An Introduction To Criminological Theory and The Problem of Causation PDFDrive

An Introduction to Criminological
Theory and the Problem of Causation

Jason Warr
An Introduction
to Criminological
Theory
and the Problem
of Causation
Jason Warr
University of Lincoln
Lincoln, United Kingdom
ISBN 978-3-319-47445-8 ISBN 978-3-319-47446-5 (eBook)

DOI 10.1007/978-3-319-47446-5
Library of Congress Control Number: 2016954267
© The Editor(s) (if applicable) and The Author(s) 2016

This book was advertised with a copyright holder in the name of the publisher in error, whereas
the author holds the copyright.
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher,
whether the whole or part of the material is concerned, specifically the rights of translation,
reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in
any other physical way, and transmission or information storage and retrieval, electronic
adaptation, computer software, or by similar or dissimilar methodology now known or here-
after developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are
exempt from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information
in this book are believed to be true and accurate at the date of publication. Neither the
publisher nor the authors or the editors give a warranty, express or implied, with respect to
the material contained herein or for any errors or omissions that may have been made.
Cover illustration: Pattern adapted from an Indian cotton print produced in the 19th century
Printed on acid-free paper
This Palgrave Macmillan imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
This book is dedicated to Ben, Shadd and Alison for inviting
me to the Cropwood Conference on the Effects of Imprisonment
in 2004 and to John Irwin, Barbara Owen, Craig Haney, Paul
Hirschfield, Pat Carlen and Joe Sim who I met there.
I would not have walked this path without the encouragement
and the kicks up the arse that you all gave me.
And, of course, Kate.
PREFACE
This book arose from my own confusion when first introduced (indoctri-
nated?) into the field of criminology. Moving from one discipline to
another always results in two fundamental issues: firstly, being confronted
by one’s own lacuna of knowledge – you quickly become starkly aware of
how ignorant you are of the new field of study. The second is that you
cannot help but compare the new subject, and its ways of doing things, to
the one that you have just left. The conjunction of these two epistemic
positions resulted in a Gordian knot of befuddlement. The issue?
Criminology has some problems!
Before coming to criminology, I had begun studying philosophy with the
Open University before then going on to study Philosophy, Logic and
Scientific Method at the London School of Economics and Political Science.
The one thing that really marks out philosophical discourse is the examination
and explication of assumptions which underpin the ideas with which we
wrestle. Some may see this as academic pedantry (I have some sympathy for
that) but what it does do is ensure that you make manifest the logics with
which you construct your argumentation as well as attempt to remove/
eliminate semantic ambiguity. I carried this mindset with me into the study
of criminological thought – where often I found myself trying to uncover, and
discuss, the assumptions with which criminology is laden. I must have being a
nightmare to teach. For instance, when studying rational choice theories of
crime, I wanted to know what was meant by rationality in any given theory – it
is not a straightforward notion and is an assumptively and value laden concept.
I also wanted to know what role akrasia (Aristotle’s theory of weakness of will)
had in terms of rational choice theories? As I noted – a nightmare to teach.
vii
viii PREFACE
However, the questions and confusion remained – often the more I read
the more these issues became entrenched. Nevertheless, I was (being a
geeky academic) determined to untangle my befuddlement. One of the
fundamental issues I had, which I needed to untangle, has resulted in this
book. It seemed evident that there had been a major shift in modes of
thinking in terms of causes of crime, yet there seemed to be little literature
that expounded an adequate reason for this shift. That theory had changed
was a given, and the theories themselves were laid out in various texts and
to varying degrees, but the process of that change, the mechanics if you
will, had not really been explored. Upon investigation, I became aware
that the logics of causation were not really discussed in any systematic way
in even the most advanced theories on the causation of crime. As with the
concept of rationality, the assumptions underpinning causation were not
being made explicit; as such the problems inherent to causal theorising
were not being adequately addressed in criminology.
Here, then, are the central two themes of this book: what are the
philosophical problems of causation and how have these issues impacted
on the course and history of criminological thought. The book itself is
designed to be an introductory text to these issues. Causal reasoning and
the problems associated with it are incredibly complex. This book does not
presuppose any knowledge of logic nor the philosophy of science and is
thus aimed at both students of criminology and criminological theorists.
The argument contained herein is set out in a simple linear format exam-
ining why a poor understanding of causation is a serious issue for crimin-
ology; what the main causal conception within criminology has been; what
the problems are with this conception; how an implicit rejection of this
conception has shaped the history of criminology; and the proposal of a
form of causal reasoning that would suit contemporary theories of crime
causation and which solves some of the identified problems.
How to use this book. The exploration of these themes is designed to help
the reader think about the complexity and difficulty presented by causal
reasoning when considering theories of crime, designing their own
research and writing up method chapters or even analysing data and
constructing a theory of crime. The book consolidates the problems of
causation into a small, reader friendly format that will also be useful for
those working in the field of policy development in terms of how research
and theory can be assessed as well as for those who may have to give expert
testimony on crime causes. Whilst it may be possible to dip into individual
chapters to glean information as you need it, to use the text as a reference
PREFACE ix
guide if you will, I would recommend reading the argument through fully
first. Also, the book is to be enjoyed – and to this end, I have made this
examination of a crusty old philosophical problem accessible (hopefully)
to all.
CONTENTS
1 Introduction 1
2 Criminology and the Problem of Causation 7
3 Humean Causation and Crime Theory 13
4 Deviant Causal Chains, Refutation and Other Problems 21
5 Humean Causation and the History of Criminology 29
6 Paradigm Drift and Criminological Theory 43
7 INUS Conditions and Criminological Theory 55
8 Consequences 71
9 Conclusion 77
Appendix A 81
References 85
Index 91
xi
CHAPTER 1
Introduction
Abstract This chapter introduces the book and explains that causation and
causal explanations are central to human understanding. The chapter also
begins to explain why the problem of causation has been such a thorn in the
side of both philosophy and the natural/social sciences. It notes that the
field of criminology essentially has three core aims: defining crime, explain-
ing how crime occurs and deciding what to do about crime. It explains the
aetiological crisis that has beset criminology and how we are failing our
own discipline. It also explains why this is an issue for criminology students
and criminological theorists and how they may use the book. It also
explains how and why each chapter is set out.
Keywords Aetiology Causation Criminology Theory
A spectre has haunted the history of theoretical criminology. As Wikström

(2006) notes, this ‘spectre’ has been ‘ . . . that criminological theory, by
and large, has not been able to fully address the problems of causation and
explanation’ (p. 61). Causality is an integral part of the social sciences
(Hollis 1994) and should be accorded a high priority because it stands as
the bedrock for the ‘understanding of social phenomena and the building
of an explanatory science’ (Marini and Singer 1988: 347). Criminology, as
a discourse has fundamentally a threefold aim, which is to: (1) define the
nature of crime; (2) uncover the root causes of crime; and (3) resolve what
© The Author(s) 2016 1

J. Warr, An Introduction to Criminological Theory and the Problem
of Causation, DOI 10.1007/978-3-319-47446-5_1
2 AN INTRODUCTION TO CRIMINOLOGICAL THEORY AND THE PROBLEM OF . . .
it is that we ought to do about the issue of crime. However, there is

something of an aetiological crisis in criminology as the discourse has
failed to provide reliable causal accounts (Young 1986). Weisburd and
Piquero (2008) further extend this point when they note in their study
that the explanatory power of even the most sophisticated criminological
theories, employing the most robust and advanced statistical models/
tests, put forward in the last century and a half is very low.
As a consequence, one of the core challenges of criminology, establish-
ing how and why crime is caused, has not been adequately met. We are, to
a degree, failing in our own discipline. This of course poses problems in an
industry where the double curses (obsessions?) of impact and research
excellence hover above us like an academic sword of Damocles. It is my
contention that we need to address this problem head on by opening up to
discussion, from novice undergraduate student through to seasoned crim-
inological theoretician, the issues and complexities of the problem of
causation.
If we are to open up this discussion more widely to the field, then we
must start somewhere. This is what this book is for – to begin the discus-
sion. There are, fundamentally, two problems associated with causation
and causal explanations within theoretical criminology. Firstly, as in the
wider context of the natural and social sciences, there is a poor under-
standing of causal mechanisms. Secondly, the implicit adoption of an
overly simplistic causal model leads, inevitably, to inadequate integration
of levels of explanation and the failure of causal explanations. This book is
an introduction to these issues and as such is concerned with exploring not
only the issues and problems raised by causation for theorising but also the
consequences of these issues and problems on the history, present and
future of criminological discourse.
So, who is this book for? Essentially, I have attempted to make what is a
horrendously complex set of issues accessible to all with some understand-
ing of the nature of social science. Therefore, this book is for both the
criminology undergraduate interested in the history and development of
their chosen subject, the research student who is beginning their project
and needs to consider these issues when drafting their methods chapter
and the seasoned practitioner, who is engaged in advanced aetiological
theorising who may need a (unfortunately not so) simple reference guide
on the problems of causation. This book, being an introduction, is by no
means exhaustive in its exploration of the complexities of causal reasoning
but is intended to give all readers an understanding of the problems, how
1 INTRODUCTION 3
this impacts on the field of criminology and what we need to think about
when constructing/deconstructing criminological theory.
Before beginning in earnest, it must be noted that this book is not
intended as a critique of particular criminological theories or of particular
theorists. All theories have problems and all theories have merit – whether
that be in explanatory power, utilisation or impact. However, such con-
siderations are beyond the scope of this particular book. Instead, what this
book is concerned with is the apparent shift in aetiological thinking that
has accompanied the progress of criminological thought. Here I am
interested in charting the apparent drift from one form of aetiological
thinking to another in the modern history of our field. I am not particu-
larly interested in what particular theorists have to say on crime and its
causation, rather how they construct their argumentation in terms of
explicating causal explanations of crime. There are no value judgements
placed on the works being analysed – all have played a part in the devel-
opment and progress of the field to which I am an adherent and which I
seek here to examine. As such, I am more interested in the mechanics of
their argumentation, their navigation of ontology and epistemology, their
dissatisfaction with method and their conclusions than the content, if you
will. This is a historical and technical overview of the field. In that regard,
though this book is firmly designed for criminologists and social scientists
it has its roots in the abstract aetiological discourses to be found in the
philosophy of the natural and social sciences. It is also why this book is not
an exhaustive examination of criminological theory in the period under
discussion but is instead an examination of representative forms of causal
explanations in criminological thinking in this period.
However, before explicating these arguments it is important to note
that the problem of causation is not restricted to criminology. Cause-
and-effect-type constructions in language, general conversation and
scientific/non-scientific practice are both pervasive and, in some regards,
structurally necessary for human endeavours (Gerring 2005). Hume
(1999, 2003), writing in the eighteenth century, notes that we all,
from childhood onwards, have an inherent concept of causation which
underpins both our learning and navigation of the world. Much of our
quotidian rituals and behaviours are predicated on this conception, from
preparing breakfast to getting to the office where I type up this manu-
script. We come to understand, through direct experience or teaching,
that one element or factor can have a direct causal impact on some other
element or factor: for instance, we learn that a heat source will make
a pan hot and that if we touch the rim of the pan with our bare hands, it
will cause us to experience both a burning pain and burning injury. This
is, seemingly, not that complicated.
Nevertheless, as Ehring (1997) highlights, when you attempt to analyse
and communicate to others the exact nature of that cause and effect
relationship and how it works we soon run into problems. Is it the trans-
ference of heat that causes the sensation and injury or the length of time
contact is maintained? Is it the length of time or the amount of respective
surface areas of hand and pan that are in contact? Is it the absence of
mitigating factors? Is it that you decided to put the pan on a heat source
or turned on the heat source? Is it that the gas/electric of the heat source is
working? Was it that your favourite song came on the radio and you got
distracted? Is it a conjunction of all these elements? If so, where do we draw
the boundaries or how do we exclude elements that are not causal? Are any
elements not causal? Is the fact that you were born at all relevant to the fact
that you got burnt whilst cooking? If so, how and in what way? Once we
begin to unpick the nature of the cause/effect relationship it becomes
exponentially complex. The reason for that is because causation is, rather
than being seemingly simplistic, brain-achingly complex (Clark 1998). The
author and philosopher Douglas Adams, in his book Dirk Gently’s Holistic
Detective Agency (1987), even noted that the complexities of causal rela-
tions ‘defy analysis’. Thankfully, this is not quite true.
Causality and its attendant problems have then been the subject of
philosophical and scientific discourse for centuries (Marini and Singer
1988). In fact, causatives and causal explanations are, to a degree, tied
inherently to the very construction of language and as such are a
significant source of research and debate not only in the field of linguis-
tics but in most empirical sciences and endeavours (Song 2013). If you
think about it modern technology, medicine, entertainment, engineer-
ing, everything that makes the modern world what it is would not exist
if someone had not made a causal link between two phenomena.
Therefore, causation, its construction, the manner in which it shapes
theoretical thinking and thus shapes theory development is a problem
that besieges the sciences as a whole (Davidson 1994). This is even
more true in human sciences where those elements, factors or phenom-
ena which may be causal are much more complex or fluid than the
rigidity of natural laws found in some other fields of study (Mele 2007).
Much of the philosophy of the natural, as well as the social, sciences
have thus been concerned with addressing this specific problem
1 INTRODUCTION 5
(Mackie 1974). What follows draws heavily upon this literature and
inter-contextualises it with regard to the history of criminological
thought.
The book itself is divided into three broad sections. The first section
will explore three central themes: the first of these is concerned with
explicating why a poor understanding of causal mechanisms is a problem
for criminological theorising. This will involve the discussion of how an
inadequate understanding of causal mechanisms may cause theories to fail
as causal explanations, how the adoption of a simplistic causal framework
renders a theory open to damaging criticism, how confusion arises
between causes and correlates and how all these can adversely affect the
manner in which criminologists approach specific explananda. The second
theme identifies the implicit causal mechanism employed in much of the
history of criminology as the Humean regularity/chain model (as based
upon the critique of human knowledge by David Hume (1999, 2003))
and shows why this is an inadequate model of causation. Lastly, a number
of examples, from the history of criminology, Classicism, Biological
Positivism, Strain theory and Labelling theory, will be explored highlight-
ing both the manner in which the Humean model has shaped these
theories and the impact upon the validity of those theories by adopting
this inadequate causal mechanism.
The second section will focus upon the apparent ‘paradigm drift’, as
exampled by such theories as Routine Activity/Rational Choice, Life
Course and other integrated explanations, which has occurred within the
theoretical discourse. I will argue that the reason for this ‘drift’ is an
inherent rejection of the simplistic and inadequate Humean chain mechan-
ism implicitly employed in previous criminological theorising. I will argue
that the manner in which these advanced criminological theories formulate
their argumentation and construct their causal explanations is a distinct
rejection of the problems that beset their forebears and which are identified
in this book. I will also argue that though this rejection has occurred, and
that explanations mentioned in this context have achieved a more profound
level of integration, and thus causal explanation, there has still neither been
any identification nor adoption of an explicit or adequate model of aetio-
logical explanation.
The third and final section is divided into two parts: the first is con-
cerned with identifying an adequate mechanism of causation. I will argue
that the suggested mechanism, an adaptation of Mackie’s (1965) INUS
model, is an adequate causal model for criminology theorising. I will show
that this particular model works by explicitly defining not only that which
can be counted as causal but also how it must fit into the structure of the
mechanism. Also, I will explicate not only how this model is more suited
to contemporary integrated forms of theorising but also how the INUS
conditions solve many of the problems, such as the confusion between
correlates and causes, that besets extant theorising. I will also explore some
of the criticisms and problems that this causal explanation may face.
Secondly, I will explore some of the potential consequences of explicitly
adopting the INUS model of causation. I will highlight the impact that
such an adoption would have upon the depth of integration, and the scope
of argumentation, needed in order to satisfy the internal strictures of the
model.
Lastly, it must be noted that this book does draw on some logical
symbols in order to highlight the nature of theory construction under
discussion. This can often put readers off (it often does me) and thus can
detract from the reader’s enjoyment of the text. All I can say here is that I
have simplified this as much as is possible and kept this symbolising to a
minimum and only where it is necessary. I have also explicated what the
symbols mean every time they are used. I am therefore hoping that this
does not get in the way and detract from the reader’s pleasure (a dubious
assumption for such a text, I know) of this book.
Let us begin . . .
CHAPTER 2
Criminology and the Problem of Causation
Abstract The purpose of this chapter is to establish the problem with

