Full Download The Philosophy of Quantitative Methods: Understanding Statistics Brian D. Haig PDF

Full download test bank at ebook ebookmass.
com
The Philosophy of Quantitative

Methods: Understanding Statistics
Brian D. Haig
CLICK LINK TO DOWLOAD
https://ebookmass.com/product/the-philosophy-
of-quantitative-methods-understanding-
statistics-brian-d-haig/
ebookmass.com
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
The Code of Putinism Brian D. Taylor
https://ebookmass.com/product/the-code-of-putinism-brian-d-
taylor/
Quantitative Methods for Business 13th Edition
https://ebookmass.com/product/quantitative-methods-for-
business-13th-edition/
An Introduction to the Philosophy of Religion 4th

Edition Brian Davies
https://ebookmass.com/product/an-introduction-to-the-philosophy-
of-religion-4th-edition-brian-davies/
Quantitative Methods for Business, 5th Edition Donald

Waters
https://ebookmass.com/product/quantitative-methods-for-
business-5th-edition-donald-waters/
Understandable Statistics: Concepts and Methods 12th
Edition
https://ebookmass.com/product/understandable-statistics-concepts-
and-methods-12th-edition/
All Talked Out: Naturalism and the Future of Philosophy

J. D. Trout
https://ebookmass.com/product/all-talked-out-naturalism-and-the-
future-of-philosophy-j-d-trout/
Quantitative Techniques in Management 5th Edition Dr.

N. D. Vohra
https://ebookmass.com/product/quantitative-techniques-in-
management-5th-edition-dr-n-d-vohra/
Understanding Basic Statistics 8th Edition Charles

Henry Brase
https://ebookmass.com/product/understanding-basic-statistics-8th-
edition-charles-henry-brase/
(eBook PDF) Statistics for Political Analysis:

Understanding the Numbers Revised Edition
https://ebookmass.com/product/ebook-pdf-statistics-for-political-
analysis-understanding-the-numbers-revised-edition/
i
The Philosophy
of Quantitative
Methods
ii
S E R I E S I N U N D E R S TA N D I N G S TAT I S T I C S
S. NATASHA BERET VAS Series Editor
S E R I E S I N U N D E R S TA N D I N G M E A S U R E M E N T
S. NATASHA BERET VAS Series Editor
S E R I E S I N U N D E R S TA N D I N G Q U A L I TAT I V E R E S E A R C H
PATRICIA LEAVY Series Editor
Understanding Statistics Video as Method

Anne M. Harris
Exploratory Factor Analysis
Leandre R. Fabrigar and Focus Group Discussions
Duane T. Wegener Monique M. Hennink
The Philosophy of Quantitative The Internet
Methods Christine Hine
Brian D. Haig
Diary Methods
Validity and Validation Lauri L. Hyers
Catherine S. Taylor
Oral History
Patricia Leavy
Understanding Measurement
Using Think-Aloud Interviews and
Item Response Theory Cognitive Labs in Educational Research
Christine DeMars Jacqueline P. Leighton
Reliability Qualitative Disaster Research
Patrick Meyer Brenda D. Phillips
Understanding Qualitative Fundamentals of Qualitative Research

Research Johnny Saldaña
Autoethnography Duoethnography
Tony E. Adams, Stacy Holman Jones, Richard D. Sawyer and Joe Norris
and Carolyn Ellis Analysis of the Cognitive Interview in
Qualitative Interviewing Questionnaire Design
Svend Brinkmann Gordon B. Willis
Evaluating Qualitative
Research: Concepts, Practices,
and Ongoing Debates
Jeasik Cho
iii
Brian D. Haig
THE PHILOSOPHY
OF QUANTITATIVE
METHODS
1
iv
3
Oxford University Press is a department of the University of Oxford. It furthers
the University’s objective of excellence in research, scholarship, and education
by publishing worldwide. Oxford is a registered trade mark of Oxford University
Press in the UK and certain other countries.
Published in the United States of America by Oxford University Press

198 Madison Avenue, New York, NY 10016, United States of America.
© Oxford University Press 2018
All rights reserved. No part of this publication may be reproduced, stored in

a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by license, or under terms agreed with the appropriate reproduction
rights organization. Inquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above.
You must not circulate this work in any other form

and you must impose this same condition on any acquirer.
CIP data is on file at the Library of Congress

ISBN 978–0–19–022205–5
9 8 7 6 5 4 3 2 1
Printed by WebCom, Inc., Canada
v
Contents
Acknowledgments . . . . . . . . . . . . . . . . . . ix
CHAPTER 1 Introduction . . . . . . . . . . . . . . . . . . . . . . 1
The Philosophy of Quantitative Methods . . . . . . . 2

Scientific Realism and Its Methodology . . . . . . . . 3
Theories of Scientific Method . . . . . . . . . . . 6
Book Overview and Chapter Summary . . . . . . . . 7
A Note for the Reader . . . . . . . . . . . . . . 9
References . . . . . . . . . . . . . . . . . 11
CHAPTER 2 Exploratory Data Analysis . . . . . . . . . . . . . . . . . . 13
Introduction . . . . . . . . . . . . . . . . . 13
What Is Exploratory Data Analysis? . . . . . . . . 15
Two Methods of Exploratory Data Analysis . . . . . . 17
The Four Rs of Exploratory Data Analysis . . . . . . 18
Exploratory Data Analysis and Scientific Method . . . . 20
Exploratory Data Analysis After Tukey . . . . . . . 29
Resampling Methods . . . . . . . . . . . . . . 30
A Philosophy for Teaching Data Analysis . . . . . . 35
Conclusion . . . . . . . . . . . . . . . . . 37
vi
vi : Contents
Further Reading . . . . . . . . . . . . . . . 37
References . . . . . . . . . . . . . . . . . 38
CHAPTER 3 Tests of Statistical Significance . . . . . . . . . . . . . . . . 41
Introduction . . . . . . . . . . . . . . . . . 41
Null Hypothesis Significance Testing: Psychology’s
Textbook Hybrid . . . . . . . . . . . . . . 43
The Neo-Fisherian Perspective . . . . . . . . . . 46
The Error-Statistical Perspective . . . . . . . . . . 49
What Should We Think About Tests of Significance? . . . 55
Conclusion . . . . . . . . . . . . . . . . . 60
Further Reading . . . . . . . . . . . . . . . 60
References . . . . . . . . . . . . . . . . . 61
CHAPTER 4 Bayesianism . . . . . . . . . . . . . . . . . . . . . . 65
Introduction . . . . . . . . . . . . . . . . . 65
Bayesianism in Psychology . . . . . . . . . . . 67
Bayesian Confirmation Theory . . . . . . . . . . 68
Bayesianism and the Hypothetico-Deductive Method . . 72
Bayesianism and Inference to the Best Explanation . . . 73
Two Common Criticisms of Bayesianism . . . . . . 75
What Should We Think About Bayesian
Confirmation Theory? . . . . . . . . . . . . 78
A Neo-Popperian Philosophy of Bayesian Statistics . . . 80
Conclusion . . . . . . . . . . . . . . . . . 86
Further Reading . . . . . . . . . . . . . . . 87
References . . . . . . . . . . . . . . . . . 88
CHAPTER 5 Meta-Analysis . . . . . . . . . . . . . . . . . . . . . 91
Introduction . . . . . . . . . . . . . . . . . 91
Glass’s Rationale for Meta-Analysis . . . . . . . . . 93
Meta-Analysis and Scientific Discovery . . . . . . . 101
Meta-Analysis and Phenomena Detection . . . . . . 105
Meta-Analysis and Scientific Explanation . . . . . . 106
Conclusion . . . . . . . . . . . . . . . . . 110
Further Reading . . . . . . . . . . . . . . . 111
References . . . . . . . . . . . . . . . . . 112
vi
Contents : vii
CHAPTER 6 Exploratory Factor Analysis . . . . . . . . . . . . . . . . . 117
Introduction . . . . . . . . . . . . . . . . . . . . 117
Exploratory Factor Analysis and Scientific Inference . . 119
The Principle of the Common Cause . . . . . . . . 122
Methodological Challenges to Exploratory
Factor Analysis . . . . . . . . . . . . . . . 128
Exploratory Factor Analysis and Other
Factor Analytic Methods . . . . . . . . . . . 135
Conclusion . . . . . . . . . . . . . . . . . 139
Further Reading . . . . . . . . . . . . . . . 140
References . . . . . . . . . . . . . . . . . 140
CHAPTER 7 Conclusion . . . . . . . . . . . . . . . . . . . . . . 143
Chief Lessons Learned . . . . . . . . . . . . . 143

A Final Word . . . . . . . . . . . . . . . . 148
Index . . . . . . . . . . . . . . . . . . . . . . 149
vi
ix
Acknowledgments
In preparing this book, I have made use of previously published material

from the following sources:
Haig, B. D. (2005). Exploratory factor analysis, theory generation, and
scientific method. Multivariate Behavioral Research, 40, 303–329.
Haig, B. D. (2012). The philosophy of quantitative methods. In T. D.
Little (Ed.), Oxford handbook of quantitative methods (Vol. 1, pp. 6–30).
New York, NY: Oxford University Press.
Haig, B. D. (2016). Tests of statistical significance made sound. Educational
and Psychological Measurement, 77, 489–506.
I am grateful to the journals and publishers for allowing me to make use
of this material.
x
xi
The Philosophy
of Quantitative
Methods
xi
1
1
Introduction
the ph i l osophy of research methods is an area of knowl-

edge that receives limited attention in behavioral research method-
ology and science education. The majority of students and research
practitioners in the behavioral sciences obtain the bulk of their
knowledge of research methods from textbooks. However, a casual
examination of these texts shows that they tend to pay little, if any,
serious regard to the philosophy of science and its bearing on the
research process. As Thomas Kuhn pointed out more than 50 years
ago (Kuhn, 1962/1996), textbooks play a major role in dogmati-
cally initiating students into the routine practices of normal science.
Serious attention to the philosophy of research methods would go a
considerable way toward overcoming this uncritical practice.
This book is concerned with the philosophical foundations
of research methods. In particular, it undertakes a philosophical
examination of a number of different quantitative research meth-
ods that are prominent in, or relevant for, the conduct of research
in the behavioral sciences. The methods submitted to critical
examination are exploratory data analysis, statistical significance
testing, Bayesian confirmation theory and statistics, meta-analysis,
and exploratory factor analysis. I introduce these methods, and
explain their selection, in the overview section that follows.
2
2 : The Philosophy of Quantitative Methods
The Philosophy of Quantitative Methods

