelligence
of the children. Children have the advantage of numerous
playmates; and one person can supervise the play of a dozen of
them with less fatigue than the mother of one is likely to feel at the
end of a day in the average home.
The Russians have already taken some steps in this direction by
establishing both nurseries and schools in connexion with certain
factories. From what I can gather of their policy, it would seem that
they regard the care and education of children as being very much
the concern of the whole community. They look upon childbearing as
a service to the community, but they do not appear to take the view
that women should be required to perform this service at the
expense of their independence, for they have instituted a system of
subsidies for pregnant and nursing working mothers, with rest-
periods before and after confinement, and a subsidy during
confinement amounting to the daily subsidy multiplied by fifteen.[29]
I have already indicated in the preceding chapter what it seems to
me would be the course of a free people in this matter of reconciling
the care of children with the greatest possible freedom for both
parents. It seems hardly necessary to call attention to the obvious
fact that the question is simply that of placing the care of the young
in the hands of those who are interested in it and fitted for it, instead
of forcing it willy-nilly upon either sex through a traditional
expectation and a traditional division of labour. In a free society,
those parents who wished to pursue careers incompatible with the
actual care of young children would avail themselves of the services
of substitutes, as the well-to-do classes do at present; and they
might do so with even greater confidence because, as I have
remarked, those engaged in caring for and teaching the young would
do so as a matter of interest primarily and only secondarily as a
means of livelihood. There is another important consideration to be
taken into account, and that is, that in a free society the problem of
reconciling the occupations of the parents with their personal
supervision of their children would be much easier to solve; for their
hours of labour would be greatly decreased. It is only where
production must support an enormous amount of idleness and waste
that it is necessary to overwork producers.
It is possible, of course, that the institution of economic freedom
might check the present tendency of women to engage in gainful
occupations outside the home. It most certainly would if the vast
increase of opportunity which it offered were reserved exclusively for
men; but to bring about this result it would be necessary for
traditional anti-feminist prejudices to survive much more strongly
than they do today. The position of women has too radically changed
to admit of their exclusion from direct participation in the benefits of
economic freedom; therefore if they resigned the increased
economic opportunities that it offered them, and withdrew to the
sphere of domesticity, they would do so as a matter of choice. Why
should we not expect them to choose the exclusive domesticity
which might be rendered possible through the increased earning
power of men? They probably would, where it suited their taste to do
so; but one of the most powerful incentives to do so would no longer
exist, namely: the desire for economic security. Women, to be sure,
are not exempt from the characteristic willingness of humankind to
live by the exertions of others; but I would remark that there is this
difference between the person who does this indirectly, through
legalized privilege, and the person who depends directly on the
bounty of another: that the former is independent and the latter is
dependent. Women are not strangers to the human desire for
freedom; and when the fear of want is allayed they are quite likely to
prefer an easy and secure self-support to the alternative of economic
dependence. Moreover, economic freedom would set domesticity in
competition with the interests of women rather than their needs; for it
would set all people free to engage in occupations that interested
them, whereas at present the vast majority do whatever offers them
a living. Under these circumstances it might reasonably be expected
that the number of women who would continue in business and in
industrial and professional pursuits, even after marriage and the birth
of children, would greatly increase.
Indeed, if we postulate an economic system under which every
human being would be free to choose his occupation in accordance
with his interests, I see no more reason to suppose that women
would invariably choose domesticity than to suppose that all men
would choose blacksmithing. Under such a régime I doubt that even
the power of the expected which affects them so strongly at present,
would long continue in an effectiveness which it has already begun
to lose. Women, I think, might be expected to choose their
occupations with the same freedom as men, and to look for no
serious interruption from marriage and the birth of children. There
are a good many women at present who very ably reconcile
motherhood with a chosen career. I think we might expect to find
more of them rather than fewer, in a free society. One thing is
certain, and it is the important thing: they would be free to choose. If
it be woman’s nature, as some people still believe, to wish to live at
second hand, then in a free society they will freely make that choice,
and no one can complain of it—unless it be the men on whom they
elect to depend. However, to assume from past experience that they
do want to live at second hand is to assume that all the social and
legal injustices which have been employed to force them to do so,
were unnecessary; and when have Governments and communities
wasted their power in exercising compulsion where no compulsion
was needed?
FOOTNOTES:
[26] Ellis: Man and Woman. 5th ed. p. 14.
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!
textbookfull.com

Multivariate Analysis for the Behavioral Sciences Kimmo Vehkalahti 2024 Scribd Download

Uploaded by

Copyright:

Available Formats

Multivariate Analysis for the Behavioral Sciences Kimmo Vehkalahti 2024 Scribd Download

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Multivariate Analysis for the Behavioral Sciences Kimmo Vehkalahti 2024 Scribd Download

Uploaded by

Copyright:

Available Formats

Download the Full Version of textbook for Fast Typing at textbookfull.

Multivariate Analysis for the Behavioral Sciences

Download More textbook Instantly Today - Get Yours Now at textbookfull.com

Multivariate analysis for the behavioral sciences 2nd

Applied Power Analysis for the Behavioral Sciences 2nd

Statistics for the behavioral sciences Gravetter

Statistics for the Behavioral Sciences Frederick J.

Statistical Concepts for the Behavioral Sciences Harold O.

Essentials of Statistics for the Behavioral Sciences

Statistics for the Behavioral Sciences 4th Edition Susan

Introduction to Quantitative Data Analysis in the

Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)

Wim J. van der Linden Tom Snijders

Aims and scope

Recently Published Titles

Handbook of International Large-Scale Assessment: Background, Technical Issues,

For more information about this series, please visit: https://www.crcpress.com/go/ssbs

Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)

c 2019 by Taylor & Francis Group, LLC

No claim to original U.S. Government works

Printed on acid-free paper

International Standard Book Number-13: 978-0-8153-8515-8 (Hardback)

Trademark Notice: Product or corporate names may be trademarks or registered trademarks,

Library of Congress Cataloging-in-Publication Data

Names: Everitt, Brian, author. | Vehkalahti, Kimmo, author.

Visit the Taylor & Francis Web site at

and the CRC Press Web site at

Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)

Brian dedicates the book to the memory of his parents,

Kimmo dedicates the book to Sirpa, the love of his life.

Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)

1 Data, Measurement, and Models 1

2.5 Conditioning Plots and Trellis Graphics . . . . . . . . . . . 48

3 Simple Linear and Locally Weighted Regression 63

4 Multiple Linear Regression 83

5 Generalized Linear Models 113

6 Applying Logistic Regression 123

Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)

6.5 Driving and Back Pain: A Matched Case–Control Study . . 134

7 Survival Analysis 139

8 Analysis of Longitudinal Data I: Graphical Displays and

9 Analysis of Longitudinal Data II: Linear Mixed Effects

10 Analysis of Longitudinal Data III: Non-Normal Responses 189

Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)

10.3 Using Generalized Estimating Equations

11 Missing Values 209

12 Multivariate Data and Multivariate Analysis 225

13 Principal Components Analysis 239

Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)

13.6 Rescaling the Principal Components . . . . . . . . . . . . . 248

14 Multidimensional Scaling and Correspondence Analysis 267

15 Exploratory Factor Analysis 295

Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)

16 Confirmatory Factor Analysis and Structural Equation

17 Cluster Analysis 341

18 Grouped Multivariate Data 365

Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)

In some respects this book is a second edition of Multivariable Modeling and

• As the main part of a formal statistics course for advanced undergraduates

Libros de Estadística-Ciencia de Datos|Statistics-Data Science Books (PDF)