which the book is concerned. As such the focus of this chapter is to supply
answers to the following question: why is a poor understanding of causal
mechanisms, and causal explanations, a problem for theoretical criminol-
ogy? I then set out the four main reasons why a poor understanding of
causal argument is indeed a problem for criminology and how these issues
can impact on understanding, theory design, data analysis and the age-old
problem of distinguishing between causes and correlates.
Keywords Cause Correlation Criminological Problems
With such a complex topic it is not suitable to jump straight into the deep
end of the discussion. I agonised over where to start this book – should it
begin with the nature of causation? Or perhaps differing types and the
complexities of causal reasoning? Or maybe the history of causality in the
natural and social sciences? I have decided to not to do any of these as
these discussions could actually take us away from the specific problem of
causation in criminological theory. Therefore, the purpose of this chapter
is to establish the problem with which this book is concerned. I shall, as
any good philosopher/scientist does, begin with the Socratic method and
explicate the central question of why. As such the focus of this chapter is to

of Causation, DOI 10.1007/978-3-319-47446-5_2
supply answers to the following question: why is a poor understanding of

causal mechanisms, and causal explanations, a problem for theoretical
criminology?
There are fundamentally four main arguments why a poor understand-
ing of causal mechanisms, and causal explanations, is a problem for crim-
inology. These four arguments are: firstly, an inadequate, or folk attributive
(Hitchcock 2007), understanding of causal mechanisms can result in the
failure of posited causal explanations (Russell 1913); secondly, the adop-
tion of a simplistic causal framework renders a theory open to damaging
criticism; thirdly, confusion and/or conflation can arise between factors
that are causal and factors which may merely be correlates. This is especially
pertinent with the recent growth of the ‘death of theory’ thesis proposed by
some involved in big data research (Cowls and Schroeder 2015); and,
lastly, it can adversely affect the manner in which criminologists approach
specific explananda – the element that is in need of explanation, in this
instance: crime.
The first of the arguments listed previously is that an inadequate, or folk
attributive (the kind of simple, or common sense, ideations of cause/
effect), understanding of causal mechanisms can result in the failure of
posited causal explanations. Why would this be the case? The problem here
is that, unfortunately, as noted previously causation is not a simple matter.
However, within criminology, as with all social sciences, very little thought
is given by the majority of practitioners to the complexities involved in this
notion (Wikström 2006). Instead, common practice is usually dominated
or ‘guided by a natural or intuitive idea of causality’ that lead theorists to
attempt to discover ‘directional relationships that are not spuriously deter-
mined’ (Marini and Singer 1988: 348). Such observed (even scientifically
observed) relationships result in assertive notions of causation that rely on
distinctive (as opposed to merely background) factors, conditions or ele-
ments (Hitchcock 2007). This intuitive conception of causation in crim-
inological thinking is evident in the assumptions underpinning both
classical and positivist iterations of the discourse (Einstadter and Henry
2006).
However, the adoption of such a simplistic, or ‘folk attributive’, notion
of causation and causal relations can easily result in a situation where a
proposed theory of causation does not actually explain, account or capture
the reality of the causal relationships discussed or asserted (Blackburn
1990). This is primarily due to the inadequate integration of levels of
explanation that is possible with simplistic notions of causation. One
2 CRIMINOLOGY AND THE PROBLEM OF CAUSATION 9
example here is the (rightfully) much-maligned Broken Windows theory

(Wilson and Kelling 1982) which posits a notion of crime causation
predicated upon a reductio to social disorder. As an environment becomes
increasingly disordered, so too do those who inhabit the environment.
This societal and environmental atrophy inevitable leads to, firstly,
deviance and then criminality, which leads to further environmental dis-
order that then leads to further deviance and so on. The problem with
such a causal theory, forgetting for a moment that it is horrendously
circular, is that it cannot possibly account for other possible factors that
may account for the aetiology of crime, such as structural or individual
issues. The terms of the account are far too narrow. In such a situation the
theory will fail what I refer to here as the test of aetiological extensivity –
the extent to which the causal elements, mechanisms and relations of a
given set of phenomena are satisfactorily accounted for by a theory – and
will therefore, fail as a causal explanation (Woodward 2003).
A further, yet more subtle, problem is that a poor understanding of
causal mechanisms and systems of explanation can result in a situation
where the line of reasoning/argumentation adopted is not compatible
with the conclusions drawn. Hempel (1965) argues that there are funda-
mentally two types of scientific explanation: those explanations which are
based upon a deductive model and those which are based upon a probabil-
istic, or inductive, model (I shall discuss abductive models of explanation
later). Each type of explanation leads to its own specific, and exclusive, kind
of conclusion. The deductive model results in conclusions that are implicitly
universal or generalisable (I will return to this point later) whereas a prob-
abilistic, or inductive, model results in conclusions that are necessarily
couched in terms of statistical probability, or degrees of certainty
(Hempel 1965). One type of explanation cannot result in the others kind
of conclusion and to confuse the two can result in false or erroneous
conclusions (Salmon 1984). A poor understanding of causal explanations
can result in just this kind of confusion. Most commonly where a theory/
explanation is constructed through the deductive/universal model yet
entails an inductive/probabilistic conclusion.
The second argument is that the adoption of a simplistic causal frame-
work renders a theory open to damaging criticism or even logical refuta-
tion. There are two points to be made with regard to this. The first is that
logical refutation is a particular problem, especially in light of the fact that
many causal theories inherently result in conclusions that take the form of
causal generalisations or law-like statements. As is the case with any
generalisation, they can be open to logical refutation through the means of

a single counterexample (Wittgenstein 1921). These particular arguments
shall be discussed in more technical depth in the later chapter on deviant
causal chains as the Duhem-Quine thesis adds complexity to this refuta-
tional notion (Harding 1976). However, if the adopted framework results
in a causal generalisation such that criminality, for instance, is to be
explained by a given set of conjoined antecedent conditions, then a single
instance where the conditions are present but criminality does not occur
would be sufficient to refute the generalisation and thus pose significant
problems for the theory. In essence, the theory fails to do what it says on
the tin.
The second point, and one that is related to the next argument, is that
even if the conclusions seem robust and are not open to direct refutation
through counterexample, if there is the reliance, due to the adoption of
this particular framework, upon a conjoined set of antecedent conditions
as the causal factor in explaining criminality, or some other phenomena,
then it is difficult, if not impossible, to establish which, if not all or indeed
any, of the antecedent conditions are actually/partially causal or not. This
can once again cause a theory to fail the test of aetiological extensivity and
lead to the criticism of the theory as an adequate explanatory tool.
The third argument is concerned with the perennial problem of the
confusion that arises between those phenomena, events or conditions
which are the actual cause of some other phenomenon, event or condition
and those that are merely, or even coincidentally, correlated with it.
Interestingly, Sowel (1995) notes in passing that when learning the subject
of statistics one of the first things that people are taught is that correlation
does not equate to causation – yet this is also one of the first things that is
forgotten by researchers, scientists and ‘experts’. His argument, largely focus-
ing on economics and politics, is cynically tongue-in-cheek but yet pertinent.
He argues that this simple fact is often either subsumed under research
interests or sacrificed on the altar of expedience – it benefits those with
structural power to make causal claims predicated on (perhaps spurious)
correlations if this suits their policy making. As noted, this is a rather cynical
account of public science but the basic truth still holds – correlation does not
equal causation yet is often presented as doing so. In terms of the argumenta-
tion being made here, once again, there are two issues relevant to this
problem. The first is that this confusion most readily occurs when the adop-
tion of a simplistic causal framework or mechanism leads to the reasoning
error traditionally known as ‘post hoc ergo propter hoc’ (Warburton 2000). This
2 CRIMINOLOGY AND THE PROBLEM OF CAUSATION 11
reasoning error is when it is deduced that because one event comes after, and
is regularly observed to come after, some other particular event then the
second event must occur because of the preceding one. However, without a
robust explanation of the mechanics of this relationship and a detailed
examination of the processes being inferred for such a move the assumption
that the second event was caused by the first cannot be justified (Honderich
1995). The problem is that simplistic causal frameworks do not allow for such
a robust aetiological argument in defence of this move.
The second issue is that just because two, or more, sorts of events are
correlated, that is, whenever one is found the other is usually also found,
this does not mean that there is necessarily a causal connection between
them. The classical philosophical example to highlight this point is with
regard to the high degree of correlation that exists between shoe size and
vocabulary (Warburton 2000). It is true to say that people with larger shoe
sizes tend to have much larger vocabularies than those with smaller shoe
sizes. However, shoe size does not, indeed cannot, cause any person to
have any vocabulary size. It is just that adults tend to have bigger feet than
children and coincidently have larger vocabularies. Another example from
the field of criminology is with regard to criminality, age and gender
(Steffensmeier and Allan 2000). Though it is accepted that there is a high
degree of correlation between being a teenage male and criminality, it
cannot be argued that being male and a teenager is a cause of criminality;
or that being female and not a teenager is a cause of abstention from
criminality. Gender and age are just not causal elements or factors in this
regard. The problem is that a simplistic causal framework, or mechanism,
does not allow for the distinction between what is a causal element and one
that is merely a correlate.
However, an added complication here is that, as Cowls and Schroeder
(2015) note, there is a growing but problematic contention in big data
research that developmental and causal theory is superfluous and that the
identification of correlational factors is all that is necessary for scientific
understanding. Initially outlined in Anderson’s (2008) paper, the notion is
posited that in an age of mass, or petabyte, data the discovery of correla-
tional relationships have such significant explanatory power that these
supersede quotidian understandings of causation and explanation. The
conclusion drawn from this logic is that causal theories are no longer
relevant. However, even if the explanatory power of such analyses is accu-
rate there are a number of problems – not all areas of scientific investigation
can generate petabyte-type datasets. Thus, many theorists must depend on
traditional models of scientific explanation. Also, if developing treatments

or interventions predicated upon the research then it is often necessary that
distinctions are made between causal factors and correlates (Humphries
2014). Not doing so could have rather nasty consequences – imagine
predicating an invasive pharmaceutical-based sex offender treatment pro-
gramme on research where the offending cause was deemed irrelevant.
The final argument is really concerned with the adverse impact that the
previously mentioned three arguments can have on the manner in which an
unwary researcher will approach their specific explananda. The first point to
be made is that there are different types of causation (intrinsic, extrinsic, self
and teleological determination, etc.) and not all causal frameworks can
account for all types of causation (Woodward 2003). Also, the different
types of causation can require different approaches to both empirical
research and analyses (Marini and Singer 1988). Without this recognition
a researcher could run the risk of failing to adopt an appropriate causal
framework leading to a misdiagnosis of causation and an inappropriate
form of analysis (Gerring 2005). Also, without an appropriate understand-
ing of causal mechanisms there is no way, as we have seen, that when
analysing data, the distinction between causes and correlates can be
made. As such there is the danger of making the ‘post hoc ergo propter hoc’
reasoning error and drawing causal generalisations that are open to coun-
terexample and refutation. All these dangers can await the unwary and
uninformed researcher and result in a final causal theory of crime that
fails all tests of aetiological extensivity and is, therefore, inadequate for
purpose.
CHAPTER 3
Humean Causation and Crime Theory
Abstract This chapter explores the relevance of Hume’s analysis and cri-
tique of causal explanation to the field of criminological theory. This for-
mulation of causal reasoning, based on an ‘if x then y’ construction, is
known as the Humean regularity, or chain, model of causation. It is argued
here that this form of reasoning has been implicitly employed throughout
the history of the social and criminological sciences. Utilising, and introdu-
cing the reader to, an adapted form of symbolic logic this chapter explores,
in some depth, the technical and formal structure of the Humean model of
causation. Finally, by employing a somewhat simplified example of an
imagined Control theory, I show how this model works in terms of the
structuring and creation of criminological theory.
Keywords Causation Control Hume Logic Regularity Theory
This chapter is concerned with identifying the causal mechanism, or

framework, that has been implicitly employed throughout the history of
criminological discourse. It shall be argued that the particular framework
that has dominated criminological theorising, as it has in the wider context
of the social sciences, is based upon Hume’s analysis of human knowl-
edge/understanding and is therefore known as the Humean regularity, or
chain, model of causation (Beauchamp and Rosenberg 1981). In order for
this argument to be made it is necessary to explore, in some depth, the
technical and formal structure of the Humean model of causation. Once

of Causation, DOI 10.1007/978-3-319-47446-5_3
this has been done it shall be shown, by means of a somewhat simplified

example of an imagined Control theory, how this model works in terms of
criminological theorising.
However, before starting to explore this model of causation it is first
necessary to say something about the formal language of logic that shall be
employed in this, and in other, chapters. This particular form of representation
has been chosen for demonstrative purposes. Though it will be necessary to
employ the formal symbols of logic in order to actually model the causal
frameworks discussed within these chapters, this has been kept to an absolute
simplistic minimum. No doubt, philosophical logicians will read this and
scream at a perceived injustice to which I subject their formal modulations.
For this I can but only offer my humble apologies. Nevertheless, as was noted
previously this is an introductory text for students and practitioners of crimin-
ology (and related/cognisant fields) and thus familiarity with first, second or
even higher orders of logic cannot be assumed and thus are avoided. What
symbols and logical schema that may be employed are as follows:
• The conjunction: (p ^ q) which reads: both p and q

• The disjunction: (p v q) which reads: either p or q
• The material conditional: (p → q) which reads: if p then q
• The bi-conditional: (p → q) which reads: p if and only if q
• The negation: ¬p which reads: not p
• The universal quantifier ∀x which reads: that for every instance
of x such and such is the case.
• The existential quantifier ∃x which indicates: at least one instance of
x where such and such is the case1
It was stated previously that this chapter is to explore the causal frame-
work that has been implicitly adopted in the history of criminology. This
is an important point to note, the causal framework that runs through
most theorising has not been chosen explicitly by those theorists who are
employing it. Marini and Singer (1988) continually make the point that
the practice that has dominated causal theorising throughout the social
sciences, including criminology (see Einstadter and Henry 2006), has
been this natural or intuitive notion of causation, that is simplistic and,
seemingly, fit for purpose. It is argued that the notion of causality that
has dominated the social sciences is based upon the structure of x causes y
if it is the case that y will not occur unless x does (Marini and Singer
3 HUMEAN CAUSATION AND CRIME THEORY 15
1988; Beauchamp 1974). It is for this reason that the Humean model of
causation has been identified as that which implicitly underlies the causal
theorising within the social sciences. This is because the Humean model
is based upon Hume’s exploration, and criticism, of man’s intuitive and
natural propensity to assign causal relations to the world around him in
the form of conditional statements (Beauchamp 1974, 1999).
What then is this Humean model of causation? Hume (1999, 2003)
argued that there are three axioms, if you will, to causation and causal
inference. These three axioms are as follows:
1. Contiguity in time – the idea that in order for a causal relationship to exist
between any two phenomena/events they must be connected in time.
2. Priority in time – that where one phenomena/event is seen as a
‘cause’ and a second is seen as an ‘effect’ the ‘causal’ phenomena/
event must have priority in time.
3. Constant conjunction – that whenever one event occurs it always
produces an event that appears to be the effect and that, in alike
circumstances, the same effect will always follow (Hume 2003).
It is this that causes Hume’s account to be sometimes referred to as his