Historically, philosophers of science have given research methods
in science limited attention, concentrating mostly on the nature
and purpose of theory in the physical sciences. More recently,
however, they have shown an increased willingness to deal with
methodological issues in sciences other than physics, particularly
biology, but also psychology and related behavioral and social sci-
ences to some extent. In short, there is a developing literature in
contemporary philosophy of science that can aid both our under-
standing, and use, of a variety of research methods and strategies
in psychology. Increasingly, the philosophy of science contributes
important methodological insights that are impossible to ignore
when coming to grips with research methods. Increasingly, the
philosophy of science is becoming a philosophy for science. At the
same time, a miscellany of theoretically oriented psychologists, and
behavioral and social scientists more generally, has produced work
on the conceptual foundations of research methods that helps illu-
minate those methods. The work of both professional philosophers
of science and theoretical scientists deserves to be included in a
philosophical examination of behavioral research methods.
The three major philosophies of science that bear on psychol-
ogy are empiricism, social constructionism, and scientific realism
(Greenwood, 1992; Manicas & Secord, 1983). Nineteenth-century
British empiricism had a major influence on the development of
British statistics in the first half of the twentieth century (Mulaik,
1985). The statistical methods developed in that intellectual milieu
remain an important part of psychology’s statistical research prac-
tice. For example, Karl Pearson’s product moment correlation coef-
ficient was taken by its founder to be the quantitative expression
of a causal relation viewed in empiricist terms. Similarly, Fisher’s
endorsement of inductive methods as the proper view of scien-
tific method stemmed from a commitment to the empiricism of
his day. Even in the current postpositivist philosophical climate,
authors of research methods textbooks sometimes portray quan-
titative research as essentially positivist in its empiricist commit-
ments (Yu, 2006). Among other things, positivism restricts its
attention to what can be observed and regards theories as instru-
ments that organize claims about observables but do not explain
them by appeal to hidden causes.
3
Introduction : 3
Qualitative methodologists also often bolster their preferred

conception of qualitative research by comparing it with an unflat-
tering positivist picture of quantitative research. They tend to
adopt the philosophy of social constructionism, which is opposed
to the traditional notions of truth, objectivity, and reason, main-
taining that our understanding of the world is determined by
social negotiation. In one or other of its various forms, it is the
philosophy of choice for many qualitative researchers, and it tends
to be employed by those who are opposed, or indifferent, to quan-
titative methods.
Scientific Realism and Its Methodology

The book adopts a scientific realist perspective on research meth-
ods, although its emphasis is not always evident. Scientific real-
ism, like all philosophies of science, is the subject of considerable
debate, and it is opposed by many antirealist positions (princi-
pally, the philosophies of empiricism and social constructivism).
Nonetheless, with justification, it remains the dominant philoso-
phy of science to this day (Psillos, 1999). It is also the tacit philoso-
phy of most working scientists. This fact, combined with its current
emphasis on the nature of scientific practice, makes scientific real-
ism the philosophy of choice for science.
Scientific realism comes in many forms. Most versions of sci-
entific realism display a commitment to at least two doctrines: (1)
that there is a real world of which we are part and (2) that both the
observable and unobservable features of that world can be known
by the proper use of scientific methods. Some versions of scien-
tific realism incorporate additional theses (e.g., the claims that
truth is the primary aim of science, and that successive theories
more closely approximate the truth), and some will also nominate
optional doctrines that may, but need not, be used by scientific
realists (e.g., the claim that causal relations are relations of natural
necessity; see Hooker, 1987). Others who opt for an “industrial
strength” version of scientific realism for the physical sciences are
more cautious about its successful reach in the behavioral sciences.
In philosophy, J. D. Trout (1998), for example, subscribes to a mod-
est realism in psychology, based on his skepticism about the disci-
pline’s ability to produce deeply informative theories like those of
the physical sciences. In psychology, James Grice (2011) presents a
4
philosophy of moderate realism to underwrite his novel methodol-

ogy of “observation oriented modeling.” This philosophy maintains
that things have essences, that their natures are knowable, and that
a strategy of modeling can be used to integrate knowledge about
the systems under study. Grice shows in a general way how philos-
ophy of science can make an important contribution to scientific
methodology.
Scientific realism boasts a rich conception of methodology,
which is of considerable help in understanding and guiding
research. The resourcefulness of realist methodology is suggested
in the following description of its major characteristics (cf. Haig,
2014; Hooker, 1987; Nickles, 1987): First, realist methodology has
three major tasks: to describe how methods function; to evaluate
methods critically against their rivals; and to recommend how to
use particular methods to pursue chosen research goals. I hope
that my concern with these tasks is evident in the treatment of the
methods in the following chapters.
Second, realist methodology is critically aim oriented. At a
broad level, it recommends the pursuit of valuable truth, explan-
atory understanding, and effective control as primary research
goals; it is also concerned with the mutual adjustment of meth-
ods and research goals. At a more specific level, my discussion
of methods attempts to give due recognition to their appropriate
research goals.
Third, realist methodology is naturalistic; that is to say, it is a
substantive domain that uses the methods of the various sciences
to study methods themselves. The error-statistical perspective pre-
sented in Chapter 3 is a philosophy of statistics that sits squarely
within the naturalistic tradition in modern philosophy. Proctor
and Capaldi (2001) advocate a naturalistic approach to methodol-
ogy in psychology in which the empirical justification of method-
ological ideas is emphasized.
A fourth feature of realist methodology is that it is both gen-
erative and consequentialist. Generative methodology involves
reasoning to, and accepting, knowledge claims in question from
warranted premises. Exploratory factor analysis is a prominent
example of a method in psychology that involves a generative jus-
tification of the factorial hypotheses to which it gives rise. By con-
trast, consequentialist methodology focuses on reasoning from
knowledge claims in question to their testable consequences. The
5
Introduction : 5
widely used hypothetico-deductive method, with its emphasis on

predictive accuracy, clearly exhibits a consequentialist approach to
justifying knowledge claims.
Fifth, realist methodology acknowledges the need for two quite
different approaches to justifying knowledge claims. In philoso-
phy, these are commonly known as reliabilism and coherentism.
With reliabilism, a belief is justified to the extent that it is acquired
by reliable processes. In general, the innumerable methods that
contribute to the detection of empirical phenomena are concerned
with reliabilist justification. With coherentism, a belief is justi-
fied in virtue of its coherence with other beliefs. Thagard’s (1992)
theory of explanatory coherence (which is not considered in this
book) is used for the comparative evaluation of scientific theories
and embodies an illuminating coherentist perspective on knowl-
edge justification. These two forms of justification are different,
complementary, and of equal importance.
As a sixth feature, realist methodology regards science as a
problem-oriented endeavor in which problems are conceptual-
ized as constraints on their effective solution (Haig, 1987; Nickles,
1981). On this formulation, the constraints are actually constitu-
tive of the problem itself; they characterize the problem and give it
structure. Further, by including all the constraints in the problem’s
articulation, the problem enables the researcher to direct inquiry
effectively by pointing the way to its own solution. In a real sense,
stating the problem is half the solution! This focus on research
problems holds more promise for research inquiry than the cus-
tomary talk about research questions.
Finally, realist methodology takes the researcher’s makeup as a
“knowing subject” seriously. Among other things, the researcher
is regarded as a satisficer who makes heavy use of heuristics to
guide inquiries. McGuire (1997), for example, discusses many use-
ful heuristics that can be employed to facilitate the generation of
hypotheses in psychological research.
Scientific realist methodology undergirds a wide variety of
methods, strategies, and heuristics that have been successfully
used to produce worthwhile knowledge about both empirical
phenomena and explanatory theories. If quantitative researchers
in psychology fully engage this literature, they will find resources
for enhancing their understanding of research methods and the
proper uses to which they can be put.
6
Theories of Scientific Method

Modern science is a multifaceted endeavor. A full appreciation of
its nature needs to consider the aims it pursues, the theories it pro-
duces, the methods it employs, and the institutions in which it is
embedded. Although all of these features are integral to science,
science is most illuminatingly characterized as method. Method is
central to science because much of what we have learned from sci-
ence has been acquired through use of its methods. Our scientific
methods have been acquired in the course of learning about the
world; as we learn, we use methods and theorize about them with
increased understanding and success. Applied to science, method
suggests the efficient, systematic ordering of inquiry. Scientific
method, then, describes a sequence of actions that constitute a
strategy to achieve one or more research goals. Relatedly, scien-
tific methodology denotes the general study of scientific methods
and forms the basis for a proper understanding of those methods.
Modern scientific methodology has given considerable attention
to a number of general theories of scientific method. Here, I sketch
three theories that figure in the chapters that follow: inductive
method, hypothetico-deductive method, and abductive method.
These theories of method provide different orientations to the
more specific research methods considered in Chapters 2–6.
The idea that scientific method involves inductive reasoning
goes back at least to Aristotle and was given heavy emphasis by
Francis Bacon and John Stuart Mill. Inductive reasoning takes dif-
ferent forms. For example, it is to be found in the fashioning of sta-
tistical generalizations, in the Bayesian assignment of probabilities
to hypotheses, and even in the reasoning involved in moving from
data to hypotheses in the hypothetico-deductive method. In psy-
chology, the radical behaviorism of B. F. Skinner is a prominent
example of a research tradition that makes use of an inductive con-
ception of scientific method.
The most popular account of method in science is the
hypothetico-deductive method. It has come to assume hegemonic
status in the behavioral sciences and places a heavy emphasis on
testing hypotheses in terms of their predictive success. Relatedly,
the use of traditional statistical significance test procedures in
psychology is often embedded in a hypothetico-deductive struc-
ture. With the hypothetico-deductive method, the scientist takes
7
Introduction : 7
a hypothesis or a theory and tests it indirectly by deriving from it

one or more observational predictions that are amenable to direct
empirical test. If the predictions are borne out by the data, then
that result is taken as a confirming instance of the theory in ques-
tion. If the predictions fail to square with the data, then that fact
counts as a disconfirming instance of the theory.
According to the abductive theory of method (Haig, 2014), sci-
entific inquiry is a problem-solving endeavor in which sets of data
are analyzed to detect robust empirical regularities, or phenom-
ena. Once detected, these phenomena are explained by abduc-
tively inferring the existence of underlying causal mechanisms.
On positive judgments of the initial plausibility of these explana-
tory theories, attempts are made to elaborate on the nature of the
causal mechanisms in question. This is done by constructing plau-
sible models of those mechanisms by analogy with relevant ideas
in domains that are well understood. When the theories are well
developed, they are assessed against their rivals in respect of their
explanatory goodness. This assessment involves making judg-
ments of the best of competing explanations. This abductive theory
of method can serve as a useful framework for locating a number
of more specific research methods within its fold.
Book Overview and Chapter Summary