Regularity theory.2 Hume famously used the example of one billiard
ball striking another to illustrate the manner in which causation is often
perceived, and how the three axioms mentioned previously can easily be
deduced from the observation of physical phenomena.
In essence Hume’s argument is as follows: billiard Ball A moves
towards a second billiard ball, Ball B, and at a certain point Ball A strikes
Ball B, from that point on Ball B moves. Hume argues that as there is no
interval between the shock (Ball A striking Ball B) and the motion (Ball B
moving), this shows that there must be contiguity in time between the
two events. Also, that the motion of Ball A and the striking of that ball
with Ball B (the causal affect) were prior to the motion of Ball B (the
caused effect) then this is indicative that the cause in a cause/effect event
must have priority in time to the effect. Finally, that whenever one event
occurs, Ball A striking Ball B for instance, it always produces an event
which appears to be the effect, Ball B moving, and that in alike instances,
Ball X striking Ball Y, the same effect follows – this is his notion of
constant conjunction (Hume 1999, 2003).
Hume adds a further notion, that of inductive expectation, to these issues
of contiguity in time, priority in time and constant conjunction. So that
whenever a particular event occurs, such as Ball A striking Ball B, if we have

experience of a previous alike event then we will expect or anticipate, the
effect, Ball B moving. It is this inductive expectation, or so Hume argues,
that instils in us the propensity to impose causal relations upon the events
that we perceive in the world around us, and draw causal generalisations
from our investigations.
This process of causation as described by Hume can be re-expressed in
terms of antecedent conditions and consequential phenomena. For a
causal relationship to be perceived there must be at least one phenom-
ena/event/condition that is antecedent to a second phenomenon/event/
condition which is the consequent. This relationship can be formalised by
way of the material conditional where if some x condition occurs then
another condition y must necessarily occur. Thus, when this conditional
statement is formalised it takes the form:
p!q
Other than Regularity theory, this notion of causality is often referred

to as the ‘chain model’ or ‘Humean chain ontology’ (Beauchamp and
Rosenberg 1981). The reason for this is that in any cause/effect event the
two conditions, or ‘atoms’, from which the event is constructed, must
necessarily be connected (though this is a contested notion - see Schaffer
(2000)) by that relationship. If they are not necessarily connected in this
manner then the two atoms may just be correlated (I will discuss this
further), as with the shoe size/vocabulary example from earlier.
Therefore, in this instance we have at least two conditions that act as the
atoms within a cause/effect event. However, as Strawson (2014) notes
when it comes to causation it is recognised that a cause/effect event rarely
involves just the dyadic atoms thus far described. Instead there is usually a
series, or a set, of antecedent events/conditions that are involved in the
production of some consequent. As such, and because of the first two
axiomatic strictures, contiguity and priority in time, imposed by this
model, these antecedent conditions must necessarily be conjoined regres-
sively. Thus when expressed formally the causal statement will look
like this:
ðp^q^rÞ ! c
Now that there is some indication of how Humean causation is perceived

it is time to illustrate how this model works in practice. Let us take a very
simplistic version of Hirschi’s Control/Bond theory (Hirschi 1969). This
theory presupposes that the motivations and propensity for criminality and
antisocial behaviour are widespread throughout contemporary Western
societies. That being the case the question that naturally follows is: why
do the majority of people not commit crimes, why is criminality confined
to a relatively few individuals?3 The answer that control theorists reach is
that there are social bonds, such as good and healthy family relations,
investment in one’s community and stable employment that exist which
lead people to either desist or abstain from criminal behaviour. Therefore,
it is the absence of these ‘bonds’ which causes criminality and antisocial
behaviour.
In order to explore how Humean causation is implicit to this theory
it is necessary to formalise and model the argument in the language of
logic that was outlined earlier. As was highlighted, Humean causation
can be represented in terms of the material conditional; and that where
there is more than one causal element the antecedent condition
becomes a set, or chain, of conjoined elements. If in this instance
the social ‘bonds’ and criminality are represented by the following
variables:
Good and healthy family relations = P

Personal investment in community = Q
Stable employment = R
Criminality = C
The domain is people = x
Then there are two potential consequences, or outcomes, implicit to this

argument. As such there are two models for this argument. The first will
look like this:
Argument 1: ðPx^Qx^RxÞ ! : Cx
Where the argument will read: if a person has good and healthy family
relations and has personal investment in the community and has stable
employment then that person will not be a criminal. The second way in
which this argument can be modelled is as follows:
Argument 2: ð: Px^ : Qx^ :RxÞ ! Cx
Where the argument will read: if a person does not have good and healthy
family relations and they don’t have an investment in the community and
they are not in stable employment then they will be a criminal.
Of course, a further point here is that as Marini and Singer (1988) note
in the social sciences the structuring of causal reasoning adopted attempts
to strengthen this Regularity theorem by introducing a degree of bi-
conditionality to the schema such that it becomes: x causes y if it is the
case that y will not occur unless x does. In terms of formalisation this
would look like:
ðX ! YÞ $ ðY $ XÞ
Here, the ‘unless’ becomes expressed as a double bi-conditional (Suppes

1999) and as such the overall schema reads: if x then y if and only if it is the
case that y if and only if x. In terms of the Control theory already discussed
this would equate to saying that: if a person has good and healthy family
relations and has personal investment in the community and has stable
employment then that person will not be a criminal if and only if it is the case
that they will not be a criminal if they have good and healthy family relations
and have investment in the community and have stable employment.
You may notice that, other than being very long winded and complex, this
is a slightly odd formulation of the expression. The reason for the oddness is
that in some senses it involves a circularity when translated from dialectical
discussion into formal systems. The consequent in this bi-conditional is really
just a restated form of the antecedent. This is not the arena in which to
discuss Sextus Empiricus’ (2000) contention that all deductions involve a
necessary and vicious circularity as the conclusion entails nothing outside of
the premises, though I will discuss this issue to some degree in later chapters.
It is, nevertheless, sufficient to note that this peculiarity does arise and seem
to incorporate a degree of question begging – the notion that the premises
presuppose the conclusion (Jacquette 1993). However, in this sense (in
language rather than formal logic) it does not equate to a logical fallacy but
rather a rhetorical quirk of reinforcement (Tennant 2005). To some degree
that can be seen to be Hume’s inductive expectation. Therefore, the critique
of the logical sceptic has little pertinence for this account, nor is the notion of
question begging in its purest form a damaging one for the contentions
being made. Nevertheless, to avoid over complication of the forthcoming
schema and a need to constantly write out the seeming question begging,
due to the formalisation, the bi-conditionality of the arguments from here on
in will be taken as a given rather than expressed.
Though this has been a somewhat potted and simplified exploration of
the base principles of Control theory, obviously Hirschi’s theory is more
complex than this, it has nevertheless suited in providing an example of the
manner in which Humean causation is implicit to the reasoning employed
by those constructing this kind of theory. However, as was previously
stated there are problems with this kind of reasoning.
NOTES
1. For further details, see Appendix A. For an accessible and complete account
of logic see Howson (1997)
2. There are, however, varying interpretations of Hume’s account and not all
of them see it as a regularity account. See Hume and Thick Connexions, by
Simon Blackburn, references.
3. Of course, this in and of itself is a huge assumption in some criminological
theorising. Contraventions of law are widespread and not limited to only a
specific subset of the population. It is unlikely that there is anyone in the
general population who has not infringed upon the law at some point in
their life.
CHAPTER 4
Deviant Causal Chains, Refutation

and Other Problems
Abstract This chapter examines the core problems associated with the
Humean model of causation and highlights why this is an inadequate
form of causal reasoning for criminology and the social sciences. The
chapter discusses two central themes: firstly, the Humean model’s relation-
ship to causal generalisations and its subsequent vulnerability to open and
direct refutation through the problem of deviant causal chains – where a set
of antecedent conditions do not produce the expected consequent. This
section also involves a discussion of abduction – inferring to the best
explanation and explains why, though intended, this is not what occurs in
the social sciences. Secondly, the chapter will explore the Humean model of
causation and how it relates to the infamously tangled causation/correlation
problem. I explain how the very nature of the Humean construction pre-
vents the discernment of any condition that may be causal from those which
are merely correlational. Finally, this chapter will explore and explain just
why this formulation of causal reasoning is not a suitable mechanism for
understanding the causes of crime.
Keywords Causal Complexity Deduction Humean Refutation

Theorising
This chapter is concerned with exploring why it is that the Humean model
of causation, that which has dominated criminological theorising, is an
inadequate form of causal reasoning. As such there shall be two forms

of Causation, DOI 10.1007/978-3-319-47446-5_4
of argument presented. The first shall explore the Humean model’s vul-
nerability to open and direct refutation through the problem of deviant
causal chains, especially with regard to causal generalisations. The second
argument shall focus upon the causation/correlation problem as it relates
specifically to the Humean model of explanation.
The problem of deviant causal chains is perhaps the main problem for
Humean regularity models of causal explanation. This is because it is this
problem that exposes the weaknesses at the heart of the conception. There
are two ways in which deviant causal chains can arise with each resulting in
a different consequence. The first, from here on in referred to as Type 1, is
where all the antecedent conditions are present, and conforming to the
axiomatic strictures of the model, and are therefore true, but yet the
expected consequent does not follow rendering the conditional statement
false. The second, Type 2, is where the antecedent conditions are not
present but yet the consequent still arises. As the causal chains of the
Humean model are constructed formally in terms of the conjunctive and
conditional declarative schemas they are subject to the rules of truth
functional logic. As such the consequences of deviant chains of causation
are governed by the truth functional tables expounded by Wittgenstein in
his seminal Tractatus Logico-Philosophicus (see Appendix A).
In order to explore deviant chain Type 1 we shall return to the simpli-
fied Control theory example used in the previous chapter, which will allow
us to see more clearly how this deviant causal chain can arise. Firstly, it
could be the case that there are good and healthy family relations and
there is personal investment in the community and there is stable employ-
ment but yet criminality still occurs:
Argument 1:2: ðPx^Qx^RxÞ ! Cx
Alternatively, it could be the case that there is an absence of good and

healthy family relations and there is no investment in the community and
there is no stable employment but yet criminality does not arise:
Argument 2:2: ð: Px^: Qx^: RxÞ ! :Cx
Both of these examples are instances where the antecedents are present
and correct and yet, in neither case, does the expected consequent occur.
In terms of truth functionality this means that the antecedents are true but
yet the consequent is false. As the Wittgensteinian truth tables1 indicate,
4 DEVIANT CAUSAL CHAINS, REFUTATION AND OTHER PROBLEMS 23
this is the only instance (T → F) where the material conditional is false and
invalid. Therefore, both of these two arguments are instances of the
counterexample move and can be thought of as direct refutations of the
theory in question. They both render the theory logically invalid. Neither
example is an outlandish one; both are quite within the imaginable realms
of possibility and could be open to empirical verification.
The Type 2 deviant causal chain can arise when the antecedent condi-
tions are false or not present but yet the consequent still arises (F → T). In
terms of truth functionality this means that the antecedent conditions are
false but yet the consequent is true – this does not render the conditional
statement logically invalid but does mean, in corporeal terms, that one’s
reasoning has gone awry. In terms of the Control theory this very much
looks like the second iteration of the Type 1 problem discussed previously
but is logically very different – remember the only time that the condi-
tional schema is entirely false is if the antecedent conditions are true whilst
the consequent is false and this is not what occurs in the Type 2 problem.
Though this is a deviant causal chain it does not necessarily mean that a
refutation has occurred though it is still damaging for the proposed
theory. Instead of a direct refutation it is evidence that either the ante-
cedent conditions have been completely misidentified as having a causal
relation to the consequent or that the wrong inferential schema has been
employed which results in the failure of the theory to capture the causal
relation between antecedent conditions and consequent. This latter pos-
sibility is fairly common, especially where a theory follows the Humean
structure, whereas the former is extremely rare in the natural and social
sciences.
To return to the first description of deviant causal chains the question
arises of how, and by what means, do these refutations become possible?
There are two related reasons for this. The first is that the Humean model
implicitly entails, due to both the constant conjunction and the inherent
inductive expectation (and the circularity discussed in the last chapter),
that a causal relation of the like described will hold in all alike/future
circumstances. This leads, naturally, to the drawing of causal generalisa-
tions which are expressed as law-like universalisations (McIntyre 1994):
8x ðPx^Qx^RxÞ ! : Cx
Where, for example, the schema will read: in all instances if a person has
good and healthy family relations and has personal investment in the
community and has stable employment then that person will not be a
criminal. Or for the second implication of this argument:
8x ð: Px^:Qx^:RxÞ ! Cx
Where the argument is that in all instances if a person does not have good
and healthy family relations and they don’t have an investment in the
community and they are not in stable employment then they will be a
criminal. However, it is rarely possible to make a true universalisation in
the social sciences whilst using such models of explanation because they
fail to capture the complexities, and potential relativism, of human, or
social, interactions (Habermas 1988). Put simply, the model is too
simplistic.
One form of argument that may seemingly offer a way out of this is if
you move away from the universal quantifier and instead utilise an
existential one. In such cases the argument would be that: in at least
one instance (or if you are being very generous in your meaning – there
are instances when . . . ) if a person does not have good and healthy family
relations and they don’t have an investment in the community and they
are not in stable employment then they will be a criminal. However, the
problem here is immediately evident – such an argument reduces this
analysis to a single (or a handful of) referent(s) and that then fails the
Humean notion of constant conjunction. This is the final axiom of
Hume’s account of causation – wherever the antecedent condition is
observed the consequent is also always observed. Without the constant
conjunction (and indeed the inductive expectation) the account fails to
be causal. Thus, one of the most problematic issues identified by Hume
and relating to this form of causal reasoning is that, if your theory is
formulated/constructed in this way, as a theorist you are stuck with the
problems of universalisation and refutation.
Perhaps it is the case that because the Humean model results in this
kind of universalisation that it is, in fact, not this model that has dominated
criminological theorising as very few theorists explicitly make that kind of
universal causal claim. Instead most theorists have couched their conclu-
sions more in terms of probability or degrees of certainty claims rather
than outright universals. The claim here then is that statistical or inductive
modelling is the method employed in criminological causal theory and
that this allows for causal reasoning to be consciously uncoupled from the
burden of deduction with its problematic inherent universalisation. In this

sense what contemporary social science, including criminology, involves is
not formal systems of deduction, as is posited here, but rather forms of
abduction or abductive reasoning which involves statistical mechanisms to
infer to the best explanation (Magnani 2001; Gauderis 2013).
Abductive logic, or abductive reasoning, will be most familiar to readers
not from any scientific work but rather the character Sherlock Holmes. His
famous maxim ‘ . . . when you have eliminated the impossible, whatever
remains, however improbable, must be the truth?’ (Conan-Doyle 1890: 93)
encapsulates the notion behind abductive reasoning. As Walton (2005)
notes this form of reasoning extends, or is generated from, the data and
represents the best possible explanation for that data. Fundamentally one
looks at the given set of facts in a case (or the evidence if you will) and
utilises these to formulate an explanation that captures, or accounts for, all
of the seen data. If an account does not capture the data then it is to be
rejected as a false account and the reasoning must continue until an
account, even if it is an improbable one, fits the data. It is common in
forensic and criminal investigation (hence Conan Doyle’s utilisation of it
for Sherlock Holmes) but is also utilised in most scientific fields – even if
this goes unrecognised. In this sense, this form of reasoning is mostly
recognised as underpinning discovery science and allows for the formula-
tion of hypotheses which can then be formally tested (Gauderis 2013).
Abductive reasoning can generate a phenomenally large number of
possible explanations and as such then is also related to, and indeed must
contain a healthy inclusion of, Occam’s Razor. This heuristic (also known
as the principle of parsimony) fundamentally, and rather simply, states that
entities (or indeed assumptions) must not be multiplied beyond necessity
(Sober 1981). This is important for abductive logic as it offers a means of
limiting the number of potential explanations by excising those that
(in order to capture the data) include extraneous assumptions or ad hoc
justifications. In effect, the razor slices away those explanations which are
unnecessary to capture the given explananda. These two forms of reasoning
then, when combined, allow for the best possible (i.e., least assumptively
laden) inference to be made.
To return to the point – the problem has been posed is it not deducti-
vism that dominates criminology but instead this adbuctivism which
allows the criminologist, a la William of Occam, to slice away potential
and unnecessary causal conditions until left with those perceived as best fit
for the explananda – crime. Such a move would, to a degree, solve the
problems thus far described and therefore render this book superfluous.
However, as Magnani (2001) notes in general, and Weisburd and Piquero
(2008) point out explicitly in criminology, not only do the forms of
probabilistic/statistical models utilised to infer to the best explanation
largely fail in terms of their explanatory power, but also they do so because
their logics are fundamentally deductive. Hypothesis creation may well
involve inductive (Popper 1968) or abductive (Walton 2005) processes
but statistical analyses, in general, do not (Burgess 1998). I am not
attacking the utilisation of statistical methods; they have many benefits
in terms of rendering a more complete understanding of crime phenom-
ena. No, it is not the statistical model that is the problem per se but rather a
failure to appreciate the manner of the logics which underpin them and
the consequences they induce for causal explanations (Cummins 1995).
Furthermore, a problem that arises from this failure to appreciate these
logics is, as was pointed out earlier, that the notion of causality that has
dominated the social sciences is based upon the structure of x causes y if it
is the case that y will not occur unless x does (Woodward 2003; Pearl
2000; Marini and Singer 1988; Beauchamp 1974). This is specifically the
deductive, as opposed to the probabilistic, or inductive, form of scientific
explanation as described by Hempel (1965) and thus results in conclu-
sions that are implicitly universal. Nor is this particular model abductive in
structure – here the consequent is drawn specifically from the premises, or
antecedents, wrought. There is no possibility of alternatives as one is tied
into the premise schema being employed. As previously noted, the deduc-
tive type of explanation cannot result in a probabilistic kind of conclusion
and to do so can result in false conclusions (Salmon 1984). Therefore,
even though theorists may not wish to make such explicitly strong uni-
versal claims, and believe that their statistical/abductive modelling pre-
vents this, the fact is that they are tied into a specifically deductive model
which forces such universal conclusions upon them.
The second related reason regards the causation/correlation problem.
Refutation through deviant causal chains/counterexamples arises due to
the fact that at least one, if not all, of the antecedent conditions whilst true
are not causal in nature at all but purely correlated with the other condi-
tions and the consequent. In this sense the antecedent set of conditions
will be true in terms of logical truth functionality but yet result in a false
conclusion which then results in the Type 1 deviant chain. The problem is
that with the Humean model, unless you take a very narrow realist reading
of the model, there is no way to accurately distinguish which conditions
are correlates and which are actually causal regardless of statistical weight
given to specific conditions (Millican 2009). This arises because of the
simplistic foundation upon which the Humean model is built and that it
does not allow for any meaningful integration of the levels of explanation
that may allow conditions that are correlates to be distinguished from
those that are causal. This is because with the Humean model all the
antecedent conditions are necessarily conjuncted; with each condition
being contiguously and temporally related to those that come before
and those that come after up to the point that the chain of conditions is
sufficient to cause the particular phenomenon in question. The conditions
cannot be explored with any other relations than those dictated by the
model; they are tied into the chain as it were, even if in reality the
conditions are related in some other way.
To a certain degree the problem that the Humean model has with
distinguishing between causes and correlates is also responsible for both
aspects of the second type of deviant causal chain. If none of the ante-
cedent conditions are causal but weak correlates, then it can be the case
that in terms of the Humean structure the antecedent conditions are false
but the consequent still arises. Or it can be the case that the causal, and
correlational, relations that exists between the antecedents and the con-
sequent are more complex than the Humean structure can hope to capture
or explain.
NOTE
1. See Appendix A.
CHAPTER 5
Humean Causation and the History

of Criminology
Abstract This chapter is concerned with tracing Humean causation

through the history of Criminological theorising in a more systematic
way. As such four classic theories from the history of criminological
thought shall be given a somewhat brief exposition and the underlying
deductive Humean causal framework will be exposed and made explicit.
The four examples that have been chosen for exposition are Classical
theory, Biological Positivism, Strain theory and Labelling theory. These
particular forms of theory have been chosen due to the evident framework
of causation that underlies them and how this is largely representative of
the type of thinking, employed logics and theory construction that has
dominated criminological theorising up until the late twentieth century.
Keywords Classicism Causation Humean Labelling Positivism

Strain
The previous chapters have identified the problems that can arise due to an
inadequate understanding of causal explanations, the form of causation
that has been implicitly adopted in the field of criminology and the
weaknesses of that form of causation. This chapter is concerned with
tracing Humean causation through the history of criminological theoris-
ing in a more systematic way. As such four classic theories from the history
of criminological thought shall be given a somewhat brief exposition and

of Causation, DOI 10.1007/978-3-319-47446-5_5
the underlying deductive Humean causal framework will be exposed and

made explicit. The four examples that have been chosen for exposition are
Classical theory, Biological Positivism, Strain theory and Labelling theory.
These particular forms of theory have been chosen due to the evident
framework of causation that underlies them and how this is largely repre-
sentative of the type of thinking, employed logics and theory construction
that has dominated criminological theorising up until the late twentieth
century.
However, before starting it needs to be pointed out that in each
instance the work of just one proponent of each of the classic theories
shall be explored. The reason for this is, firstly, a pragmatic matter of
space and time; and, secondly, that the chosen proponents are repre-
sentative of the method and form of theorising specific to their theory
type. Also, the theories and argument of their proponents will only be
explored to the extent that the inherent causal framework is exposed.
Though this may result in a somewhat superficial exploration of the
theory itself, which cannot be helped, it is nevertheless both necessary
and sufficient for the purposes of exposing the internal logic of the
theory and thus for this book.
The first of the theories to be explored is that known as the Classical
theory. This theory was first outlined by Cesare Beccaria in his book
An Essay on Crimes and Punishments, which was first published in 1764
(Beccaria 1775). The book was, fundamentally, concerned with criticising
the systems not only of law and punishments extant at the time of his life
(Cullen and Agnew 2006) but also of divorcing explanations of crime
from ecclesiastical authority and historical notions of sin (Harcourt 2013).
During the course of that book, he outlined a theory of crime causation
that dominated explanations of crime for more than a century and which
has seen a resurgence in the twentieth century in the form of deterrence
and rational choice theories of crime.
Drawing upon the themes proposed within Aristotle’s Nicomachean
Ethics (man as rational actor) (1999), Epicurus’ Principal Doctrines (man
acts upon hedonistic/utility calculations) (1993), and more specifically
Hobbes’ Leviathan (the social contract) (1991), and embracing the
Enlightenment era’s rejection of ecclesiastical superiority/authority his
theory, in its simplistic form, states that man is a rational being who is
concerned with the subjective maximisation of pleasures and, obversely,
the minimisation of pains or harms. As such, crimes are rationally com-
mitted in order to maximise some form of utility; therefore, punishments
5 HUMEAN CAUSATION AND THE HISTORY OF CRIMINOLOGY 31
necessarily need to be designed so as to ‘prevent the criminal from doing

further injury to society and to prevent others from committing the like
offence’ (Beccaria 1775: 23). He further argues that the mode chosen to
implement punishment must be that which can be administered swiftly, is
proportionate to the crime and which will ‘make the strongest and most
lasting impression’ on both the offender and those that stand as witness to
the punishment (p. 23).
There are really two strands of reasoning that go to make up the theory
of Beccaria. The first is concerned with rationality and the role that this
plays in crime causation. The second is with regard to punishment and the
role that it can play in crime prevention. Each of these strands shall be
formalised and examined, with regard to the underlying notion of causa-
tion, in turn.
The first strand is, as we have seen, concerned with the role of ration-
ality in crime causation. In order to formalise and examine this notion
more thoroughly it is necessary to make explicit the underlying premises.
These premises are:
1. That man is a rational actor

2. That individual utility maximisation can be achieved through crime
Now, let premise 1 be represented by the variable A and premise 2 be

represented by the variable B, and C shall, as before, represent criminality
and x the domain of people. The argument can then be formalised as:
ðAx^BxÞ ! Cx
Where the argument will read: if a person is a rational actor and that
person’s utility maximisation can be achieved through crime then that
person will commit crimes. In this instance it can be clearly seen that
Beccaria’s notion of crime causation fits the deductive Humean model
perfectly.
The second strand of the theory regards the role that punishment can
play in crime prevention. Once again it is necessary to make explicit those
premises that underlie the argument. These premises are:
1. That man is a rational actor

2. That punishment minimises individual utility
In order to maintain some continuity with the schema outlined previously,

let premise 1 be represented by the variable A and premise 2 be repre-
sented by the variable ¬B. The argument thus formalised is:
ðAx^: BxÞ ! :Cx
This time the argument reads: if a person is a rational actor and punish-
ment minimises a person’s individual utility then that person will not
commit crimes. Again it is evident that the causal framework underlying
this aspect of the theory is the Humean model that has been discussed
thus far. As such this strand, as with the first, is open to the problems
that beset the Humean model. Even if you combine the premises to
give a fuller, or more exhaustive, account of the theory it does not
matter because the truth functionality of the overall schema remains the
same. It is still a conditionally structured theory and subject to the same
problems.
Both strands, separate or combined, are vulnerable, due to the condi-
tional construction, to the problem presented by the Type 1 deviant causal
chain and counterexample refutation. For instance, it may well be the case
that man is a rational actor and that utility maximisation can be achieved
through crime but this does not mean that every person commits criminal
acts. It may also be the case that punishment minimises individual utility
yet a person may still commit a crime. Also, the simple nature of the causal
framework prevents other elements of causation, or other levels of expla-
nation, being introduced to the theory in any other way than as further
conjunctive or disjunctive in the overall conditional schema. The result
remains the same. This therefore limits the scope of the theory and that
which can be explained by using the theory.
The second historical example that shall be explored is the Biological
Positivism of Cesare Lombroso. The work of Lombroso, specifically his
The Criminal Man and later papers, was a reaction against the Classical
school of thought that had dominated explanations of crime up to the late
1800s (Gibson 2002). With the rise of the medicalised professions and
socio-cultural fascination with the natural sciences you begin to see in this
period more biological explanations for social phenomena – including
crime. Lombroso encapsulates this fashion. It is worth noting here that
where Beccaria had drawn his notions of rationality from the Classical
thought of Aristotle, Lombroso’s work in his seminal text has much more
in common with the essentialism of Plato (1994). The essential character-

istics of the individual are fixed and these determine the individual’s role
and place in society as well as their behaviour and character. Nevertheless,
they both followed similar logics in the constructions of their respective
theories. Though Lombroso’s first formulations of his biological theory
had been largely dismissed by the fourth decade of the 1900s the funda-
mental tenets upon which his theory was built live on and have had a
profound influence on many later theorists (Cullen and Agnew 2006).
The fundamental point upon which Lombroso built his theory was that
whilst man is indeed a rational actor (this is what defines humanity) crime
is not, and cannot be, a rational act. Therefore, it is not the rationality of
man and his desire to maximise his utility that causes crime; instead it is the
individual biological factors of offenders that causes them to follow their
base (i.e., not rational) desires and be criminal. In effect the criminal
cannot be ‘man’ as they have shown themselves to be acting beyond the
rationality that defines us as a species. He argues that ‘the physical and
psychic organisation’ of criminals is ‘essentially different from that of
normal individuals’ (Lombroso 1910:49). It is these essential differences
in the physical and psychological makeup of criminals that is the result of
them being the brute-like and savage atavists that he describes, which
therefore accounts for their criminality.
Once Lombroso’s theory of atavism, that criminals are more like
primitive ancestors than modern rational and enlightened man, was
refuted he changed tact and argued that biology was not the sole
cause of crime but it was the conjunction of biological factors and the
environment which led the offender type to actually offend (Gibson
2002). For the purpose of this book the more outlandish of
Lombroso’s notions shall be ignored, instead the focus shall be on the
basic tenet that it is the conjunction of atavistic biological factors and
the environment which causes crime. As with the Classical theory dis-
cussed previously, in order to uncover the causal framework employed
by Lombroso, and of Biological Positivism in general, it is necessary to
expose, or make explicit, the premises upon which the theory is based.
These premises are:
1. There are those in society who are biologically different to normal

humans
2. People with these biological differences tend towards criminality
3. Environmental factors can be conducive to criminality
In this instance, let premise 1 be represented by the variable indicator E,

premise 2 by the variable F and premise 3 by G. When the argument is
formalised it looks like this:
ðEx^Fx^GxÞ ! Cx
Where the argument will read: if a person is biologically different to

normal humans and people with this biological difference tend towards
criminality and that person is in environmental circumstances conducive
to criminality then that person will commit a crime, or crimes. As with the
previous example, this theory is a prime example of a theory dominated by
the Humean model and so too is beset with the problems inherent to that
model. In fact, this particular theory is particularly vulnerable to both
types of deviant causal chains. The most obvious example would be
Type 1 where the three antecedent conditions hold but a person does
not commit crimes or there is a person who is as ‘normal’ as others in
society but yet commits crimes. In both circumstances the antecedent
conditions are true but the consequent false. However, there are other
ways in which this ‘chain’ can be deviant. One such is:
ðEx^Fx^:GxÞ ! Cx (1)
This argument is: if a person is biologically different to normal humans

and that people with this biological difference tend towards criminality
and that the person does not live in an environment conducive to crimin-
ality then that person will still commit a crime, or crimes. This instance is
an example of Type 2 deviant chains where the antecedent conjuncts result
in a false set of antecedent conditions but the consequent is still true.
Again the situation arises whereby the causal mechanism that underlies the
theory is inadequate, or unfit, for purpose.
One point that should be noted here is that Positivistic accounts of
crime are often accused of being deterministic in nature and this determin-
ism is reason enough for their refutation. This is partly how Lombroso’s
theories were originally debunked. This argument is predicated on the
notion that humans are not fixed entities and that our character is not
‘essential’. In effect people change – and a theory that presupposes that
our behaviour and character is determined and unchanging is patently
false. To some degree this is a fair point – however, the basis of the
argument tends not to be predicated on the logic employed in the con-

struction of argument but rather in the ontologically realist position from
which much positivistic thought emerges. As Agnew (1995) notes, this
determinism is tied to the notion that there are measurably causal factors
which can be examined in the analysis of crime/deviancy. However, even
here you see the conflation of the ontological position with the logic
employed in the construction of the theory. It is not the realist position
from which this determinism stems but rather the inherent universalisation
already identified earlier. It is the particular logical framework employed
that does indeed tie the positivist into a deterministic logic – but then the
same is true for any explanatory theory couched in the terms thus far
described, including Classical, Control and other non-Positivistic theories
of crime. Therefore, it is unfair to single out Positivistic theories for this
problem when most fall into this trap of determinism due to their logical
formulation.
The third example to be examined in this section is Strain theory, which
dominated criminological theorising, and crime policy in America, in the
1950/60s (Burton jr. et al 1994). This is a somewhat more complex set of
theories than those that were exposited before. The reason for this is the
complexity and sophistication of the argumentation employed by the
various theorists working in this particular area, especially Merton
(1938). However, the theorist that has been selected here is Cohen
(1955) and his work on delinquent gangs. The reason for this is that by
highlighting his version of Strain theory the fundamentals of strain, gen-
eral strain and anomie theory will come to light.
The central focus of Strain theory has been to explain why only certain
aspects, individuals or groups, within societies are more prone to crimin-
ality than other groups and individuals. The conclusion drawn by strain
theorists is that those who are prone to criminality are so because they are
pressured by societal structures into such a position (Burke 2001). Here
the cause is less to do with individual agency but rather the influence or
confluence of structural factors. Ordinarily the argument is that these
groups or individuals are pressured into crime when they are prevented
from achieving their society’s cultural goals, wealth or class status for
example, through legitimate means (Cullen and Agnew 2006). The
novelty of these approaches is that they look beyond the individual to
the role of society, of culture, of political structure and environment to
look for the causes of criminality. The basic premise is that humans do not
operate in a vacuum – the very society in which humans live shape their
interaction, not only with other humans but with the society as a whole. If
crime exists as a problem in society, then it is to society that we need to
look for causes.
Cohen (1955), one of the main proponents of Strain theory, argued in his
work on delinquent boys that the strain, caused by the blockage of goal
achievement, came about not purely because of the barriers to monetary
success that young lower-class, or working-class, males experienced. Instead
it was the wider goal of middle-class status, including peer respect as well as
monetary success, and the denial of legitimate, as well as illegitimate, means
to achieve that goal that caused the strain that Cohen considered the ‘pre-
cedent’ to criminality. However, he argues that this ‘strain’ is not the sole
cause of a person, or group, turning to crime. Cohen argues that ‘strained’
individuals are unlikely to engage in any criminality unless they join, or are
co-opted into, a delinquent subculture whose social mores are conducive to
crime (Cohen 1955). These subcultures arise because though lower- and
working-class males may achieve monetary success through illegitimate
means they cannot achieve the desired middle-class status through the
same means and therefore cannot achieve their goal. As a consequence,
these individuals ‘gravitate’ towards each other and establish alternate, delin-
quent, status systems in which they can achieve success (Agnew 2000).
As before, this exposition of Cohen’s Strain theory is somewhat sim-
plistic. However, it does suffice to expose the salient premises and the
causal structure upon which the theory is built. These premises are:
1. The goal of a lower- or working-class person is middle-class status