This book undertakes a critical, in-depth examination of a selec-
tion of well-known, or otherwise important, quantitative research
methods that are, or can be, used in behavioral science research.
The book is interdisciplinary in nature and draws from varied
literatures in research methodology, including the philosophy of
science and statistical theory. As such, it is intended to serve as a
useful complement to other books in the Understanding Statistics
series. For example, the conceptual treatment of the method of
exploratory factor analysis offered in Chapter 6 of the present book
fits well with Fabrigar and Wegener’s (2012) book in the series,
Exploratory Factor Analysis.
In writing this book, my primary goal is to examine the concep-
tual foundations of a range of behavioral research methods. Some
of them are well known. Others are seldom considered by behav-
ioral science methodologists and researchers. A conceptual under-
standing of those methods is facilitated by presenting them in
8
relation to prominent accounts of scientific method where appro-

priate. The critical nature of the book is a natural consequence of
dealing squarely with a conception of research methodology that
is sponsored by the philosophy of scientific realism.
Chapter 1: This introductory chapter provides key ideas that
should help make sense of the treatment of the five methods dealt
with in the book. It begins by highlighting the importance and
relevance of philosophy of science for understanding quantitative
methods. It then gives a brief overview of the prominent philoso-
phy of scientific realism, with particular emphasis on the nature of
scientific methodology. After that, three major theories of scien-
tific method are sketched because they figure in some of the ensu-
ing chapters. Finally, an overview of the book’s contents is provided
before providing a note to the reader.
Chapter 2 focuses mainly on the nature, role, and importance
of exploratory data analysis in behavioral research, although
some attention is also given to the companion movement of
computer-intensive statistics and its use of a reliabilist approach
to justifying the knowledge claims it produces. Four perspectives
on exploratory data analysis are presented, as they are shaped by
different accounts of scientific method. One of these, the abduc-
tive theory of scientific method, locates exploratory data analy-
sis in a multistage model of data analysis. Finally, John Tukey’s
outline of a philosophy for teaching data analysis is presented as
an important part of an overall philosophy of exploratory data
analysis.
Regarding Chapter 3, although widely used in behavioral science
research, tests of statistical significance are poorly understood. In
this critical examination of tests of significance, I discuss the ques-
tionable use of a popular hybridized form of significance testing in
psychological research before outlining two plausible views of tests
of significance: the neo-Fisherian and error statistical perspectives.
These are judged to be superior to the hybrid version, especially
that sponsored by the error-statistical account, which is a coherent
philosophy of statistics. It is suggested that tests of significance can
play a useful, if limited, role in research.
The subject of Chapter 4 is Bayesianism, which comprises both
a philosophical theory of scientific confirmation and an influ-
ential perspective on statistics. I describe the nature of Bayesian
9
Introduction : 9
confirmation theory and assess its difficulties. I then compare it

to two rivals: the hypothetico-deductive method and inference to
the best explanation. In addition, I present, and evaluate, a neo-
Popperian philosophy of Bayesian statistics, which is offered as an
alternative to standard Bayesian modeling practice.
The primary concern of Chapter 5 is with the conceptual foun-
dations of meta-analysis. The examination centers on large-scale
issues having to do with meta-analysis and the nature of science.
I give considerable space to presenting the conception of inquiry
embodied in the underlying rationale of Gene Glass’s approach to
meta-analysis. I then examine David Sohn’s provocative argument
that meta-analysis is not a proper vehicle of scientific discovery.
After that, I consider the role of meta-analysis in relation to the
different processes of phenomena detection and scientific explana-
tion. In doing so, I examine the extent to which meta-analysis can
properly be said to contribute to scientific progress.
Chapter 6 examines the logic and purpose of exploratory factor
analysis. It is argued that the common factors of exploratory factor
analysis are not fictions, but latent variables best understood as
genuine theoretical entities. This realist interpretation of factors is
supported by showing that exploratory factor analysis is an abduc-
tive generator of elementary theories that exploits an important
heuristic of scientific methodology known as the principle of the
common cause. The importance of exploratory factor analysis is
affirmed, and it is argued that it can be usefully combined with
confirmatory factor analysis.
The concluding Chapter 7 assembles a number of important les-
sons learned from the preceding chapters before emphasizing the
need for further work in the philosophy of research methods.
A Note for the Reader

The books in the Understanding Statistics series are fairly short in
length. Thus, because I wanted to examine each method in this
book in some conceptual detail, a limited number of methods were
selected for consideration. Briefly, the reasons for my selection
were as follows: Exploratory data analysis has a major role in pat-
tern detection, and despite being advocated by the eminent statis-
tician John Tukey for more than 50 years, it has not found a regular
10
place in the behavioral research method curriculum. The same can

be said for computer-intensive resampling methods, which only
arrived on the scene with the advent of high computing power. The
reasons for the selection of tests of statistical significance should
be obvious. They are overused in behavioral research and yet are
poorly understood. It is high time that these failings were put right.
The Bayesian approach to quantitative thinking earns its place in
the book because it stands as the best-known theory of scientific
confirmation, as well as the major rival school of thought to fre-
quentist tests of statistical significance. Although the advocacy of
Bayesian methods is on the rise, they also fail to figure regularly in
the methods curriculum. Meta-analysis is a comparatively recent
development in scientific methodology, but it has quickly become
the dominant approach to reviewing primary empirical studies in
the behavioral sciences. However, its conceptual foundations are
only occasionally addressed. Finally, the long-standing method
of exploratory factor analysis has been widely used in psychol-
ogy, and many other sciences, for more than 70 years. Yet, its deep
structure is seldom considered by those who describe it. It stands
as our best example of a method that is well suited to the genera-
tion of explanatory hypotheses and theories.
The amount of space given to a description of each method
varies considerably. Well-known methods, such as tests of sta-
tistical significance and exploratory factor analysis, receive little
exposition. Less well-known methods, such as exploratory data
analysis and resampling methods, receive more. Partly for this
reason, and partly because of the paucity of a philosophical lit-
erature on these methods, their treatment in this book will seem
less philosophical.
Finally, none of the methods considered in the book receives
a full examination of its conceptual foundations. Relatedly, the
reader should not expect from this book a series of definitive
assessments about how one should understand and use the differ-
ent methods. The book was written primarily as a stimulus for the
reader to develop personal thinking about the methods considered
in a manner that goes beyond what is contained in usual book pre-
sentations. The “Further Reading” section provided for each chap-
ter should help the reader to extend thinking well beyond what the
chapters themselves contain.
1
Introduction : 11
References
Fabrigar, L. R., & Wegener, D. T. (2012). Exploratory factor analysis. New York,
NY: Oxford University Press.
Grice, J. (2011). Observation oriented modeling: Analysis of cause in the behavioral
sciences. New York, NY: Academic Press.
Greenwood, J. D. (1992). Realism, empiricism, and social constructionism.
Theory and Psychology, 2, 131–151.
Haig, B. D. (1987). Scientific problems and the conduct of research. Educational
Philosophy and Theory, 19, 22–32.
Haig, B. D. (2014). Investigating the psychological world: Scientific method in the
behavioral sciences. Cambridge, MA: MIT Press.
Hooker, C. A. (1987). A realistic theory of science. New York, NY: State University
of New York Press.
Kuhn, T. S. (1996). The structure of scientific revolutions (3rd ed.). Chicago,
IL: University of Chicago Press (originally published 1962).
Manicas, P. T., & Secord, P. F. (1983). Implications for psychology of the new phi-
losophy of science. American Psychologist, 38, 399–413.
McGuire, W. J. (1997). Creative hypothesis generating in psychology: Some useful
heuristics. Annual Review of Psychology, 48, 1–30.
Mulaik, S. A. (1985). Exploratory statistics and empiricism. Philosophy of Science,
52, 410–430.
Nickles, T. (1981). What is a problem that we might solve it? Synthese, 47,
85–118.
Nickles, T. (1987). ’Twixt method and madness. In N. J. Nersessian (Ed.), The
process of science (pp. 41–67). Dordrecht, the Netherlands: Nijhoff.
Proctor, R. W., & Capaldi, E. J. (2001). Empirical evaluation and justification
of methodologies in psychological science. Psychological Bulletin, 127,
759–772.
Psillos, S. (1999). Scientific realism: How science tracks the truth. London,
England: Routledge.
Thagard, P. (1992). Conceptual revolutions. Princeton, NJ: Princeton University Press.
Trout, J. D. (1998). Measuring the intentional world: Realism, naturalism, and
quantitative methods in the behavioral sciences. New York, NY: Oxford
University Press.
Yu, C. H. (2006). Philosophical foundations of quantitative research methodology.
Lanham, MD: University Press of America.
12
13
2
Exploratory Data
Analysis
Exploratory data analysis is an attitude, a flexibility, and a