2. Goal blockage causes ‘strain’
3. ‘Strained’ individuals will group together to form alternate, delin-
quent, status systems in which they can achieve success
4. Delinquent status systems are conducive to criminality
In this instance, premise 1 is represented by the variable indicator H,

premise 2 by the variable I, premise 3 by J and premise 4 by K. When
the argument is formalised it looks like this:
½ðHxÎxÞ^ðJx^KÞ ! Cx
Where the argument will read: if the goal of a lower- or working-class

person is middle-class status and the blockage of this goal causes a person
to be strained and strained individuals will group together to form
alternate, delinquent, status systems in which they can achieve success and
delinquent status system are conducive to crime then the strained lower-/
working-class person will turn to criminality.
Though this schema is more complex, with one set of conditions
referring to the individual and a second referring to situations, than
those that have been highlighted before it is still evident that the under-
lying structure is that of the Humean model where all antecedent condi-
tions, or sets of conditions, are conjunctively related. As such this theory,
though more sophisticated, still falls foul of the problem of Type 1 deviant
causal chains because of the underlying structure. In this case the theory is
particularly vulnerable to direct refutation due to counterexample if it can
be shown that the antecedent conditions hold true but yet the conclusion
is false.
½ðHxÎxÞ^ðJx^KÞ ! :Cx
Where there is a person of a lower- or working-class status and has the

goal of middle-class status and who is blocked from this goal and is
therefore ‘strained’ in the Cohen sense and it is true that ‘strained’
individuals will group together to form alternate, delinquent status
systems in which they can achieve success and those delinquent status
systems are conducive to crime but yet the person does not turn to
crime. Again this refutational move arises due to the inherent universa-
lisation of the deductive schema adopted. This is not the fault of the
theorists as they are implicitly following the logical structures that had
underpinned the science of which they are adherents. Even if you were
to be more generous in your interpretation of the antecedent sets in
terms of truth functionality you would still end in a situation where the
theory is vulnerable to becoming a Type 2 deviant causal chain. The
sophistication of their argumentation illustrates their dissatisfaction with
the causal arguments that they have previously been exposed to (I shall
return to this point in the next chapter).
The last of the historical examples is that of Labelling theory. Labelling
theory arose in the mid to late 1950s as a natural extension of the urban/
ecological theories of the Chicago School and has survived in a limited, or
subjugated, context ever since. This theory was also a different type of
theorising from that which had gone before. All the theories thus far
examined had as their starting point, and central loci, either the offender
directly or their placement in society. Labelling theory differs in that it
rejects the offender as the central and determining loci and focuses instead
on the behaviour of those within the wider society who label and seek to
control deviancy and ‘offenders’ (Cullen and Agnew 2006).
In its most simplistic form Labelling theory argues that crime is not a
manifestation of individual differences but that it is the efforts at social
control employed by wider society that produces the social circum-
stances that trap individuals onto the pathways of criminality. These
efforts at social control inevitably result in an individual being publicly
‘tagged’, identified as a criminal, segregated and ‘othered’ (Becker
1963). This process entrenches the perception, not only in society, of
an individual as a ‘criminal’, but also in the mind of the person thus
labelled. There is an adhesiveness to the labels imposed which mean that
the individual thus ‘tagged’ cannot escape the label and thus it begins to
define how others interact with that individual and, eventually, how
that individual is able to interact with society. The conjunction of
these two perceptive realities results in a situation where an individual
is trapped into a very particular life course – one characterised by
deviancy and crime.
Lemert (1951, 1972) recognised, and disagreed with, this simplistic
formula, which often portrayed offenders as victims of the wider society
who were deterministically carolled into a life of crime. Lemert’s version
of Labelling theory involved his famous notion of primary and second-
ary deviance. He argued that that in the first instance deviance may
occur for a number of reasons, some of which may be individual and
some of which may be situational or environmental. These ‘primary’
deviations are seen by the offenders as peripheral to their identity and
social role and are therefore not internalised as such (Lemert 1972).
Secondary deviance occurs when the individual no longer perceives their
deviance as peripheral to their social role but instead perceives that
deviance as integral to their subjective identity and role in society. The
person’s life and identity thus becomes organised, and entangled,
around the facts of their deviance and deviant identity (Lemert 1951).
He further argues that the shift from primary to secondary deviance
occurs largely due to the reaction of others. A process, or cycle, of
individual deviance, discovery and overt negative public reaction (cen-
sure) is repeated until such a point that the individual’s deviance is
accepted, internalised, reinforced and entrenched.
As with the previous example this is quite a complex argument invol-
ving both individual and societal factors. A further complication is that
there is a notion of temporality, or of events taking place over a period of

time. The central premises to this theory are as follows:
1. There is some form of minor deviation.

2. Minor deviance is met with social penalties.
3. Further acts of deviance are committed.
4. Further deviance is met with stronger penalties and rejection.
5. Deviance becomes serious.
6. Serious deviance is met with formal action by the society and the
stigmatizing of the deviant.
7. Secondary deviance occurs; the deviance is internalised by the
individual.
In this instance, let premise 1 be represented by L, premise 2 by M,

premise 3 by N, premise 4 by O, premise 5 by P, premise 6 by Q, premise
7 by R and, as before, criminality by C. So, when formalised, the argument
is structured like this:
½ðLx ! MÞ^ðNx ! OÞ^ðPx ! QÞ ! Rx ! Cx
Where the argument reads: if a person commits a minor act of deviance

then this will be met with social penalties and if that person commits
further acts of deviance then these will be met with stronger penalties and
rejection and if deviance becomes serious then this will be met with formal
action by the society and the stigmatising of the deviant; if all this occurs
then the deviance will be internalised by the individual and secondary
deviance will manifest; if this occurs then the individual will be a criminal.
Obviously this is a somewhat simplified version of the argument as there
may well be numerous instances of deviance and reaction that occur before
the shift to secondary deviance occurs. However, the version given is
sufficient to highlight the structure of the argument.
It is worth noting here that the previous ‘simplification’ may also play a
part in why the problem of causation remains in, and why logical formula-
tions of causation have not been considered as pertinent for, sociological
and, of course criminological, reasoning. It is difficult enough to commu-
nicate ideas to others through the written word anyway. If one is to write
in such a way as to fully explicate the nuances of the logic of one’s
argument in its entirety (which is what I am trying to extract/highlight
here) then you soon run into the problem that it becomes unwieldy,
unreadable and (let us be honest here) ridiculous. If you were to do this,
people would not read your works. Therefore, we seek shorthand mechan-
isms of communicating our thoughts and ideas. These logical formula-
tions become hinted at or become underlying assumptions of our
argumentation. They sit in the background both propping up and, in
terms of causal reasoning, undermining the theory at the same time.
This is another reason why the logic of causation in the special sciences
is so implicit to our theory construction.
Returning to Lemertian Labelling theory, as can be seen earlier the
construction of this argument is far more complex than those that have
been examined before. However, it is also evident that the underlying
structure is the same as that which has dominated all the theories thus far.
As with the others this too is vulnerable to the Type 1 deviant causal chain
and may also be open to Type 2. An instance of Type 1 would be when all
the conditions that are involved in the process of shifting from primary to
second deviance occur but yet an individual does not succumb to crimin-
ality. This constitutes a direct invalidation of the argument. Although, and
this is a complexity of the logic involved here and the semantic ambiguity
of the notion of deviance, it may in fact be the case that in order for the
level of secondary deviance to occur the individual in question is already
‘criminalised’. The problem with this notion is that if you add this into the
schema and the argumentation you end up with a hopeless circularity and
the question begging identified in earlier chapters. Fundamentally, and in
short, a person is labelled criminal because they are criminal. This too
would be devastating for the theory.
An example where a Type 2 deviant causal chain could occur would
be if a person engaged in various types of primary deviance, was never
sanctioned due to never being caught, and then became so entrenched
in their deviance that they internalised their deviance and embraced
criminality. In this instance it is not the reaction of wider society that
leads to secondary deviance and criminality. Another possible example
would be someone who had never performed any deviance and thus had
never been sanctioned so no entrenchment of a deviant identity but yet
still went on to commit a criminal act. Both are examples where the
schema fails because it is possible for the antecedent conditions to be
false yet the consequent true.
This chapter has been concerned with tracking the causal structure
implicit to four theories that between them have dominated, to a greater
or lesser extent, the early history of criminological discourse. It was found

that the implicit, or underlying, structure was that of the Humean model
that was identified in an earlier chapter. As such all of the theories here
examined suffer the same problems. Problems that arise solely because of
the implicit adoption of the Humean model of causation.
CHAPTER 6
Paradigm Drift and Criminological Theory
Abstract This chapter is concerned with the second section of this

book: namely, the apparent paradigm shift/drift, in relation to causal
explanations, that has occurred, and continues to occur, within the dis-
course of criminology. Utilising the notion of paradigm shift as introduced
by Thomas Kuhn, this chapter examines why criminological theory has
changed and developed. As such, various forms of criminological theory
(including Life Course, Routine Activity and Developmental Life Coursè)
that have tried to move away from the Humean form of theorising will be
explicated. It will be shown how this theory development has led to dis-
satisfaction with contemporary causal modelling and the search for ever
greater explanatory power. It is this facet of theorising that is evidence of a
paradigm shift within the field of criminology.
Keywords Developmental Kuhn Paradigm Routine Scientific

Theory
The previous chapters, and the first broad section of this book, have been
concerned with identifying the model of causation that has dominated the
history of criminology and exposing the inherent weaknesses of this model
and the consequences of those weaknesses for the theories that have
adopted it as their causal framework. This chapter is concerned with the

of Causation, DOI 10.1007/978-3-319-47446-5_6
second section of this book: namely, the apparent paradigm shift/drift,

in relation to causal explanations, that has occurred, and continues to
occur, within the discourse of criminology.
The term ‘paradigm shift’ has been borrowed from Thomas Kuhn
(1962, 1977) and refers to a major shift of methodological approach or
epistemic thought that can occur in science. The term ‘paradigm shift’
brings to mind quite a violent process of almost instantaneous change.
A Game of Thrones type usurpation of dominance – the theory is dead,
long live the theory! With regard to certain ‘revolutions’ within the
natural sciences, the shift from geocentricity to heliocentricity that
occurred with Copernicus or from Newtonian to Einsteinian physics
for example, this may, if the truth is stretched a little, indeed be the
case (Kuhn 1977). However, this abrupt shift in methodology is quite
rare and has been challenged even in the philosophy of the natural
sciences. It is thought that a much more likely, and prosaic, scenario is
that it is more usual for there to be a slow and gradual process of change
such that ‘paradigm drift’ may be a more appropriate phrase.
Paradigm shifts/drifts occur in the sciences when the extant methodol-
ogies are no longer adequate to account for, capture or explain the specific
problems, or phenomena, under investigation (Brush 1996). Gradually,
and increasingly, the old methods of theorising will be exposed as inap-
propriate and new methodologies, or epistemes, shall emerge to replace
them, which enrich understanding of extent phenomena and perhaps result
in novel discoveries (Kuhn 1977). Eventually, new systems/structures,
evolved from and superior to that which came before, shall become the
dominant episteme or methodology. Really, this is a natural process of
theory evolution – assumptions are challenged, theory adapts, hypotheses
are falsified or disproven, theory adapts – eventually the dominant theory
that emerges may contain nuggets or remnants of the previous theorising
but little of the original systems of thought remain intact (Chalmers 2013).
They become superseded.
Brush (1996) argues that theory change in the social sciences is com-
parable to theory change in the natural sciences but has a slightly more
mutational component. In order to make his argument he explicates the
shift in theory from Gurr’s (1970) dominant Relative Deprivation theory
to alternative theories of collective violence which include resource mobi-
lization, social-process, structural, as well as new social movement theories
which account for the same phenomena but with greater accuracy (Brush
1996). He further argues that this particular instance of theory change is a
6 PARADIGM DRIFT AND CRIMINOLOGICAL THEORY 45
symptom of the wider methodological shift away from ‘quantitative or

predictive hypothesis testing toward qualitative or descriptive explanations
of social phenomena’ (p. 526). This is important to note as it also marks a
change in ontological position in terms of social scientific investigation.
His argument rests on the notion that extant forms of social scientific
explanation were failing to capture or account for the complexities of
social phenomena and thus drastic shifts in thinking were made necessary.
Archer (2007) also argues that shifts in social science ontology may arise
due to dissatisfaction in models of explanation. Here, if realist positions of
science are adopted then they may fail to account for inherently con-
structed phenomena which are culturally/socially (and thus temporally)
dependent. This ontological debate is, of course, important in terms of the
changing nature of social science and goes some way to explaining the
degrees of dissatisfaction with theory. However, I argue that it is not so
much the ontology that has given rise to dissatisfaction but, as thus far
described, in models of causal explanation that have been traditionally
utilised. The ontology here is irrelevant if the logics employed in causal
theorising are inadequate – the dissatisfaction will always arise. Changing
ontology in order to create theory which may capture the specific expla-
nanda under investigation is pointless if that is not where the problem lies.
Apart from the emergence of more qualitative forms of explanation in
criminology, which has mirrored that occurring in the broader field of
social sciences, there has also been a broad shift in general criminological
theorising. Garland and Sparks (2000) point out that the history of
theoretical criminology has been one of dynamic change involving shifts
in theoretical dominance and influence. They trace these changes with
special attention paid to the role of modernity and late-modernity and the
impact of these social trends on criminological discourse and methodolo-
gies. Part of this process results in complexity being forced upon theories
due to the wider forms of societal, theoretical and functional changes
which criminology needs to capture. If you look beyond the individual
or structure in order to explain crime, then you must need also develop a
theory that can account for multiple influences on the individual at differ-
ing points and in differing ways.
It is the core contention of this book that there is in fact a ‘paradigm
shift/drift’ currently underway within the field of theoretical criminology.
This is not just the normal course of theory progression but a more
distinct rejection of one methodology and the search for a replacement.
It is proposed that such a process began, and has continued, due to the
dissatisfaction within the field with the extant causal explanations and
mechanisms – not competing ontologies. Also that this process is taking
place against the backdrop of the changes occurring in the wider contexts
discussed previously. The purpose of this chapter is to, somewhat briefly,
track this shift and the attempts of theorists to avoid the problems and
constraints of the Humean model of causation.
In order to make the argument described here it will be necessary to
once again look at a number of theories and show how the theorists in
question have attempted to either solve, or avoid, the problems that have
been discussed in the previous chapters.
As has been argued before, one of the main problems of the Humean
model of causation is the manner in which it limits the integration of levels
of explanation to those which occur in a lineal or conjunctive relation. This
was the problem that was highlighted by Sheldon and Eleanor Glueck in
their work on juvenile delinquency (1950). They argue right at the start of
their book that any study of the causation of delinquency that depends
solely upon one form of theoretical approach will be insufficient for the
task at hand and will result in bias, ‘weakness and incompleteness’ in causal
reasoning. Due to their explicit identification of this as a problem for
crime/deviance theorising in 1950, and their evident frustration in
attempting to overcome these problems, this shall be taken as the starting
point for the paradigmatic drift discussed herein.
The Glueck’s work on juvenile delinquency arises from a study that they
began in 1939 where they compared 500 institutionalised delinquent male
youths with a matched group of 500 non-delinquent male youths from
Massachusetts in the United States. Their major findings from this study
were published in their seminal work Unravelling Juvenile Delinquency in
1950. The dominant theories at the time of their publication were those
that were sociologically focused, such as the Strain or Labelling theories
outlined in the previous chapter. However, the Glueck’s study caused them
to disagree with the monadic approach taken by other theorists. They
concluded that juvenile delinquency, and therefore criminality, derives
from the complex interplay of a number of different forces (Laub and
Sampson 1991). These diverse forces included somatic, temperamental/
emotional and intellectual traits as well as various socio-cultural notions
such as family/home circumstances and the particulars of the inhabited
environment (Glueck and Glueck 1950).
One of the main criticisms of the Glueck’s work is that though they
argue that delinquency/crime is caused by the complex interplay between
the previously mentioned factors/forces they do not give an account of

the relations between these factors in any great detail. However, they
argued at the beginning that the causal reasoning adopted by other
theorists was inadequate and that they were impelled, by their findings,
to adopt a more multidimensional interpretation (Glueck and Glueck
1950). It is evident in their argument that the reason that the Gluecks
do not describe the relations between these factors in any great detail is
because they did not have a causal framework that allowed for the kind of
depth of explanation that their evidence demanded. They were frustrated
by what they perceived as a lack of an adequate framework upon which to
pin their data. As such, this resulted in the presentation of their conclu-
sions, with their implicit complexity of causation, without offering a
structure by which the mechanics of causation could be understood.
This is clear evidence of a causal theory arising from collated data where
the causal relations cannot be captured by the causal scheme, the Humean
model, which dominated criminological theorising at the time. This is
really the first instance to be found in the history of criminology where
theorists have found the extant means of theory construction to be inade-
quate for purpose and have sought other means of explaining the causa-
tion of crime.
Though the Gluecks were the first to actively criticise the methodology of
their day and sought a different way of accounting for causation, they were
by no means the last. Other theorists have also identified problems with the
dominant notion of causation and have attempted to avoid the inherent
problems. One such attempt has been the axiomatic approach taken by
Routine Activity theorists such as Cohen and Felson (1979). An axiom is
any given proposition that is either self-evident or beyond logical doubt – for
instance a bachelor is an unmarried man, a prime number can only be divided
evenly by itself or 1, Reviewer 2 is unnecessarily harsh (this last is of course a
joke – maybe). They are unchallengeable quanta of data or positions – they
exist rarely outside of mathematical or logic systems. An axiomatic approach
to theorising is one predicated on mathematical or logical deduction
whereby the conclusion inherently arises from each of the premises (axioms)
given – in essence the combination of axioms leads to a self-evident or
unquestionable conclusion. The aim here for Cohen and Felson was to
establish a set of axioms which can account for crime causation. Once this
move has been taken, the need to account for causation is, seemingly,
dispensed with as it becomes a deductive given; the problem is then to
argue how the circumstances in which these axioms can occur arise.
The approach taken, or the argument presented, by Cohen and