reliance on display, NOT a bundle of techniques, and should
be so taught.
—J. W. Tukey, 1980
[T]‌he time is ripe for a broad conceptualization of data
analysis that includes the principles and procedures of EDA.
—J. T. Behrens, 1997
Introduction
During the last 80 years, data analysis in statistics has placed its
major emphasis on classical statistical inference, where the pri-
mary goal is to find out whether a set of data exhibits a designated
feature of interest, associated with a probabilistic model. Such an
approach to data analysis favors confirmatory research in which
hypotheses are tested using methods such as tests of statistical sig-
nificance (the topic of Chapter 3). Unfortunately, the dominance
of this data analytic practice has had the effect of discouraging
the genuine exploratory examination of data sets in terms of their
quality and structure. Detailed explorations of data are essential
14
for detecting patterns, and it makes good sense to undertake such

explorations instead of, or before, a probabilistic model is formu-
lated and adopted.
However, since the early 1960s, a new empirical approach to data
analysis in statistics has emerged. One important part of this devel-
opment is exploratory data analysis (EDA), a process in which data
are examined to reveal potential patterns of interest (e.g., Tukey,
1977, 1980). Another important part of this empirical data analytic
movement is the advent of computer-intensive resampling methods,
which, through repeated sampling of observed data, are used to
produce a reference distribution (Efron & Tibshirani, 1993). These
modern confirmatory resampling methods, rather than traditional
confirmatory methods, can be employed as suitable complements
to exploratory methods.
In 1968, the prominent statistician and scientist John Tukey
gave an invited talk to the annual convention of the American
Psychological Association; the talk was published in the associa-
tion’s flagship journal, the American Psychologist, the following year
(Tukey, 1969). In his article, “Analyzing Data: Sanctification or
Detective Work?” Tukey argued for the need to practice EDA, in the
manner of a detective, as well as the confirmatory, or judicial, mode.
Unfortunately, psychology has, for the most part, ignored Tukey’s
advice, preferring instead to continue its focus on classical confirm-
atory methods. A comprehensive survey of all psychology PhD pro-
grams in the United States and Canada in 1986 revealed that only
20 percent of introductory statistics courses gave in-depth coverage
to the topic of EDA (Aiken, West, Sechrest, & Reno, 1990). A rep-
lication and extension of that survey in the late 1990s contained no
explicit information on this topic, though it did include coverage of
“modern graphical displays” at 10 percent. An informal inspection
of current standard textbooks on statistical methods in psychology
shows that they give little attention to the topic of EDA. Occasional
prominent calls in psychology recommending greater use of explor-
atory data analytic methods (e.g., Behrens, 1997; Wilkinson & the
Task Force on Statistical Inference, 1999) seem to have made little
difference. Clearly, the use of traditional confirmatory methods in
data analysis remains the dominant practice.
This chapter is concerned with modern data analysis. It focuses
primarily on the nature, role, and importance of EDA, although it
gives some attention to the companion topic of computer-intensive
15
Exploratory Data Analysis : 15
confirmatory methods. Because exploratory data analytic and

computer-intensive resampling methods are less well known to
behavioral scientists than the methods considered in other chap-
ters, they receive more expository attention here than do key meth-
ods from other chapters. Considerable attention is also given to
somewhat different perspectives on data analysis as they are shaped
by four different accounts of scientific method. Before concluding,
the chapter offers a brief presentation and discussion of Tukey’s val-
uable, but underappreciated, philosophy of teaching data analysis.
It should be pointed out that the present chapter focuses on
EDA in the manner of Tukey, given that such a view of EDA has
not yet found its way into the modern behavioral science methods
curriculum. The chapter does not consider the more recent explor-
atory data analytic developments, such as the practice of statistical
modeling, the employment of data-mining techniques, and more
flexible resampling methods (see, e.g., Yu, 2010).
What Is Exploratory Data Analysis?

In his landmark article, “The Future of Data Analysis” (Tukey,
1962), Tukey introduced the term data analysis to distinguish
applied statistical work from the then-dominant formal infer-
ential statistics. He characterized data analysis broadly in the
following words:
Data analysis . . . I take to include, among other things: pro-
cedures for analyzing data, techniques for interpreting the
results of such procedures, ways of planning the gathering of
data to make its analysis easier, more precise, more accurate,
and all the machinery and results of (mathematical) statistics
which apply to analyzing data.
Large parts of data analysis are inferential in the sample-
to-population sense, but these are only parts, not the whole.
Large parts of data analysis are incisive, laying bare indica-
tions which we could not perceive by simple and direct exam-
ination of the raw data, but these too are parts, not the whole.
Some parts of data analysis . . . are allocation, in the sense that
they guide us in the distribution of effort. . . . data analysis
is a larger and more varied field than inference, or incisive
procedures, or allocation. (Tukey, 1962, p. 2)
16
For Tukey, data analysis is both an empirical science and an

art; it is not mathematical statistics (Tukey, in fact, advised serious
students of data analysis that it was better that they aspire to be
first-rate scientists rather than second-rate mathematicians); data
analysis places a heavy emphasis on judgment, and it accepts satis-
factory, not optimal, solutions to problems. Adopting an appropri-
ate set of attitudes is an important part of Tukey’s approach to data
analysis. These attitudes include investigating realistic problems,
making liberal use of ad hoc informal procedures in exploratory
work, accepting the approximate nature of results, employing pro-
cedures iteratively, and including both exploratory and confirma-
tory approaches in the same analyses.
As the intellectual progenitor of modern EDA, Tukey developed
a distinctive perspective on the subject that has helped to highlight
its importance to research. It deserves to be considered as a philos-
ophy of EDA in its own right and, more broadly, as a philosophy of
data analysis (Dempster, 2003). Therefore, this brief examination of
the philosophy of EDA pays particular attention to Tukey’s thinking
on the topic. As will be seen, Tukey took EDA to include more than
descriptive statistics. Although both EDA and descriptive statistics
are concerned with the visual display of data, EDA is also concerned
with recognizing patterns in the data, tentatively forming hypoth-
eses, and adopting the attitude of a detective when analyzing data.
According to Tukey (1980), data analysis should be treated as a
two-stage, compound process in which the patterns in the data are
first suggested by EDA and then critically checked through the use
of confirmatory data analytic procedures. As already noted, EDA
involves descriptive, and frequently quantitative, detective work
designed to reveal structure or pattern in the data sets under scru-
tiny. The data analyst is encouraged to undertake an open-eyed
investigation of the data and perform multiple analyses using a
variety of intuitively appealing and easily used techniques.
The compendium of methods for the exploration of data,
many of which were developed by Tukey (1977), sometimes in
association with colleagues, is designed to facilitate both discov-
ery and communication of information. These methods are con-
cerned with the effective organization of data, the construction
of graphical and semigraphical displays, and the examination
of distributional assumptions and functional dependencies. As
noted further in this chapter, two additional attractive features
of Tukey’s methods are their resistance to changes in underlying
17
distributions and their resistance to outliers in sets of data.

Exploratory methods with these two features are particularly
suited to data analysis in psychology and the behavioral sci-
ences, where researchers are frequently confronted with ad hoc
sets of data on amenable variables, which have been acquired in
convenient circumstances.
In the next section, I briefly describe the two best known
methods of EDA—stem-and-leaf displays and box-and-whisker
plots—and then provide an overview of the four focal themes of
EDA: resistance, residuals, re-expression, and revelation.
Two Methods of Exploratory Data Analysis

The Stem-and-Leaf Display
The stem-and-leaf display is a clever, useful, and easily understood
device that can provide helpful first impressions of a set of data.
The stem-and-leaf display economically organizes values from a
set of data in numerical order, while visually displaying the shape,
spread, and distributional characteristics in the manner of a histo-
gram. Unlike a histogram, however, the stem-and-leaf-display can
be readily constructed by hand, and unlike most displays, it has the
virtue of retaining information on individual data values. A stem-
and-leaf display will decompose each value into a stem value and
a leaf value. For example, for an ordered set of data values, each
tens digit would comprise the vertical stem column, while the units
digit would be arrayed horizontally as the leaf values. Stem-and-
leaf displays can also be placed back to back to compare similar
batches of data for detailed differences.
The Box-and-Whisker Plot

The box-and-whisker plot, sometimes simply called the box plot,
is also helpful in the presentation of data. The box plot is a graph
comprising rectangular boxes and lines that display the distribu-
tional shape, central tendency, and variability of a set of obser-
vations. It is particularly helpful for indicating skewness and the
presence of outliers in a distribution. From a box plot we can
obtain the “five-number summary” of a set of data. This summary
comprises the minimum and maximum values within a set, along
with the median and the lower and upper quartiles. In effect, the
18
box plot provides an effective graphical display of the five-number

summary. Because of this, the box plot is particularly useful for
comparing two or more sets of data.
Of course, there are many more methods suitable for EDA, but
the two just noted should give a sense of how semigraphical meth-
ods, which combine both graphical and tabular information, help
meet the primary goals of exploring data.
The Four Rs of Exploratory Data Analysis

Tukey and his collaborators (Hoaglin, Mosteller, & Tukey,
1983) argued that the tools of EDA can be organized according to
four central themes: resistance, residuals, re-expression, and reve-
lation. These themes usefully help to distinguish EDA from classi-
cal inferential statistics.
1. Resistance. An analysis, or summary, of the data is resistant
if it is not sensitive to misbehaving or unusual data. This
requirement of EDA is not met when unexpected data
values mislead the summary of the bulk of the data,
as happens when data misbehave, are unusual, or are
otherwise misleading. These situations occur often enough
to make resistance an important consideration in EDA.
For example, in EDA, the median is the most commonly
used measure of central tendency because it is not affected
by outliers as much as the mean; it is a more resistant, or
robust, statistic.
2. Residuals. The careful examination of residuals is important
in EDA. They are the deviations of observations from
the value predicted by a tentative model. In EDA (and in
confirmatory data analysis), residuals are summarized to
obtain a sense of overall fit. Stated differently, EDA uses the
framework
data = fit + residual
Hence, data residuals are what remain of the data after

a summary or the fitted model has been subtracted,
according to the equation
residual = data − fit