Felson involves two notions. The first is that crime occurs when there
is a convergence of the three following axioms: (1) a motivated offen-
der; (2) a suitable target; and (3) the absence of capable, or affective,
guardianship. The second notion is that structural alterations in the
routine activity patterns, specifically changes that directly impact upon
any one of the previously mentioned axioms, can influence crime rates
(Cohen and Felson 1979). This is due to their arguing that, as with any
axiomatic system, the absence of any one of the three axioms is suffi-
cient to prevent the occurrence, or perpetration, of the consequent, in
this case a direct-contact predatory crime. As the motivation of offen-
ders is assumed to be a given, no argument is given with regard to that
axiom. However, it is the other two axioms, more specifically how the
routine activities of people can impact upon them, that are the focus of
their theory.
Though there is much more to the theory presented by Cohen and
Felson, it is their attempt to establish an axiomatic system of crime causa-
tion that is really of interest here and thus the previous description is
sufficient. This ‘move’ is interesting because it is a reversal of the usual
crime causation theory. Most theories that have thus far been examined
have had as their end goal the establishing of what factors cause people to
commit crimes and present their conclusions in accordance with that
question. As we have seen, this usually results in some form of causal
conclusions that take the form of the Humean model and therefore face
the problems that have been described. However, Cohen and Felson, in an
attempt to avoid the problems that other theories have faced, look at the
problem of causation from a different perspective. They establish at the
outset that crime is caused by the convergence of their axioms and then
seek to extrapolate, by means of syllogistic deduction, further conclusions
from their inferential scheme.
Cohen and Felson’s theory is a clever attempt at avoiding the problems
posed by causal reasoning by limiting their explanation to this reversed
deductive model. However, ultimately it fails because they are still depen-
dent upon the Humean model of causation in at least two ways. The first is
with regard to their axioms and the second is with their conclusions drawn
concerning the manner in which crime rates can be influenced by the
structural changes to routine activities that they describe (Cohen and
Felson 1979). To take the first point: the axiomatic schema that they
employ is still grounded in the Humean ontology. If we take the three
axioms to be represented by the following indicators, W, Y and Z, then

when the argument is formalised it will take the form:
ðWx^Yx^ZÞ ! C
Where it reads: if there is a motivated criminal and there is a suitable target

and there is an absence of affective guardianship then a crime will occur.
As with the theories presented in Chap. 5, due to the inferential scheme
employed and the implicit inductive expectation inherent to it, this argu-
ment is vulnerable to the Type 1 deviant causal chain. The second point is
that they argue that if any one of the axioms is absent then this is sufficient
to prevent the occurrence, or perpetration, of the consequent. However,
this would depend on the three axioms being not only genuinely causal but
also the sole elements of crime causation. This is a problem because where
the Humean model is adopted it is difficult to establish, either through
argumentation or empirical test, the distinction between causal and corre-
lational factors. For instance, if their theory holds then you should be able
to design out crime from any given circumstance – yet this has proved
impossible to achieve. Perhaps they would argue it is because it is impos-
sible to actually develop affective guardianship in all, or even most, circum-
stances. However, perhaps it is impossible to achieve because this last factor
plays no causal role whatsoever? The problem here is that you cannot tell.
Terrie Moffitt is another theorist who was not satisfied with the
approach offered by traditional theories and sought, in the same fashion
as the Gluecks, to establish a more comprehensive integration of levels of
explanation in her theory of crime causation. The main premise of her
account is that there are two distinct categories of offenders, those who are
persistently antisocial and those whose antisociality is temporary and
situational (Moffitt 1993). She goes further and states that these two
categories of antisocial offenders are qualitatively different and, therefore,
the causes of their criminality are in need of separate, or distinct, systems of
explanation. She defines these two distinct groups as follows: Life Course
Persistent (LCP) offenders and Adolescent-Limited (AL) offenders.
LCP offenders are marked by their continuity in antisocial behaviours.
Though the manifestations may change the entrenched personal anti-
social disposition remains a constant throughout their life course. This
transitivity across age is also matched by the consistency of their cross-
situational antisocial behaviour. No matter the social context, the LCP
type will manifest their antisociality. The cause of this, she argues, must
lie in factors that were present either during foetal gestation, the birth
process or neonatal maturation. She posits the idea that neuropsycholo-
gical anomalies may lay at the root of the antisocial behaviour of LCP
offenders. The most important point is that it is the interaction between
the effects of these neuropsychological anomalies and the post-natal
environment, particularly an inimical parenting context, which causes,
and reinforces, the entrenchment of the LCP offender’s underlying
antisocial disposition. It is the constant conjunction of the LCP offen-
der’s antisocial traits and their environment, and the consequences of
this constant interaction for them, throughout their developmental life
course, that insidiously constructs their rigidly stubborn antisocial per-
sonality and, therefore, their criminality.
Adolescent-Limited (AL) offenders are markedly different from the
LCP group in a number of ways. Some of which are: they are present in
the population in a far higher frequency than their LCP peers, as she
notes they are ubiquitous in contemporary Western societies (Moffitt
2006); they do not portray, or not to any significant level, the history of
antisocial behaviours during childhood; and their offending pattern is
marked by episodicity and discontinuity and tends to be situationally
restricted. Also, their anti-sociality contains an overt element of instru-
mentalism – they will act in ways that are antisocial if these acts will
maximise their perceived utility in given and certain circumstances but
refrain from such behaviours if the rewards of prosocial behaviours is
greater in those same circumstances.
The differences do not stop at the type, or manifestation, of antisocial
behaviour; there is also a difference in the genesis of that antisocial
behaviour. AL antisociality is a product of ‘social mimicry’ (Moffitt
1993). She explains that this ‘social mimicry’ can only account for the
antisociality of AL-type personalities if and only if that mimicry offers some
form of access to, or attainment of, some desirable resource. The resource
in this context is, she argues, maturity status (or if you are of a more
sociological frame of mind – social capital) amongst peers and the wider
societal context.
It is clear that though in certain regards the causal account offered
by Moffitt follows in the footsteps of the Gluecks, for example, by
attempting a far more complex and multi-layered, or multidimensional,
explanation her argument is still entrenched within the Humean causal
ontology. Her argument is still constructed in the format of: if X and Y
then . . . Her explanation of crime is incredibly sophisticated and, to a

degree, has weathered the sands of time empirically. Much of the more
modern or contemporary developmental (life-course) criminology is pre-
dicated on her work and the formulation of theory which includes biolo-
gical, structural, social and relational factors in the causation of crime. The
problem is that her theory construction adopts, and thus includes the same
problems as, pre-existing forms of causal theorising. Her attempt to
account for the causal complexity of her two offender group taxonomy
results in an antecedent set of lineal conjuncts that are subject to the same
problems of those theories she was attempting to move away from.
There are many other examples of theorists attempting the same ‘shift’
away from this causal methodology. Theorists such as Bernard and Snipes
(1996), Barak (1998) and Farrington (2005) all follow this trend of
steady, and implicit, rejection of the Humean model. Some even going
as far as the Gluecks in identifying and explicating their dissatisfaction with
the types of causal theorising open to them. Others, whilst attempting to
cultivate a causal explanation, leave the notions of causation and the
problems associated with it out of their argumentation altogether.
However, they all fall foul of what Marini and Singer (1988) identified
as the unintended consequences of implicit causal reckoning. Theory that
does not capture causal mechanisms at the same time as arguing that their
theory does – in this case explaining the causes of crime.
Perhaps the most, if not the first, successful attempt to completely, and
unconsciously, uncouple their theory from the problems of the Humean
ontology is the Developmental Life Course (DLC) theory proposed by
Laub and Sampson (Sampson and Laub 1993, Laub and Sampson 2003).
Their theory can be seen as a natural progression, or even theoretical
evolution, from that proposed by Moffitt and arose from their discovery
of the original data from and then a re-examination, and continuation, of
the longitudinal study of juvenile delinquency first started by the Gluecks.
Their DLC theory, though an evolved form of, is very different from
that proposed by Moffitt. They accept that life course trajectories of
antisocial behaviour may well extend back into childhood and present
on a continuum over substantive time periods (Laub and Sampson
2003) however, they disagree that this process is rigidly determined.
They argue that there is a far more complex set of factors that combine
in entrenching an individual on an antisocial pathway (Sampson and
Laub 1993) than what Moffitt allows and that people are not con-
demned to be LCP offenders. Some of the factors they describe are
those described by Moffitt and others: individual (somatic) differences,

environmental and socially contextual factors but, and this is where
they diverge from the others, it is the constant interaction of these
factors coupled with more random influences, personal choice (they
argue that individuals play active roles in the construction/manifesta-
tion of their lives (Laub and Sampson 2003)) and the absence or
obtention of social bonds, that account for life-course antisociality or
desistence.
In the explanation of persistent offending they argue that the interac-
tion between individual differences and environmental factors can lead to
social contexts that encourage antisocial behaviours. These in turn interact
with other factors, social contexts and personal choices to produce delin-
quency. Once an individual is engaging in delinquency this behaviour
weakens their social bonds which then can lead to further delinquency
and adult criminality. Adult criminality further weakens any social bonds
that the individual may have and leads to a situation where the routine
activities of that individual place them in situations where antisocial/
criminal behaviours are the norm.
However, Sampson and Laub do not accept that individuals are trapped
into a predetermined life course of persistent offending – change is as
possible as continuity. Continuity in offending is marked by instability in
all aspects (family, home, school, work, etc.) of the individual’s life which
results in the absence of adequate social bonds. On the other hand,
desistance is marked by the obtention of ‘quality’ social bonds. An exam-
ple of a ‘quality’ social bond is that of a ‘good’ marriage or partnership. In
such a relationship the individual is, due to their investment in the rela-
tionship, increasingly bonded into more prosocial behaviours and routine
activities. The consequence of this is that the routine activities where
antisocial behaviours were the norm recede into the past as does the
criminality associated with them. Desistance in this instance can either
occur by default (the non-contemplated adoption of prosocial roles) or
through personal agency (the choice not to engage in antisocial behaviour
so as not to lose relationship).
There are two aspects of this theory that truly mark its implicit
rejection of the Humean model. The first is that the relationship
between the antecedent factors is not necessarily that of lineal con-
juncts. They argue that there is a constant interplay between the diverse
sets of factors, which in some cases may be causal but in others correla-
tional, which can then produce either persistence (continuity) or
desistance in criminality. This notion, in and of itself, cannot be cap-

tured by the simplicity of the Humean model and is thus an implicit
rejection of it as the underlying causal framework. The second aspect of
the Sampson/Laub theory is the notion that, to a certain extent, crime
causation can be subjectively relative. There is no constant conjunction
nor inductive expectation, and thus no universalisation of the causal
conclusion, included in this theory. What they describe is a dynamic
process of interactions between sets of diverse factors that are not
universally deterministic. This again is a rejection of the precepts that
underlie the Humean notion of causation.
This evident rejection of simplistic causal mechanisms continues in
contemporary criminological theorising. Theorists such as Wikström
(2006) now not only point out explicitly that there is a direct need for
an understanding of causal mechanisms that can account for multidimen-
sional (and multidirectional) levels of explanation but also offer causal
theories that have implicitly adopted a causal framework that differs from
the Humean ontology. These latter rejections show that the pre-existing
models of explanation are no longer able to capture the type of causation
now being theorised. As such, newer notions of causation are being
adopted and tested and it is this process that is a hallmark of an ongoing
paradigm shift/drift.
In this chapter it has been argued that a paradigmatic shift is underway
within the field of criminology. As has been shown, theorists such as the
Gluecks, Moffitt and Sampson and Laub were dissatisfied with the causal
paradigm that had dominated much of the history of criminology and
have implicitly adopted a causal structure that is more complex and
which allows for greater integration of levels of explanation than the
Humean model. However, the problem, as Wikström points out, is that
the causal reasoning offered by even these sophisticated accounts of
crime causation is still implicitly adopted and, therefore, ill defined.
CHAPTER 7
INUS Conditions and Criminological

Theory
Abstract This chapter is concerned with proposing, or revisiting, a model

of causation that would be sophisticated enough for the purposes of
contemporary, complex causal theorising. It will be argued that the sug-
gested mechanism, an adaptation of Mackie’s INUS model, is an adequate
causal model for criminology theorising. It shall be shown that this
particular model, whilst not novel, works by explicitly defining not only
that which can be counted as causal but also how a causal factor must fit
into the structure of the mechanism. Also, it explicates not only how this
model is more suited to contemporary integrated forms of theorising that
arise in contemporary criminology but also how the INUS conditions
solve many of the problems, such as the confusion between correlates
and causes, that besets the Humean causal ontology. The chapter also
explores some of the criticisms and problems that this causal explanation
may face.
Keywords Causation Cause Correlation Explanation INUS Mackie
The last chapter explored the paradigm shift that has occurred, and con-
tinues to occur, within theoretical criminology. It was argued that the
reason for this shift was the evident dissatisfaction with the causal structure
that had dominated earlier theorising and which constrained causal theo-
rising to a lineal model. As was shown, many theorists began to adopt

of Causation, DOI 10.1007/978-3-319-47446-5_7
a more complex notion of causality than that offered by the intuitive

Humean model. However, the problem is that though more complex
notions of causation are being adopted this is still an implicit process.
There is still the need for an explicit model of causation that could be
adopted within the field of criminology and which would be sophisticated
enough to account for a multidimensional approach to causation.
This chapter is concerned with proposing (or revisiting) a model of
causation that would be sophisticated enough for the purposes of con-
temporary, complex causal theorising. It will be argued that the suggested
mechanism, an adaptation of Mackie’s INUS model (1965, 1966, 1974),
is an adequate causal model for criminological theorising. It shall be
shown that this particular model, whilst not novel, works by explicitly
defining not only that which can be counted as causal but also how a causal
factor must fit into the structure of the mechanism. Also, I will explicate
not only how this model is more suited to contemporary integrated forms
of theorising but also how the INUS conditions solve many of the pro-
blems, such as the confusion between correlates and causes, that besets the
Humean causal ontology. I will also explore some of the criticisms and
problems that this causal explanation may face.
However, before beginning an examination of Mackie’s account of
causation it is necessary to explain why this particular model has been
chosen over others that currently exist. There are many theories of
causation that exist in the literature of the philosophy of both the
natural and social sciences. Other models, for instance some of those
proposed by Kenny (1979), are so mathematically complex that though
they may be fit for criminological purpose they would become unwieldy
in the construction of a causal theory. Also, more prosaically is that they
are so mathematically complex, being dependent on multi-variant
forms of inter-relating systems of probability, that they are, at least if
you’re anything like me, so inaccessible that they would just add further
complexity to the horrendously complex nature of causal reasoning.
Perhaps one of the most intriguing alternative forms of causal reason-
ing is known as the Counterfactual Model of Causation (Morgan and
Winship 2015; Hoerl et al. 2011). This model proposes that it is possible
to determine the causes of any particular phenomena by extrapolating or
experimenting, in the case of controllable and testable hypotheses, with
the removal of supposedly causal conditions or factors. It is the testing of
the removed conditions that enables you to abduct to the best explana-
tion of what counts as a causal account. In some regards this is reasoning
7 INUS CONDITIONS AND CRIMINOLOGICAL THEORY 57
by negation – would y be the case if x did not occur? If y would not occur
given the absence of x then x cannot be causal. If x does in fact prove to
be causal, then you must need investigate in what way. In this sense, this
means of reasoning involves a bastardisation, and a negative form, of
abductive logic, Occam’s Razor and deduction.
If you remember back to the earlier chapters and the definition of
causation identified by Marini and Singer (1988) as having implicitly
dominated the social sciences – x causes y if it is the case that y will not
occur unless x does – you can see that there is a degree to which counter-
factual modelling may have been present if the latter half of this deduc-
tive logical construction had been utilised in and of itself. However, as
previously argued, the formulation, in this fuller guise, added levels of bi-
conditional complexity (or vicious circularity) to the reasoning necessary
to formulate theory and thus largely remained an assumptive inclusion. It
is often taken as a given. However, counterfactual modelling of causation
actually makes this process explicit and to a degree avoids the circularity
as you need to ensure two things: (1) that any given premise plays a
causal role in the outcome; and (2) once you have discovered that it has it
then behoves you to uncover the actual nature of that relationship. This
means that you can’t leave it at the x does indeed cause y – you have to
explain how and why. The question arises here then is: whether or not
this counterfactual form of reasoning is suitable for criminological
theorising?
If we return to the simplified Control theory utilised earlier then we can
examine the utility of this method. If you remember the second formula-
tion of the theory was if a person does not have good and healthy family
relations and they don’t have an investment in the community and they
are not in stable employment then they will be a criminal. In order to use
the counterfactual method of investigating causation we would need to
look at each of the antecedent conditions and ask whether or not a person
would become a criminal if they were absent? It is here that we run into
the first of the problems with this form of reasoning. The Control theory,
being formulated in the Humean ontology, stipulates that it is the con-
junction of these factors that results in criminality and thus you can’t test
each premise in isolation. You could attempt to resolve this by looking
at combinations of the antecedent factors so as to determine how they
interact in order to become causal. Nevertheless, if you attempt to resolve
this issue you then run into the second problem with this model. If you
remove parts of the lineal conjunction, then logically you should not have
criminality as the premises from which to deduce the consequent have

been altered (you’ve changed the theory in effect). If you do end up with
criminality, then it becomes impossible to tell whether you have a type
1 or 2 deviant causal chain with the result that the explanatory power of
the theory is compromised and thus it will fail the test of aetiological
extensivity.
Let us for a moment ignore this problem and assume that this was
possible and that we have multiple antecedent conditions with which to
work, such as in the theory of Terrie Moffitt described in the previous
chapter where there is a combination of biological, environmental and
relational factors. Technically it should be possible to work out which factors
and which combinations of factors do or do not have a causal relationship to
criminality by using this method to whittle out those factors that are not
causal. There is also a developmental aspect to this when you pose the ‘What
if’ – by asking the very questions you are in the process of developmental
theorising as you may define new theory by testing the extant one. There is
complexity here though that causes problems for the social scientist. As
Hoerl et al. (2011) point out, this is a matter, or a process, of elimination
in which you must need account for what may happen as well as what has
happened in order to ensure thoroughness. In simple theories this is not that
difficult a process (which ball on the pool table caused Ball A to move, for
example) but the greater the complexity of the theory the more difficult this
becomes – if you think that a theory may have 10 sets of variables then there
is a possibility of 1010 (1,000,000,000) combinations of variables that you
would need to posit in order to eliminate all possible counterfactuals.
However, this is a computational sleight of hand and one that may be
overcome with some sophisticated programming. No, the real problem
here lies in that it is impossible in this instance to test whether or not the
removal of these conditions would lead to criminality or not. This is the
fundamental problem with counterfactual modelling of causation in the
social sciences – it presupposes that we can manipulate the variables of a
person’s reality to posit alternative outcomes in any accurate way. In order
for us to do so these variables would need to be fixed and immutable in
their effects – this is rarely the case in human sciences. The very variables
with which we, as criminologists, need to construct causal theory are
inherently different from chemicals in a test tube, bacteria in agar or
atoms in the Large Hadron Collider. In each of these circumstances it is
possible to control the variables, the environmental conditions and the
manner in which they interact and predict with accuracy what the
outcomes will be. In essence, counterfactual manipulation is possible. The

social sciences, and in particular criminology, are very different. We do not
work with neat (i.e., controllable) systems – our fields of investigation are
messy due to their human-laden nature. If we can’t control the variables
we can’t be accurate in predicting/positing what the consequence is if we
hypothesise a situation where some, or combinations of some, of the
variables are removed.
Of course, this is what experimental criminology attempts to investi-
gate. They attempt to establish controllable variables. However, as a field
there are limits not only to what is practical and achievable but also to
what is ethical. For instance, if you really wanted to test the hypotheses of
a Control theory of crime then you would need to ensure that some
people have poor family relations, no stable employment and no invest-
ment in the community, whilst others did and see if they did indeed
become criminal. You could match pre-existent populations (which is
what we do try and do) but then you’re not really in control of the
variables that you are trying to test. At best you can assign them prob-
ability scores but then you are making some unjustifiable assumptions
about these variables which could severely hamper what results you draw.
Here you are rendering your own theory/research to a level of Rosenthal
Effect/Bias (Martin 1994) that is, in effect, an act of scientific self-harm.
It is for these reasons that the counterfactual model is not seen as being a
particularly good method of causal reasoning in criminology for anything
but the most basic theories.
It is also why Mackie’s INUS conditions model has been chosen as
(potentially) a suitable model for criminology. There are a number of
reasons and the first is that, in terms of accessibility, it is one of the easiest
models of causation to not only explicate but also utilise that will be
adequate for criminological theorising. Secondly, unlike some of the
mathematical models it does not rely upon a lineal (or axiomatic) struc-
ture and can, therefore, cope with multidimensional explanations and
avoid certain deductive consequences. Thirdly, either this model, or
adaptations of it, have been adopted in other branches of the social
sciences and have been found to be adequate (Marini and Singer
1988). The reason for this is that though the model provides an adequate
structure around which to build causal reasoning it also forces a deeper
level of argumentation in differentiating causal from non-causal factors
which promotes better theorising. Fourthly, it is this model that most
closely resembles the type of theorising that is currently occurring
amongst those who have rejected, or shifted away from, the simplistic
Humean causal structure.
Firstly, what is Mackie’s account of causation? The central point to
Mackie’s account of causation is that some event/phenomena can only be
counted as causal if and only if it can be explained in terms of INUS
Conditions (Mackie 1974). An INUS condition is where a ‘ . . . cause is,
and is known to be, an insufficient but necessary part of a condition which
is itself unnecessary but sufficient for the result’ (Mackie 1965: 247). His
classic example is that of a fire that breaks out in a house. The conclusion,
by the investigating experts, is that the cause of the fire was an electrical
short circuit. He states that, in and of itself, the electrical short circuit is an
insufficient but necessary part of the cause of the fire. There needs to be
further conditions, such as the presence of flammable material and the
absence of an adequate sprinkler system, for instance, which, in and of
themselves, are unnecessary but sufficient, even minimally sufficient, for the
fire to actually occur. There needs to be a conjunction of these two sets of
circumstances, one in which the cause is insufficient but necessary and the
other in which the conditions present are unnecessary but sufficient, in
order for the fire to occur.
He formalises this notion in the following way: let A stand for the INUS
condition, P the result, X those circumstance which are to be conjoined with
A, and Y the disjunction of all the possible minimally sufficient conditions.
From this you get the formalised concept that ‘A [the short circuit] is an
INUS condition of P [the house fire] iff, for some X [the presence of
flammable material] and for some Y [the absence of an adequate sprinkler
system], (AX or Y) is a necessary and sufficient condition of P, but A is not a
sufficient condition of P and X is not a sufficient condition of P’ (Mackie
1965: 248).
This seems somewhat complex However, it can be slightly adapted so
that it can be formalised in the following manner:
½ðA ! PÞ $ ððA^ ðXvYÞÞ ! PÞ
Where this will read: if the short circuit (A) occurs then the house fire
(P) will occur if and only if it is the case that if the short circuit occurs
(A) and there is a presence of flammable material (X) or the absence of
an adequate sprinkler system (Y ) then the house fire occurs. In this

instance it is implicit to the last conditional that the conjunction of the
short circuit and the presence of flammable material or the absence of
an adequate sprinkler system amounts to a situation both necessary and
sufficient for the house fire to occur.
The description that is given above is that which Mackie gives for
singular causal claims. For general causal claims he adds the notion of a
causal field which is, essentially, the domain of all possible features intro-
duced by a general causal claim. In such cases the formula is (AX or Y ) is a
necessary and sufficient condition of P in F; where F is the causal field
(Mackie 1965). This means that if a theory demands that there be more
than one causal factor, which as we have seen is usual in the social sciences,
then the causal field (F) is, essentially, the set of all factors which are causal
in the INUS sense. However, for every single factor, if they are to be
considered causal, then they too must be able to be explained in terms of
the INUS conditions.
In order to show how the INUS model could work in practice, let’s
take a simplistic imaginary criminological theory and subject it to causal
analysis in the INUS sense. Professor B has conducted a longitudinal study
on delinquency and crime and her data has led her to conclude that
criminality is caused when a person is raised with inimical parenting in
an environment that is conducive to antisocial behaviour and has non-
supportive schooling. In order for Professor B to claim that inimical
parenting is causal in the INUS sense she must show, by way of her
evidence and argumentation, two things. Firstly, that her theory satisfies
the formal INUS model. Secondly, that inimical parenting is in fact an
insufficient but necessary condition which is part of a set of conditions, an
environment that is conducive to antisocial behaviour and non-supportive
schooling, that are, in and of themselves, unnecessary but sufficient to
allow criminality to occur.
In order to see if Professor B’s theory does satisfy the model it is
necessary to formalise it and check to see if it remains logically valid
once formalised. Therefore, if the premises are:
1. Inimical parenting: represented by the indicator I

2. An environment that is conducive to criminality: represented by E
3. Non-supportive schooling: represented by S
And, as before the domain of people is represented by x and criminality by

C then the argument formalises as:
½ðIx ! CxÞ $ ððIx ^ ðEx v SxÞÞ ! CxÞ
The argument then reads: a person who has inimical parenting will be
a criminal if and only if it is the case that, if a person has inimical
parenting and lives in an environment conducive to criminality or does
not have supportive schooling then they will be a criminal. If, however,
professor B had discovered other causal factors, such as neuropsycholo-
gical anomalies (N) and goal deprivation (G), then each of these factors
would too have to be explained in terms of the INUS conditions and
would go to make up the causal field (F). In such a case the formal model
would be:
½ððIx; Nx; GxÞ ! CxÞ $ ðððIx; Nx; GxÞ ^ ðEx v SxÞÞ ! CxÞ
You will note that the factors in this causal set are not conjuncts of each
other. The reason for this, and this is one of the things that sets this model
apart from the Humean one, is that it is not the conjunction of these
factors that are causal it is that each of the factors is causal, in terms of the
INUS model, in and of themselves. If, as with this case, the causal field
only contains three factors then they can be represented individually
within the model. However, if there are multiple causal factors then it is
easier to represent the set of causal factors by the set indicator (F) and then
denote the factors which go to make up F:
F ¼ðI; N; G; H; O . . . Z Þ
½ðF ! CxÞ $ ððF ^ ðExv SxÞÞ ! CxÞ
Now that there is some understanding of the INUS model and how it
could work in practice, it is necessary to discuss the benefits of this model
over that of the Humean model. The first is really that which defines the
theory and is the very specific manner in which a factor can be considered
causal. Unlike the Humean model, Mackie has captured the notion that
causal phenomena do not take place in a vacuum. All causal events take
place in a matrix of other factors, conditions and phenomena which form
the background environment in which a causal event can develop. His
model is designed to capture this notion. The set of unnecessary but yet
sufficient, even minimally sufficient, conditions can contain all those fac-
tors, conditions and phenomena that go to make up the backdrop in
which the event/phenomena under investigation has taken place. The
Humean model does not allow for this kind of notion, all conditions or
factors must be conjuncts of each other in the antecedent set.
This leads me to the second advantage of the INUS model and it is one
that was mentioned earlier. It was pointed out that with the Humean
model the antecedent set of conditions involved the conjunction of all the
conditions in that set. As we have seen, this led to the problem that if one
of those conditions was not in fact causal then it led to a situation
where the antecedent set was false and could result in a Type 2 deviant
chain. The INUS model does not rely upon the conjunction of the
conditions/factors to be causal as each of the factors is causal, in terms
of the INUS model, in and of themselves. This means that the INUS
model explicitly avoids the problem of Type 2 deviant chains by virtue of
the construction of the model.
The third advantage is that this model also avoids the problems
posed by Type 1 deviant chains. As we have seen the Type 1 deviant
causal chains arise where all the antecedent conditions are present, and
conforming to the axiomatic strictures of the model, and are therefore
true, but yet the expected consequent does not follow and is thus false.
This is an example of the counterexample move and direct refutation.
This is a problem specific to the Humean model because it is expressed
in term of the material conditional. The INUS model is expressed in
terms of the bi-conditional and as such it cannot be the case that one
half of the expression is true whilst the other false and remain valid.
If one side of the formal expression is true and valid then by virtue of
the bi-conditional so is the other; likewise, if one side is false then so is
the other. This means that any theory validly expressed in the terms of
the INUS model is not vulnerable to Type 1 deviant causal chains and
thus cannot be falsified by means of a counterexample. The only way in
which the INUS model can be falsified is if every claimed causal factor,
when conjoined with the set of unnecessary but sufficient conditions, is
shown, in fact, not to amount to a situation both necessary and suffi-
cient for the outcome to occur. In such a case, the bi-conditional will be
false and therefore the causal condition would not be causal in the
INUS sense.
A further point, and one that is directly related to the one made
previously, is that there is no inductive expectation with the INUS
model. As was shown before, the Humean model implicitly entails
that, due to the inherent inductive expectation and the implicit con-
sequence of its deductive construction, a causal relation of the type
described will hold in all alike circumstances. This leads, naturally, to
the drawing of causal generalisations which are expressed as law-like
universalisatios:
8xðPx ! QxÞ
Where for every single case of x if it has the property P then it will have
the property Q; this is obviously open to refutation by means of a single
counterexample. However, as the INUS model does not contain this
inductive expectation there is no inherent need to extrapolate to an
universalisation. In fact, with the INUS model a causal conclusion
would only take the form of an existential conclusion:
9x½ðPx ! QxÞ $ ðPx ^ ðRx v SxÞÞ ! QxÞ
Where there is at least one case of x such that if it has the property P then it
will have the property Q if and only if it is the case that if it has the property
P and the property R or S then it will have the property Q. The advantage
here is that not only does it avoid the problem of refutation by a single
counterexample, but it also means that it can capture singular causal events
as well as more complex general causal relations. Another benefit here is
that as the model does not rely on a conjunction of INUS causal elements,
if any of the factors that are included in the field F are discovered after all
not to be causal in the manner described by the model then they can be
removed without damaging the overall schema.
This allows for theorising which can be adaptive to data – as new data
or new analysis exposes that a factor is or is not causal in the INUS sense
then it can be included/removed from F without drastically damaging
the theory. Let us imagine that Professor B, an open-minded scientist,
had with advent of new data discovered that neuropsychological anoma-
lies (N) were not in fact causal in the INUS sense as first thought (but yet
inimical parenting and goal deprivation (G) were still). Then it becomes
a simple matter (though not so simple in terms of retracting the original

paper and writing up a new one) of removing N from the set of condi-
tions that go to make up the causal field (F). In such a case the formal
model would now be:
½ððIx ; GxÞ ! CxÞ $ ðððIx; GxÞ ^ ðEx v SxÞÞ ! CxÞ
If Professor B had also found, upon closer inspection of the data, that
there was a labelling effect (L) in play that was also an insufficient but
necessary condition then this too could be added into the theory without
in any way damaging the overall schema being employed to construct
theory. What this then allows for is not only an adaptive theory for under-
standing and describing crime but also for one to weather the rigours of
theory evolution. The very logics employed allow the theorist to adapt
theory without damaging it.
The third benefit of this model is that it forces a distinction between
those elements/factors that are causal and those that are correlational. It
does this by explicitly defining what constitutes a causal factor. A factor,
some event/phenomena, is only to be counted as causal if and only if it can
be explained in terms of INUS conditions (Mackie 1974). If the factor or
element does not satisfy this stricture, then it cannot be considered as causal
and if it is not considered as causal then it must only be either correlational
or auxiliary. Therefore, with the INUS model any factor that is to be placed
within the causal field set has to be causal in the INUS sense, any other
factor is perceived as correlational and is thus placed within the set of
conditions that are unnecessary but sufficient, even minimally sufficient,
for the event/phenomena under investigation to actually occur.
Lastly, this model has an advantage over the Humean model as it most
closely resembles the type of causal reasoning that is being implicitly
adopted by those theorists who seek to explain crime causation through
multidimensional explanations. For example, if we examine Wikström’s
(2004) Situational Action theory of crime causation we see that his line of
reasoning closely resembles, and in fact can be expressed in terms of, the
INUS model.
The theory proposed by Wikström in the early 2000s is somewhat rich
and involves a complex multidimensional, and to a degree multidirec-
tional, integrated form of crime explanation. Part of this complexity
arises from the identification of various important factors that must be
included in a theory of crime causation. Not least of all an explicit

definition of crime, which he defines as an act, or procession of acts,
that involve the breaking of moral rules which are enshrined in a coun-
try’s system of laws (Wikström 2004). Crime itself is a rather difficult
philosophical issue or set of issues, that many theorists ignore and yet
which is central to any theory of crime causation. Further to this is the
inclusion of an explanation of action theory and the role of choice and
habituation in it.
However, regardless of the complexity of this Situational Action
theory of crime causation it can, in a succinct and quite simplified way,
be summed up as the following: a crime is an act of moral rule breaking.
The conjunction of an individual’s psycho-physical characteristics and
experiences not only constructs the manner in which they perceive their
action alternatives but also determines the way in which they process and
evaluate environmental input. A person’s action, or inaction, is a ‘con-
sequence of how that person perceives their action alternatives and makes
their decisions when confronted with the particularities of a setting’
(Wikström 2006: 76). Therefore, if an individual commits a crime it is
a result of the decision, determined by their perception of their action
alternatives, made when confronted by a particular environmental
setting.
Now that there is some understanding of Wikström’s theory let us see
if it does fit the INUS model. There are a number of central notions to
this theory:
1. Crime (an act of moral rule breaking)