19
As leftover parts of the data, residuals still need attention,

which may involve a reanalysis and refit.
3. Re-expression. Re-expression is the term used in EDA
for rescaling or transforming data. Re-expression is
common in EDA because data are often collected in
a manner based on convenience, or habit, rather than
careful attention to scaling. EDA looks to find more easily
interpreted scales through re-expression. Re-expressions
that lead to symmetric distributions are preferred for
they promote the interpretation of general linear models,
improve comparisons across groups, and reflect the
structure of the sampling distributions. In psychology,
logarithmic, arcsine, and reciprocal transformations are
often used.
4. Revelation (Display). Displays are often graphical (and
semigraphical) in nature. They allow the data analyst to
see the behavior of the data and, as the analysis proceeds,
also the behavior of the residuals and various diagnostic
measures. EDA emphasizes the frequent use of displays
to ensure that unexpected features of the data are not
overlooked. Traditional EDA emphasizes a number of
relatively simple numerical and graphical techniques for
displaying data. For example, box-and-whisker plots are
often used to represent univariate data. However, significant
advances in graphics and data visualization in more recent
times have enabled data analysts to construct all manner
of revealing displays of complex phenomena (e.g., Chen,
Härdle, & Unwin, 2008).
Psychology’s attitudes toward each of the four Rs deserve

comment. Checking for resistance, or robustness, of meth-
ods to violation of their assumptions with sets of data is done
much less frequently than should be the case. The framework,
residual = data – fit, is widely employed in psychology, but almost
entirely in the context of confirmatory data analysis (I comment
further on Tukey’s reluctance to accord models a role in EDA).
Re-expression of data through transformation is frequently done
in psychology, although typically with limited knowledge of preva-
lence of different types of distribution in its varied subject domain
20
(Micceri, 1989). Finally, the importance of revelation, or data dis-

play, in EDA affirms the adage that a picture is worth a thousand
words, an insight that is founded on the fact that humans (and
other vertebrates) are primarily visual creatures (Gould, 1994).
Recently, psychology has given more attention to the value of good
data displays (e.g., Lane & Sándor, 2009), although the major focus
is seldom on EDA.
I turn now to an examination of the philosophical foundations
of EDA by considering the role it plays in four different accounts
of scientific method.
Exploratory Data Analysis and Scientific Method

In his writings on data analysis, Tukey emphasizes the related ideas
that psychology is without an agreed-upon model of data analysis,
and that we need to think more broadly about scientific inquiry.
In his address to the American Psychological Association men-
tioned previously, Tukey (1969) presented the following anony-
mous excerpt from a prominent psychologist for his audience to
ponder. I quote in part:
I have the feeling that psychology is currently without a dom-
inant viewpoint concerning a model for data analysis. In the
forties and early fifties, a hypothetico-deductive framework was
popular, and our mentors were keen on urging the design of “cru-
cial” experiments for the refutation of specific predictions made
from one or another theory. Inductive empiricism was said to be
disorderly and inefficient. You and I knew then, as we know now,
that no one approach is uniformly most powerful. (p. 90)
It is not surprising that mention is made of hypothetico-deductive
and inductive inquiry in the quotation for these two outlooks are
generally acknowledged as the two most influential conceptions of
scientific method in the history of science (Laudan, 1981). In what
follows, I consider EDA in relation to the hypothetico-deductive
and inductive methods of science. I then discuss Tukey’s proposed
framework of inquiry, which he believes properly accommodates
EDA, before considering EDA in relation to an abductive theory of
scientific method. My treatment should be understood as endors-
ing the assertion in the quotation that there is no one account of
scientific method that is best for all occasions.
21
Exploratory Data Analysis and

the Hypothetico-Deductive Method
According to the standard conception of the hypothetico-deductive
method, a scientist takes a hypothesis or a theory and tests it by
deriving from it one or more observational predictions, which are
amenable to a direct empirical test. If the predictions are borne out
by the data, then that result is taken as a confirming instance of the
theory. If the predictions fail to square with the data, then that fact
counts as a disconfirming instance of the theory.
Most psychological researchers continue to undertake their
research within the confines of this conception of hypothetico-
deductive method. Witness their heavy preoccupation with the-
ory testing, where confirmatory data analyses are conducted on
limited sets of data gathered in accord with the dictates of the test
predictions of theories. In this regard, psychologists frequently
employ tests of statistical significance to obtain binary decisions
about the credibility of the null hypothesis and its statistical or sub-
stantive alternative (see Chapter 3). However, the heavy use of tests
of statistical significance in this way strongly discourages research-
ers from looking for more interesting patterns in the data of poten-
tial interest. Indeed, the continued neglect of EDA in psychological
research occurs in good part because there is no acknowledged
place for such work in the hypothetico-deductive conception
of inquiry (Wilkinson & the Task Force on Statistical Inference,
1999). It is important to understand that the hypothetico-deduc-
tive method itself is not at fault here for it is a confirmatory pro-
cedure, not an exploratory procedure. Rather, it is the dominating
use of the hypothetico-deductive method by researchers that
clouds their ability to appreciate EDA as an important element of
scientific research.
Exploratory Data Analysis and the Inductive Method

The most popular characterization of inductive scientific method
maintains that inquiry begins by securing observed facts, which
are collected in a theory-free manner. These facts provide a
firm base from which the scientist reasons “upward” to hypoth-
eses, laws, or theories. The reasoning involved takes the form of
enumerative induction and proceeds in accordance with some
2
governing principle of inductive reasoning. This rather simple

view of inductive method can be defended in a moderate form. In
psychology, the radical behaviorism of B. F. Skinner (1984) makes
use of a nonstatistical inductive conception of scientific method.
The major goals of Skinner’s conception of inductive method are,
first, to detect empirical generalizations about the subject matter
of interest and then to systematize those empirical generalizations
by assembling them into nonexplanatory theories.
I think the worth of the inductive method as a model for data
analysis is dismissed too quickly in the previous quotation from
Tukey (1969). The major limitation of the inductive account of
scientific method lies not so much with its perspective on data
analysis but with the prohibition of the formulation of explan-
atory theories by many of its proponents. It will be seen shortly
that a conception of inductive method is embedded in the broader
abductive account of scientific method.
Exploratory Data Analysis and Tukey’s Model of Inquiry

Tukey (1980) believes that statisticians and data analysts have given
too little attention to broad concerns about inquiry. He maintains
that much data analysis proceeds according to the following linear
conception of confirmatory research:
Q uestion → Design → Collection → Analysis → Answer
Tukey argues that this model of inquiry is incomplete and that it

neglects the following set of questions and answers, all of which
have to do with data exploration. He writes:
1. How are questions generated? (Mainly by quasi-theoretical
insights and the exploration of past data.)
2. How are designs guided? (Usually by the best qualitative
and semiquantitative information available, obtained by
exploration of past data.)
3. How is data collection monitored? (By exploring the data,
often as they come in, for unexpected behavior.)
4. How is analysis overseen; how do we avoid analysis that the
data before us indicate should be avoided? (By exploring the
data—before, during, and after analysis—for hints, ideas,
and sometimes, a few conclusions-at-5 percent/k.). (p. 23)
23
Tukey argues that, to pose and answer these questions, and

indeed to implement properly confirmatory research, we need to
reorganize the model of inquiry along the following lines:
Ides → Question →← Design → Collection → Analysis → Answer
This is to say, we “begin” not with a properly formulated question,

but an idea of a question that cannot be given an answer until it
is specified in terms of appropriate constraints. And, to do this
requires exploration, which if we are successful, will lead to a cir-
cumscribed question that warrants attempted confirmation. For
this reason, Tukey (1980) maintains that “finding the question is
often more important than finding the answer” (p. 23).
Clearly, if science is to derive maximum benefit from data anal-
ysis, it needs to take seriously both its exploratory and its confirm-
atory modes. It has already been noted that Tukey regards data
analysis as a two-stage compound process: exploratory followed
by confirmatory. However, in his more precise moments, Tukey
speaks of data analysis as a three-stage process, where the stages
lie on a continuum of data analysis (Tukey, 1972): The first stage is
that of EDA, where the investigator seeks to learn what is going on
in the data. The second stage Tukey calls rough confirmatory data
analysis. Here hypotheses are refined and rough tests are carried
out, often using estimation techniques such as confidence inter-
vals. In the third stage, known as strict confirmatory data analy-
sis, the investigator tests well-specified hypotheses using modern
robust statistical methods. Thus, it should be clear that, for Tukey,
confirmatory data analysis is just as important as EDA. The heavy
focus in his writings on EDA is more a function of their compara-
tive neglect than their greater importance.
Exploratory Data Analysis and the Abductive Method

As stated earlier, in the abductive theory of method (Haig, 2005,
2014), scientific inquiry proceeds as follows: Guided by evolving
research problems, sets of data are analyzed to detect robust empir-
ical regularities, or phenomena. Once detected, these phenomena
are explained by abductively inferring the existence of underlying
causes that are thought to give rise to the phenomena. Here, abduc-
tive inference involves reasoning from claims about phenomena,
24
understood as presumed effects, to their theoretical explanation

in terms of underlying causes. Upon positive judgments of the ini-
tial plausibility of these explanatory theories, attempts are made to
elaborate on the nature of the causal mechanisms in question. This
is done by constructing plausible models of those mechanisms by
analogy to relevant ideas in domains that are well understood. When
the theories are well developed, they are assessed against their rivals
with respect to their explanatory goodness. This assessment involves
employing criteria specifically to do with explanatory worth.
It should be apparent, even from this brief sketch, that the abduc-
tive theory is considerably broader than both the hypothetico-
deductive and inductive alternatives (and all three accounts go
beyond Tukey’s near-exclusive concern with data analysis). The
breadth of the abductive theory enables it to operate as a frame-
work theory within which an extensive array of data analytic and
theory construction methods and strategies can be usefully located.
Exploratory Data Analysis in a Multistage

Model of Data Analysis
The important place of EDA in the abductive theory of method can
be appreciated by describing its role in the process of phenomena
detection. Phenomena are relatively stable recurrent general fea-
tures of the world that we seek to explain (Haig, 2014; Woodward,
1989), and their detection frequently involves an inductive process
of empirical generalization. This inductive process of phenomena
detection reserves an important place for the exploratory analysis
of data. In detecting phenomena, one is concerned to extract a sig-
nal from the noise of data, and for this, the intensive search of large
amounts of data is frequently required. This is precisely because
securing a heavy information yield from our data is likely to pro-
vide potentially interesting data patterns that might turn out to be
genuine phenomena. In this context, data mining is encouraged,
and the capabilities of exploratory techniques in this regard often
make them the appropriate methods of choice.
A more fine-grained appreciation of the role of EDA in the discov-
ery of empirical phenomena can be gained by outlining a multistage
model of data analysis with EDA as one of its stages. This model,
which is featured in the abductive theory of method, describes one
of a number of ways in which empirical phenomena can be detected.
25
The model comprises the four sequenced stages of initial data analy-
sis, EDA, close replication, and constructive replication.
Initial Data Analysis. The initial examination of data (Chatfield,
1985) refers to the first informal scrutiny and description of data
that is undertaken before EDA proper begins. It involves screen-
ing the data for their quality. Initial data analysis variously involves
checking for the accuracy of data entries, identifying and dealing
with missing and outlying data, and examining the data for their fit
to the assumptions of the data analytic methods used. Data screen-
ing thus enables one to assess the suitability of the data for the type
of analyses intended. The initial analysis of data has much in com-
mon with Tukey’s approach to EDA. However, these two related
data analytic endeavors serve different primary functions (data
screening and pattern detection, respectively), and I restrict initial
data analysis to the preliminary scrutiny of data before exploratory
EDA (in Tukey’s sense) begins.
Exploratory Data Analysis. Given that EDA is the major focus
of this chapter, it suffices to say here that it plays an indispensable
role in the detection of patterns in data that are the springboard
to the eventual discovery of phenomena, or robust empirical
regularities.
Close Replication. Successfully conducted exploratory analyses
will suggest potentially interesting data patterns. However, it will
normally be necessary to check on the stability of the emergent
data patterns though use of appropriate confirmatory data analysis
procedures. Computer-intensive resampling methods such as the
bootstrap, the jackknife, and cross validation (Efron & Tibshirani,
1993) constitute an important set of confirmatory procedures that
are well suited to this role. They are briefly discussed in material
that follows.
Constructive Replication. In establishing the existence of phe-
nomena, it is often necessary to undertake both close and con-
structive replications. The statistical resampling methods just
mentioned are concerned with the consistency of sample results
that help researchers achieve close replications. By contrast, con-
structive replications are undertaken to demonstrate the extent to
which results hold across different methods, treatments, and occa-
sions. In other words, constructive replication is a triangulation
strategy designed to ascertain the generalizability of the results
identified by successful close replication (Lindsay & Ehrenberg,
26
1993). Constructive replication, in which researchers vary the

salient study conditions, is a time-honored strategy for justifying
claims about phenomena.
The four-stage model of data analysis just outlined assists in the
detection of phenomena by attending in turn to the different, but
related, tasks of data quality, pattern suggestion, pattern confirma-
tion, and generalization. To repeat, the role of EDA in this process
is that of pattern detection.
Exploratory Data Analysis and Abductive Inference

Having commented on the role of EDA in the process of phe-
nomena detection from the vantage point of the abductive theory
of scientific method, an important question remains: What is the
nature of the relations between EDA and abduction? In psychol-
ogy, I (Haig, 2013) and Behrens, Dicerbo, Yel, and Levy (2013)
(see also Behrens & Yu, 2003) have commented on philosophical
aspects of EDA. We hold contrasting views about the relevance
of abductive reasoning as a core component of the philosophy of
EDA. Behrens and his coauthors think abduction provides the
“core logic” of EDA. I disagree. In this section, I say why I think
their position is mistaken, and that their charge that mine is “a
particularly disturbing” view of EDA (Behrens et al., 2013, p. 39)
is unfounded.
Abduction as a form of inference is not well known in academic
circles. Broadly speaking, abduction is concerned with the gener-
ation and evaluation of explanatory hypotheses. In this sense, it
contrasts with the more familiar ideas of inductive and deductive
inference. Behrens et al. (2013) begin by taking their cue from the
philosopher-scientist Charles Peirce and state that abduction is the
form of inference involved in generating new ideas or hypotheses.
However, surprisingly, Behrens et al. do not stay with Peirce on
this matter. Instead, they closely follow Josephson and Josephson
(1994) and characterize abductive inference according to the fol-
lowing pattern of reasoning (p. 39):
D is a collection of data (facts, observations, givens).
Hypothesis H explains D (would if true, explain D).
No other hypothesis explains D as well as H does.
Therefore, H is probably correct.
27
Patently, this argument schema does not describe the abduc-

tive process of hypothesis generation. Instead, it characterizes
the abductive form of reasoning known as inference to the best
explanation. Inference to the best explanation is used in science
to appraise competing theories in terms of their explanatory
goodness (Thagard, 1992). For the schema to capture abductive
hypothesis generation, the third premise, which refers to com-
peting hypotheses, would have to be deleted, and the conclusion
would be amended to say that the hypothesis in question was ini-
tially plausible, not probably correct.
Despite the fact that philosophers of science sometimes speak
of abduction and inference to the best explanation as though they
were the same thing, it is important to differentiate between the
abductive generation of hypotheses and their comparative appraisal
in terms of inference to the best explanation. They are discernibly
different phases of theory construction. In short, Behrens et al.
(2013) adopt a conception of abduction that is ill-suited to expli-
cating the process of idea generation, whether it be pattern identifi-
cation through EDA or some other generative process. As a result,
they fail to make an instructive connection between their chosen
characterization of abduction and the reasoning involved in EDA.
However, my major worry is not that Behrens et al. (2013)
choose the wrong form of abduction to explicate the inferential
nature of EDA, but that they try to understand it by appealing to
abduction at all. The fundamental difference between our opposed
views can be brought out by drawing, and adhering to, the impor-
tant three-fold methodological distinction between data, phe-
nomena, and explanatory theory. Briefly, data are idiosyncratic to
particular investigative contexts, and they provide the evidence
for phenomena, which are recurrent general features of the world
that we seek to explain. In turn, phenomena are the appropriate
source of evidence for the explanatory theories that we construct
to understand empirical phenomena. I have just described one
way of detecting phenomena by outlining a multistage model of
data analysis. These stages of data analysis are concerned in turn
with assessing data quality, detecting data patterns, confirming
those patterns through use of computer resampling methods (a
prominent feature of Tukey’s conception of data analysis), and
establishing the reach of the confirmed relationships in the form
of inductive generalizations. Viewed in this context, EDA is an
Another random document with
no related content on Scribd:
The Project Gutenberg eBook of On the mode
of communication of cholera
This ebook is for the use of anyone anywhere in the United
States and most other parts of the world at no cost and with
almost no restrictions whatsoever. You may copy it, give it away
or re-use it under the terms of the Project Gutenberg License
included with this ebook or online at www.gutenberg.org. If you
are not located in the United States, you will have to check the
laws of the country where you are located before using this
eBook.
Title: On the mode of communication of cholera
Author: John Snow
Release date: February 7, 2024 [eBook #72894]
Language: English
Original publication: London: John Churchill, 1855
Credits: Richard Tonsing, MWS, (Maps provided by The UCLA

Fielding School of Public Health Website), and the
Online Distributed Proofreading Team at
https://www.pgdp.net (This book was produced from
scanned images of public domain material from the
Google Books project.)
*** START OF THE PROJECT GUTENBERG EBOOK ON THE

MODE OF COMMUNICATION OF CHOLERA ***
Transcriber’s Note:
New original cover art included with this eBook is
granted to the public domain.
ON THE
MODE OF COMMUNICATION
OF
CHOLERA.
BY
JOHN SNOW, M.D.,

MEMBER OF THE ROYAL COLLEGE OF PHYSICIANS, FELLOW OF THE
ROYAL MED. AND CHIR. SOCIETY, FELLOW AND VICE-PRESIDENT OF THE
MEDICAL SOCIETY OF LONDON.
Second Edition, much Enlarged.
LONDON:
JOHN CHURCHILL, NEW BURLINGTON STREET.
M.DCCC.LV.
LONDON:
T. RICHARDS, 37 GREAT QUEEN STREET.

PREFACE TO THE SECOND EDITION.
The first edition of this work, which was published in August 1849,
was only a slender pamphlet. I have, since that time, written various
papers on the same subject, which have been read at the Medical
Societies, and published in the medical journals. The present edition
contains the substance of all these articles, together with much new
matter, the greater part of which is derived from my own recent
inquiries.
I take this opportunity of expressing my thanks to the Registrar-
General for the facilities afforded me in making these inquiries.
I feel every confidence that my present labours will receive the
same kind consideration from the Medical Profession which has
been accorded to my former endeavours to ascertain the causes of
cholera.
Sackville Street, Piccadilly,

11 December, 1854.
CONTENTS.
PAGE
Outline of the history of cholera 1
Cases proving its communication from person to person 3
Cholera not communicated by means of effluvia 9
The pathology of cholera indicates the manner in which it is
communicated 10
Analysis of the blood and evacuations in cholera 11 and 14
Cholera is propagated by the morbid poison entering the alimentary
canal 15
Evidence of this mode of communication in the crowded habitations of
the poor 17
—— amongst the mining population 19
Instances of the communication of cholera through the medium of
polluted water, in Horsleydown 23
—— at Albion Terrace, Wandsworth Road 25
—— at Ilford and near Bath 32
—— at Newburn on the Tyne 33
—— at Cunnatore 35
—— in the Black Sea Fleet 36
—— in the neighbourhood of Broad St., Golden Sq. 38
—— at Hampstead West End (the water being carried from Broad
Street) 44
Explanation of the Map showing the situation of the deaths in and
around Broad Street, Golden Square 45
Table of attacks and deaths near Golden Square 49
Outbreak of cholera at Deptford caused by polluted water 55
Communication of cholera by means of the water of rivers which
receive the contents of the sewers 56
Influence of the water supply on the epidemic of 1832, in London 57
Table showing the mortality from cholera, and the water supply 58
Influence of the water supply on the epidemic of 1849, in London 60
Table showing this influence 62
Communication of cholera by Thames water in the autumn of 1848 66
New water supply of the Lambeth Company 68
Effect of this new supply in the epidemic of autumn 1853 69
Tables showing this effect 71 and 73
Intimate mixture of the water supply of the Lambeth with that of the 74
Southwark and Vauxhall Company
Opportunity thus afforded of gaining conclusive evidence of the effect
of the water supply on the mortality from cholera 75
Account of inquiry for obtaining this evidence 77
Result of the inquiry as regards the first four weeks of the epidemic of
1854 79
—— the first seven weeks of the same epidemic 82
Tables illustrating these results 84, 85
Inquiry of the Registrar-General respecting the effect of the water
supply of the above-mentioned Companies during the later period of
the epidemic 87
Comparison of the mortality of 1849 and 1854, in the districts supplied
by the above-named Companies 89
Effect of the water supply on the mortality from cholera amongst the
inmates of workhouses and prisons 91
Cholera in the district of the Chelsea Water Company 93
Effect of dry weather to increase the impurity of the Thames 95
Relation between the greater or less mortality from cholera in London
and the less or greater elevation of the ground 97
This relation shown to depend on the difference of water supply at
different elevations ib.
Effect of water supply on the mortality of cholera in Birmingham,
Leicester, etc. 98
—— in Exeter 99
—— in Hull 100
—— in York 101
—— in Dumfries, etc. 102
—— in Nottingham and Glasgow 103
—— in Paris and Newcastle-upon-Tyne 104
Limited inquiry respecting the effect of the water supply at Newcastle 107
Assent of the medical profession to the influence of polluted water on
the mortality from cholera 109
True explanation of this influence 110
Answers to certain objections 111
Circumstances connected with the history of cholera explained by the
principles advocated in this treatise, as,—
The duration of the epidemic in each place usually bears a direct
relation to the number of the population 115
The effect of season on the prevalence of cholera 117
Proportion of the sexes who die of cholera at different periods of 119
an epidemic, and under different circumstances
Proportion of deaths from cholera in certain occupations 121
Evidence that plague, yellow fever, dysentery, and typhoid fever are
communicated in the same way as cholera 125
Instances in which ague was caused by impure water 129
Measures required for the prevention of cholera and other diseases
which are communicated in the same way 133
Appendix, containing a list of the deaths from cholera which were
registered in the four weeks ending 5th August 1854, together with
the water supply of the houses in which the fatal attacks took place,
in all the sub-districts to which the supply of either the Southwark
and Vauxhall or the Lambeth Water Company extends 138
MAPS.
Map 1. Showing the deaths from cholera in Broad Street, Golden

Square, and the neighbourhood, from 19th August to 30th
September 1854. A black mark or bar for each death is placed in the
situation of the house in which the fatal attack took place. The
situation of the Broad Street Pump is also indicated, as well as that of
all the surrounding Pumps to which the public had access.
Map 2. Showing the boundaries of the Registrar-General’s districts
on the south side of the Thames in London, and also the water
supply of those districts.
ON THE
MODE OF COMMUNICATION OF CHOLERA.
The existence of Asiatic Cholera cannot be distinctly traced back

further than the year 1769. Previous to that time the greater part of
India was unknown to European medical men; and this is probably the
reason why the history of cholera does not extend to a more remote
period. It has been proved by various documents, quoted by Mr. Scot,[1]
that cholera was prevalent at Madras in the year above mentioned, and
that it carried off many thousands of persons in the peninsula of India
from that time to 1790. From this period we have very little account of
the disease till 1814, although, of course, it might exist in many parts of
Asia without coming under the notice of Europeans.
In June 1814, the cholera appeared with great severity in the 1st bat.
9th regt. N.I., on its march from Jaulnah to Trichinopoly; while
another battalion, which accompanied it, did not suffer, although it had
been exposed to exactly the same circumstances, with one exception.
Mr. Cruikshanks, who attended the cases, made a report, which will be
alluded to further on.
ON THE MODE In 1817, the cholera prevailed with unusual
OF virulence at several places in the Delta of the
COMMUNICATI Ganges; and, as it had not been previously
ON OF
seen by the medical men practising in that
CHOLERA.
part of India, it was thought by them to be a
new disease. At this time the cholera began to spread to an extent not
before known; and, in the course of seven years, it reached, eastward,
to China and the Philippine Islands; southward, to the Mauritius and
Bourbon; and to the north-west, as far as Persia and Turkey. Its
approach towards our own country, after it entered Europe, was
watched with more intense anxiety than its progress in other
directions. It would occupy a long time to give an account of the
progress of cholera over different parts of the world, with the
devastation it has caused in some places, whilst it has passed lightly
over others, or left them untouched; and unless this account could be
accompanied with a description of the physical condition of the places,
and the habits of the people, which I am unable to give, it would be of
little use.
There are certain circumstances, however, connected with the
progress of cholera, which may be stated in a general way. It travels
along the great tracks of human intercourse, never going faster than
people travel, and generally much more slowly. In extending to a fresh
island or continent, it always appears first at a sea-port. It never attacks
the crews of ships going from a country free from cholera, to one where
the disease is prevailing, till they have entered a port, or had
intercourse with the shore. Its exact progress from town to town cannot
always be traced; but it has never appeared except where there has
been ample opportunity for it to be conveyed by human intercourse.
There are also innumerable instances which prove the
communication of cholera, by individual cases of the disease, in the
most convincing manner. Instances such as the following seem free
from every source of fallacy.
I called lately to inquire respecting the death of Mrs. Gore, the wife of
a labourer, from cholera, at New Leigham Road, Streatham. I found
that a son of the deceased had been living and working at Chelsea. He
came home ill with a bowel complaint, of which he died in a day or two.
His death took place on August 18th. His mother, who attended on
him, was taken ill on the next day, and died the day following (August
20th). There were no other deaths from cholera registered in any of the
metropolitan districts, down to the 26th August, within two or three
miles of the above place; the nearest being at Brixton, Norwood, or
Lower Tooting.
The first case of decided Asiatic cholera in London, in the autumn of
1848, was that of a seaman named John Harnold, who had newly
arrived by the Elbe steamer from Hamburgh, where the disease was
prevailing. He left the vessel, and went to live at No. 8, New Lane,
Gainsford Street, Horsleydown. He was seized with cholera on the
22nd of September, and died in a few hours. Dr. Parkes, who made an
inquiry into the early cases of cholera, on behalf of the then Board of
Health, considered this as the first undoubted case of cholera.
Now the next case of cholera, in London, occurred in the very room
in which the above patient died. A man named Blenkinsopp came to
lodge in the same room. He was attacked with cholera on the 30th
September, and was attended by Mr. Russell of Thornton Street,
Horsleydown, who had attended John Harnold. Mr. Russell informed
me that, in the case of Blenkinsopp, there were rice-water evacuations;
and, amongst other decided symptoms of cholera, complete
suppression of urine from Saturday till Tuesday morning; and after this
the patient had consecutive fever. Mr. Russell had seen a great deal of
cholera in 1832, and considered this a genuine case of the disease; and
the history of it leaves no room for doubt.
The following instances are quoted from an interesting work by Dr.
Simpson of York, entitled “Observations on Asiatic Cholera”:—“The
first cases in the series occurred at Moor Monkton, a healthy
agricultural village, situated to the north-west of York, and distant six
miles from that place. At the time when the first case occurred, the
malady was not known to be prevailing anywhere in the
neighbourhood, nor, indeed, at any place within a distance of thirty
miles.
“John Barnes, aged 39, an agricultural labourer, became severely
indisposed on the 28th of December 1832; he had been suffering from
diarrhœa and cramps for two days previously. He was visited by Mr.
George Hopps, a respectable surgeon at Redhouse, who, finding him
sinking into collapse, requested an interview with his brother, Mr. J.
Hopps, of York. This experienced practitioner at once recognised the
case as one of Asiatic cholera; and, having bestowed considerable
attention on the investigation of that disease, immediately enquired for
some probable source of contagion, but in vain: no such source could
be discovered. When he repeated his visit on the day following, the
patient was dead; but Mrs. Barnes (the wife), Matthew Metcalfe, and
Benjamin Muscroft, two persons who had visited Barnes on the
preceding day, were all labouring under the disease, but recovered.
John Foster, Ann Dunn, and widow Creyke, all of whom had
communicated with the patients above named, were attacked by
premonitory indisposition, which was however arrested. Whilst the
surgeons were vainly endeavouring to discover whence the disease
could possibly have arisen, the mystery was all at once, and most
unexpectedly, unravelled by the arrival in the village of the son of the
deceased John Barnes. This young man was apprentice to his uncle, a
shoemaker, living at Leeds. He informed the surgeons that his uncle’s
wife (his father’s sister) had died of cholera a fortnight before that time,
and that, as she had no children, her wearing apparel had been sent to
Monkton by a common carrier. The clothes had not been washed;
Barnes had opened the box in the evening; on the next day he had
fallen sick of the disease.
“During the illness of Mrs. Barnes, her mother, who was living at
Tockwith, a healthy village five miles distant from Moor Monkton, was
requested to attend her. She went to Monkton accordingly, remained
with her daughter for two days, washed hey daughter’s linen, and set
out on her return home, apparently in good health. Whilst in the act of
walking home she was seized with the malady, and fell down in collapse
on the road. She was conveyed home to her cottage, and placed by the
side of her bedridden husband. He, and also the daughter who resided
with them, took the malady. All the three died within two days. Only
one other case occurred in the village of Tockwith, and it was not a fatal
case.” (p. 136.)
“A man came from Hull (where cholera was prevailing), by trade a
painter; his name and age are unknown. He lodged at the house of
Samuel Wride, at Pocklington; was attacked on his arrival on the 8th of
September, and died on the 9th. Samuel Wride himself was attacked on
the 11th of September, and died shortly afterwards. These comprise the
first cases.
“The next was that of a person named Kneeshaw, who had been at
Wride’s house. But as this forms one of a series connected with the
former, furnished by Dr. Laycock, who has very obligingly taken the
trouble to verify the dates and facts of the latter part of the series, it will
be best to give the notes of these cases in that gentleman’s own words.
“‘My dear Dr. Simpson,—Mrs. Kneeshaw was attacked with cholera
on Monday, September 9th, and her son William on the 10th. He died
on Saturday the 15th; she lived three weeks; they lived at Pocklington.
On Sunday, September 16th, Mr. and Mrs. Flint, and Mr. and Mrs.
Giles Kneeshaw, and two children, went to Pocklington to see Mrs.
Kneeshaw. Mrs. Flint was her daughter. They all returned the same
day, except Mr. M. G. Kneeshaw, who stayed at Pocklington, until
Monday, September 24th, when he returned to York. At three o’clock
on the same day, he was attacked with cholera, and died Tuesday,
September 25th, at three o’clock in the morning. [There had been no
cholera in York for some time.] On Thursday, September 27th, Mrs.
Flint was attacked, but recovered. On Saturday, September 29th, her
sister, Mrs. Stead, came from Pocklington to York, to attend upon her;
was attacked on Monday, October the 1st, and died October the 6th.
“‘Mrs. Hardcastle, of No. 10, Lord Mayor’s Walk, York, was attacked
with cholera on October 3rd, and died the same day. Miss Agar,
residing with her, died of cholera on October 7th. Miss Robinson, who
had come from Hull to take care of the house, after the death of Mrs.
Hardcastle and Miss Agar, was attacked, and died on October 11th. Mr.
C. Agar, of Stonegate, York, went to see Mrs. Hardcastle on October
3rd, was attacked next day, and died October 6th, early in the morning.
On Monday, October 8th, Mrs. Agar, the mother of Mr. C. Agar, was
attacked, and on the same day, one of the servants; both recovered.
They had lived with Mr. Agar. All the above dates and facts I have
verified.
“‘I am, dear Dr. Simpson, yours very truly,

“‘T. Laycock.
“‘Lendal, December 1st, 1849.’” (p. 160.)
Several other instances of the communication of cholera, quite as

striking as the above, are related in Dr. Simpson’s work.
The following account of the propagation of cholera has been
published, along with several other histories of the same kind, in a
pamphlet by Dr. Bryson.[2]
“Mr. Greene, of Fraserburgh, gives the following account of the
introduction of cholera into two villages in Scotland. Two boats, one
belonging to Cairnbulgh and the other to Inveralochy, met at Montrose,
and their crews on several occasions strolled through the town in
company, although aware that it was at that time infected with cholera.
On their passage homeward, they were obliged to put into Gourdon,
where one man belonging to the Cairnbulgh boat died on the 22nd of
September, after an illness of fourteen hours, with all the symptoms of
cholera. Several of the men of both boats were at the same time
attacked with serous diarrhœa, of which three of them had not
recovered when they reached their respective homes; nor indeed until
the first cases of the epidemic broke out in the villages.
“In Inveralochy the first case appeared on the 28th of September,
three or four days after the arrival of the boat; the sufferer, the father of
one of the crew, had been engaged in removing the cargo along with
other members of his family. Two other cases occurred in this family;
one on the 30th of September, and one on the 1st of October.
“In Cairnbulgh, the first cases appeared on the 29th and 30th of
September respectively, and both patients had also been engaged in
removing the cargo of the boat (shell-fish) belonging to that village. No
other cases appeared until the 3rd of October; so that from the 28th of
September to the 3rd of October none were attacked in either village,
but those who had come in contact with the suspected boats, or their
crews.
“The subsequent cases were chiefly among relatives of those first
attacked; and the order of their propagation was as follows. In
Inveralochy, the first case was the father of a family; the second, his
wife; the third, a daughter living with her parents; the fourth, a
daughter who was married and lived in a different house, but who
attended her father and mother during their illness; the fifth, the
husband of the latter; and the sixth, his mother. Other cases occurred
at the same time, although they were not known to have communicated
with the former. One of them was the father of a family; the second his
son, who was seized the day after his father, and a daughter the next
day.”
The following instances of communication of cholera are taken from
amongst many others in the “Report on Epidemic Cholera to the Royal
College of Physicians”, by Dr. Baly.
“Stockport. (Dr. Rayner and Mr. J. Rayner, reporters). Sarah Dixon
went to Liverpool, September 1st, to bury her sister, who had died of
cholera there; returned to Stockport on September 3rd; was attacked
with cholera on the 4th; was taken home by her mother to her mother’s
house, a quarter of a mile distant; was in collapse, but recovered. Her
mother was attacked on the 11th, and died. The brother, James Dixon,
came from High Water to see his mother, and was attacked on the 14th.
“Liverpool. (Mr. Henry Taylor, reporter.) A nurse attended a patient
in Great Howard Street (at the lower part of the town), and on her
return home, near Everton (the higher part of the town), was seized,
and died. The nurse who attended her was also seized, and died. No
other case had occurred previously in that neighbourhood, and none
followed for about a fortnight.
“Hedon. (Dr. Sandwith, reporter.) Mrs. N. went from Paul, a village
close to the Humber, to Hedon, two miles off, to nurse her brother in
cholera; the next day, after his death, went to nurse Mrs. B., also at
Hedon; within two days was attacked herself; was removed to a
lodging-house; the son of the lodging-house keeper was attacked the
next day, and died. Mrs. N.’s son removed her back to Paul; was
himself attacked two days afterwards, and died.”
It would be easy, by going through the medical journals and works
which have been published on cholera, to quote as many cases similar
to the above as would fill a large volume. But the above instances are
quite sufficient to show that cholera can be communicated from the
sick to the healthy; for it is quite impossible that even a tenth part of
these cases of consecutive illness could have followed each other by
mere coincidence, without being connected as cause and effect.
Besides the facts above mentioned, which prove that cholera is
communicated from person to person, there are others which show,
first, that being present in the same room with a patient, and attending
on him, do not necessarily expose a person to the morbid poison; and,
secondly, that it is not always requisite that a person should be very
near a cholera patient in order to take the disease, as the morbid matter
producing it may be transmitted to a distance. It used to be generally
assumed, that if cholera were a catching or communicable disease, it
must spread by effluvia given off from the patient into the surrounding
air, and inhaled by others into the lungs. This assumption led to very
conflicting opinions respecting the disease. A little reflection shews,
however, that we have no right thus to limit the way in which a disease
may be propagated, for the communicable diseases of which we have a
correct knowledge spread in very different manners. The itch, and
certain other diseases of the skin, are propagated in one way; syphilis,
in another way; and intestinal worms in a third way, quite distinct from
either of the others.
A consideration of the pathology of cholera is capable of indicating to
us the manner in which the disease is communicated. If it were ushered
in by fever, or any other general constitutional disorder, then we
should be furnished with no clue to the way in which the morbid poison
enters the system; whether, for instance, by the alimentary canal, by
the lungs, or in some other manner, but should be left to determine this
point by circumstances unconnected with the pathology of the disease.
But from all that I have been able to learn of cholera, both from my
own observations and the descriptions of others, I conclude that
cholera invariably commences with the affection of the alimentary
canal. The disease often proceeds with so little feeling of general illness,
that the patient does not consider himself in danger, or even apply for
advice, till the malady is far advanced. In a few cases, indeed, there are

Full Download The Philosophy of Quantitative Methods: Understanding Statistics Brian D. Haig PDF

Uploaded by

Copyright:

Available Formats

Full Download The Philosophy of Quantitative Methods: Understanding Statistics Brian D. Haig PDF

Uploaded by

Document Information

Original Title

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

Full Download The Philosophy of Quantitative Methods: Understanding Statistics Brian D. Haig PDF

Uploaded by

Copyright:

Available Formats

Full download test bank at ebook ebookmass.

The Philosophy of Quantitative

CLICK LINK TO DOWLOAD

The Code of Putinism Brian D. Taylor

Quantitative Methods for Business 13th Edition

An Introduction to the Philosophy of Religion 4th

Quantitative Methods for Business, 5th Edition Donald

All Talked Out: Naturalism and the Future of Philosophy

Quantitative Techniques in Management 5th Edition Dr.

Understanding Basic Statistics 8th Edition Charles

(eBook PDF) Statistics for Political Analysis:

Understanding Statistics Video as Method

Understanding Qualitative Fundamentals of Qualitative Research

Published in the United States of America by Oxford University Press

© Oxford University Press 2018

All rights reserved. No part of this publication may be reproduced, stored in

You must not circulate this work in any other form

CIP data is on file at the Library of Congress

The Philosophy of Quantitative Methods . . . . . . . 2

CHAPTER 3 Tests of Statistical Significance . . . . . . . . . . . . . . . . 41

CHAPTER 6 Exploratory Factor Analysis . . . . . . . . . . . . . . . . . 117

Chief Lessons Learned . . . . . . . . . . . . . 143

In preparing this book, I have made use of previously published material

the ph i l osophy of research methods is an area of knowl-

2 : The Philosophy of Quantitative Methods

The Philosophy of Quantitative Methods

Qualitative methodologists also often bolster their preferred

Scientific Realism and Its Methodology

4 : The Philosophy of Quantitative Methods

philosophy of moderate realism to underwrite his novel methodol-

widely used hypothetico-​deductive method, with its emphasis on

6 : The Philosophy of Quantitative Methods

Theories of Scientific Method

a hypothesis or a theory and tests it indirectly by deriving from it

Book Overview and Chapter Summary

8 : The Philosophy of Quantitative Methods

relation to prominent accounts of scientific method where appro-

confirmation theory and assess its difficulties. I then compare it

A Note for the Reader

10 : The Philosophy of Quantitative Methods

place in the behavioral research method curriculum. The same can

Exploratory data analysis is an attitude, a flexibility, and a

14 : The Philosophy of Quantitative Methods

for detecting patterns, and it makes good sense to undertake such

Exploratory Data Analysis : 15

confirmatory methods. Because exploratory data analytic and

What Is Exploratory Data Analysis?

16 : The Philosophy of Quantitative Methods

For Tukey, data analysis is both an empirical science and an

Exploratory Data Analysis : 17

distributions and their resistance to outliers in sets of data.

Two Methods of Exploratory Data Analysis

The Box-​and-​Whisker Plot

18 : The Philosophy of Quantitative Methods

box plot provides an effective graphical display of the five-​number

The Four Rs of Exploratory Data Analysis

data = fit + residual

Hence, data residuals are what remain of the data after

residual = data − fit

widely used hypothetico-deductive method, with its emphasis on

The Box-and-Whisker Plot

box plot provides an effective graphical display of the five-number