2. The perception of action alternatives
3. Action alternatives determined by the conjunction of an individual’s
psycho-physical characteristics and experiences
4. Particular settings
5. Action, or inaction, being determined by the conjunction of the
manner in which an individual perceives their actions alternatives
and the environment
If we take crime (C), moral rule guidance (M), setting (S), the perception
of action alternatives (P), an individual’s psycho-physical characteristics
(I), individual experiences (H), the environment (E) and x as the domain
of people then, though somewhat complicated, we can formalise the
theory in the following way:
½ðPx ! CxÞ $ ðððPx ≡ ðIx ^ Hx ^ ExÞÞ ^ ðE v Sx; Mx . . .ÞÞ ! CxÞ
Where it will read: a person’s perception of their action alternatives will

cause them to choose to commit a crime if and only if it is the case that
given if that perception of action alternatives (which arises from the
conjunction of their individual psycho-physical characteristics and
experiences and their environment) coupled with the environment or a
set of conditions such as the particularities of the setting, their moral rule
guidance, etc., will they then choose to commit a crime.
As can be seen, the INUS model closely resembles the theorising
undertaken by Wikström in this instance. As such, his theory avoids
many of the aforementioned problems of earlier theories due to it not
being constrained by the conditionality of the Humean model. Here the
action alternatives are insufficient but necessary and the environment,
setting and moral rule guidance are conditions that are unnecessary but
sufficient for criminality to occur.
As the Mackiean INUS model avoids the most damaging of the
problems that beset Humean causation, it seems that it is best suited to
modern criminological/social theorising. However, the model does have
its own set of problems that need to be taken into account before it is
explicitly adopted into practice. Perhaps the most serious of these pro-
blems relates to a criticism, made by Matza, upon the notion of causality
and causal elements in general (Matza 1969). He notes that one of the
problems with causal explanations is knowing how wide one must cast
the net when considering causal factors to some social phenomena.
Whether you talk of causal chains (as with the Humean model) or causal
fields (as with the Mackiean model) the problem remains as to where
does one draw the boundaries when counting those factors that have
played some causal role in the phenomena under investigation? A further
problem, especially for the INUS model, is in drawing the boundaries of
what conditions are not causal but are to be considered as those which
are unnecessary but sufficient, even minimally sufficient, for the event/
phenomena under investigation to actually occur. These are problems
that beset all scientific investigation but are particularly taxing for the
social sciences due to the nature and density of social interactions and
relations and the cultural-societal matrix within which they take place.
For instance, if you’re going to include inimical parenting as a factor in
criminality how far back do you take that? Grandparents? Great grand-
parents? Great great great grandparents, etc.? Also, if you’re going to
discuss this in terms of parenting skills do you look beyond the family to
the local community and the nation state? Do we look at how the move
from rural to industrialised and urban life changed family dynamics? Do
we then need to include the societal changes that came about from the
Enlightenment . . . Unfortunately, there is no easy or ready answer to
these problems.
A further problem for this model, as well as any other explicitly causal
explanation, is that of missed causes. The problem can best be described as
when some factor has a causal relation to some phenomena but that this
relationship is either not known or cannot be explained (Psillos 2002).
This problem arises in this instance due to the manner in which the model
differentiates between causes and correlates. Causes are only those condi-
tions that that can be explained in terms of INUS conditions and all other
factors are not to be included on the causal field. This can result in a
situation where if some factor cannot be described in terms of the model,
where the causal relation has to be explicitly known, then the factor is
discarded as causal. This can then lead to incomplete theorising and
theories. Imagine if in Professor B’s world one suspected causal factor of
crime was degrees of competitiveness instilled in children between the ages
of 3–7 but yet Professor B and her team could not measure this in any
accurate way and thus could not describe this factor adequately in terms of
INUS conditions? Strictly, this would need to be excised from the causal
field but that would be problematic for Professor B’s theory of crime
causation.
A further general point and one identified by both Hausman (1986)
and Salmon (1994), expanding the point made by Sextus Empiricus, is
that regardless of logical trickery and sophisticated argumentation to over-
come it there is actually the problem that causal models, whether they be
deductive or not, may inherently involve circularity. The reason for this
concern is that the conclusions being drawn are necessarily just a reitera-
tion of the antecedent conditions: x causes y if it is the case that y will not
occur without x. Even in the INUS model we are stating, in effect, that
criminality occurs if and only if the conditions that cause criminality occur.
It seems to be the case that it is impossible to iterate a theory of
cause without repeating the very premises that you rely upon in order to
establish the cause. However, there are two forms of circularity: one that
is vicious and one that is benign. The one described here is benign.
As Hahn (2011) notes, though some circularity is rife in the social sciences
(especially psychology), the form this tends to take is not the of the vicious
kind where it is the case that: a because of b and b because of a – here the
definiendum, that which is to be defined, (b) is inevitably involved in the
construction of the definiens (a), that which is defining. What is more
likely is the form that has been shown previously where the consequent
involves a reiteration of the antecedent rather than involving the definien-
dum (crime) in the premises. While this in and of itself does not necessarily
render the argument or theory invalid (see previous discussion), it never-
theless poses the problem that theory can be, on close inspection, specta-
cularly uninformative.
A further problem for the INUS model is the scientific notion of
falsifiability and an accusation of pseudo-science. Originating from
Popper’s (1968) notion that in order for a scientific theory to be consid-
ered actually science it must be able to be falsified – in principle. A scientific
theory generates hypotheses which can be empirically tested – gravity will
have x effect on falling bodies in y space or potassium will react with water at
z temperature in x ways, etc. For falsificationists, this is what distinguishes
true science from non- or pseudo-science. If you cannot test a theory to see
whether or not it holds some truth (at best you can only ever corroborate a
theory at a given point in time rather than prove one to be completely/
ultimately true – whereas you can disprove one) then you are in the realms
of faith rather than science. Some of the theorising that has emerged from
the special sciences, especially psychology, have been singled out for criti-
cism that theories constructed within these discourses are not disprovable
and thus can only ever by pseudo-scientific (i.e., have the façade of a science
but not the guts or bones of one). There are problems with the notion of
falsification (see Musgrave 1973; Rosenberg 2012) but for the purpose of
this introduction let us assume that the notion is relevant here. What is
interesting is that because the INUS model is somewhat retrospective in
some construction (i.e., you are analysing phenomena that have already
happened) as well as being prospectively adaptive to data change an accusa-
tion could be made that constructing a theory in this way would result in it
being impossible to falsify. Thus, rendering your theory mere pseudo-
science. This is however an unfair accusation – if you could show, through
empirical analysis, that a factor proposed as an INUS cause was in fact not
an INUS cause (Professor B and the neuropsychological condition N, for
instance) then you have falsified that iteration of the theory but not
necessarily the theory as a whole. To do that you would have to test each
INUS condition. This may be time consuming but it is, nevertheless,

possible. Which means that the theory could be falsified in principle and
as such avoids this problem too.
Lastly, there is still the thorny issue of the cause/correlate problem.
This model seemingly solves this problem but it does so by explicitly
dictating what can be counted as causal within the terms of the model as
opposed to actually distinguishing between why some element is causal
and why some other element is a correlate. Instead of solving the cause/
correlate problem, what the INUS conditions do is in fact shift the goal-
posts of the problem by redefining the usage and meaning of the term
‘cause’. However, though this may in fact be the case it could also be
possible that in order for the cause/correlate problem to be solved at all
such a move is, in fact, necessary. That is a discussion for a different book.
Also, even if the solution has come about because of goal shifting the fact
of the matter is that this model does force a distinction whereas other
models cannot.
This chapter has been concerned with looking at a suitable causal
structure/model that would be sophisticated enough for the purposes of
contemporary, complex causal theorising. It was argued that the sug-
gested mechanism, an adaptation of Mackie’s INUS model, is an adequate
causal model for modern criminology theorising because not only does it
solve many of the problems that haunted the older Humean model but
also it more closely reflected the kind of causal reasoning employed by
today’s theorists.
CHAPTER 8
Consequences
Abstract This brief chapter is concerned with a brief exploration of what

some of the possible consequences would be for explicitly adopting this
form of causal reasoning into criminological theorising. As such, the
chapter revisits the four problems identified in Chap. 2 and explains how
the INUS model either avoids or overcomes these problems. This will also
involve a brief examination of the problem posed by the Rosenthal Effect
(researcher bias) in scientific research and theory construction.
Keywords Bias INUS Reasoning Rosenthal Theorising
The last chapter was concerned with proposing a new model of causation
that would be sophisticated enough for the purposes of contemporary,
complex causal theorising in criminology. It was argued that this new
model of causal reasoning would be the Mackiean INUS conditions.
This chapter is concerned with a brief exploration of what some of the
possible consequences would be for explicitly adopting this form of causal
reasoning into criminological theorising.
However, before re-exploring these issues I just want to clarify a point.
I am not arguing here that criminological theorists need to adopt this
INUS model of causation for their work and output. That would be, on
the one hand, arrogant in the extreme and, on the other, impractical for
some. No, the purpose of introducing and exploring this particular form
of causal reasoning was to expose how this particular model, rather than

of Causation, DOI 10.1007/978-3-319-47446-5_8
the folk attributive or axiomatic models that are more widely utilised, both
exposes and solves some of the problems that have been highlighted in this
book. The purpose here is to familiarise the reader with the problems and
complexity of causal reasoning and what is necessary in order to address
some of these problems rather than impose any particular model. These
are issues that are useful to think about whether you are an undergraduate
studying the history of criminological thought or an advanced researcher
developing theories designed to explain the causes of crime.
With that thought in mind let us re-examine, in light of the arguments
made with regard to the INUS model, the four arguments presented in
Chap. 2 that focused on why a poor understanding of causal structures and
mechanisms was a problem for criminological theorising. Those four
arguments were: (1) an inadequate understanding of causal mechanisms
can result in the failure of posited causal explanations; (2) the adoption of
a simplistic causal framework renders a theory open to damaging criticism;
(3) Confusion arises between causes and correlates; and (4) it can
adversely affect the manner in which criminologists approach specific
explananda.
The first of these arguments was that an inadequate understanding of causal
mechanisms could result in the failure of posited causal explanations in at least
two ways. The first was that common practice is usually dominated, or
informed, by a natural or intuitive notion of causality, causal relations and
non-spurious connections (Marini and Singer 1988). However, the adoption
of such an ‘intuitive’ notion of causation can result in the situation where a
theory of causation does not actually explain, account or capture the reality of
the causal relationships discussed and will thus will fail the test of aetiological
extensivity – the test that allows us to examine the explanatory power of a
given theory. This is primarily due to the inadequate integration of levels of
explanation that is possible with such a simplistic notion of causation.
However, the INUS model, being a more sophisticated tool for causal expla-
nation, does allow for a far greater integration of levels of explanation and is
therefore, more capable of capturing the reality of multidimensional and
multidirectional causal relationships. This model also allows for a more
dynamic theory that can adapt to changes in data or analyses as the set of
causal conditions are not dependent upon them being necessarily conjoined.
Each causal condition is causal in and of itself in the INUS sense. As such
causal conditions can be included or discarded from the theory as needs be.
The second way in which an inadequate understanding of causal
mechanisms could result in the failure of posited causal explanations is
8 CONSEQUENCES 73
that it can result in a situation where the line of reasoning/argumentation

adopted is not compatible with the causal conclusions drawn. This is
especially true where a theory/explanation is constructed through the
deductive model, which results in conclusions that are implicitly universal,
yet entails a probabilistic conclusion. As was highlighted, one type of
explanation cannot result in the others kind of conclusion and to confuse
the two can result in false conclusions (Salmon 1984). The INUS model
solves this problem as, firstly, it does not involve the inductive expectation
of the Humean deductive model, where any given conditional relationship
is expected to hold in all alike circumstances, so the conclusions drawn
from it are not necessarily universal. Secondly, by explicitly adopting the
INUS model, or something very similar, as the causal framework of a
theory you avoid the very situation where this problem can arise. You are
not tied into making law like conclusions or statements due to the logical
construction of your modelling; instead you are free to make causal
pronouncements on a singular event, or a given set of events, as well as,
if the data warrants it, all alike events.
The second argument was that the adoption of a simplistic causal frame-
work, one such as the Humean model based upon the if p then q condi-
tional, renders a theory open to damaging criticism especially with regard
to the two forms of deviant causal chain identified in Chap. 3. Firstly, there
is the problem of the Type 1 deviant chain, a causal if/then claim where the
antecedent is true but the consequent false, which is where direct refutation
can occur due to a single counterexample. Secondly, where Type 2 deviant
chains, when the antecedent conditions are false but the consequent is true,
occur it is usually because the antecedent set involves the conjunction of all
the conditions in that set. In such a case it is difficult to establish which, if
not all or indeed any, of the antecedent conditions are actually causal
or not.
However, to take the first argument because the INUS model is
expressed in terms of the bi-conditional, not the if/then construction, it
cannot be the case that one half of the expression is true whilst the other
false. This means that any theory validly expressed in the terms of the
INUS model is not vulnerable to the Type 1 deviant causal chain. Also,
with regard to the second argument, as the INUS model does not rely
upon the conjunction of the conditions/factors to be causal, as the each of
the factors is causal, in terms of the INUS model, in and of themselves,
then it explicitly avoids the problem of Type 2 deviant chains by virtue of
the construction of the model.
The third argument concerned the perennial problem of the confusion

that arises between those phenomena, events or conditions which are the
actual cause of some other phenomenon, event or condition and those that
are merely, or even coincidently, correlated with it. There were two issues
pertinent to this problem. The first was that this confusion most readily
occurs when the adoption of a simplistic causal framework or mechanism
leads to the reasoning error traditionally known as ‘post hoc ergo propter hoc’
(Warburton 2000). The second issue was that just because two, or more,
sorts of events are correlated, that is whenever one is found the other is
usually also found, this does not mean that there is necessarily a causal
connection between them. The problem is that a simplistic causal frame-
work, or mechanism, does not allow for the distinction between what is a
causal element and one that is merely a correlate.
Neither of these two issues are problems for the INUS model. As has
been shown, this particular form of causal reasoning explicitly defines what
constitutes a causal factor. A factor, some event/phenomena, is only to be
counted as causal if and only if it can be explained in terms of INUS
conditions (Mackie 1974). If the factor or element does not satisfy this
stricture then it cannot be considered as causal and if it is not considered as
causal then it must only be either a correlated, or auxiliary, condition (Kim
1971). If the INUS conditions were to be explicitly adopted as the frame-
work for a theory then, in terms of the model, the cause/correlate pro-
blem would cease to be a problem as the distinction between the two types
of condition is imposed by the model.
The final issue was with regard to the possible adverse impact that the
previous three arguments could have on the manner in which a researcher
will approach their specific explananda and analysis. Firstly, the adoption
of a simplistic notion of causation can result in the situation where a theory
of causation does not actually explain, account or capture the reality of the
causal relationships. Also, that it can result in a situation where the line of
reasoning/argumentation adopted is not compatible with the causal con-
clusions drawn. Also, without an appropriate understanding of causal
mechanisms there is no way, as we have seen, that when analysing data
the distinction between causes and correlates can be made. As such there is
the danger of making the ‘post hoc ergo propter hoc’ reasoning error and
drawing causal generalisations that are open to counterexample and refu-
tation. As has been shown all of these problems can be either solved,
avoided or addressed if the Mackiean model of causation was explicitly
adopted as the method of causal reasoning/analysis.
8 CONSEQUENCES 75
However, more damaging than any of these issues is that an adoption

of a simplistic causal model, or even a cursory understanding of the
inherent problems outlined in this book, can even impact on the
manner in which a researcher even constructs their study – potentially
damaging their results before they have even started data collection.
The Rosenthal Effect (Martin 1994) is where the very biases of ontol-
ogy and method can be carried through to results unless the researcher
attempts to design these out.1 For instance, the argument here would
be that if Professor B had designed her study in order to investigate a
neurological cause for crime then the chances are she and her team will
find some evidence to support this hypothesis. This is not because
Professor B is a bad scientist/researcher but because the very question
that was formulated and thus being investigated is itself underpinned by
a morass of assumptions as is the manner in which she has constructed
the research, the theory she has employed, the literature which has
informed her study, the method of data collection utilised and the
protocols of analysis all of which increase the level of knowledge/
confirmation bias in the results. If Professor B is looking for N, then
chances are she will indeed find N. This is a further philosophical
problem relating to science – and not one to be discussed in full in
this book. Nevertheless, it is something that scientists do take seriously
(again if implicitly) and attempt to design out.
In experimental models of science this is normally done by introdu-
cing double blinds and strict random control trials. In more qualitative
fields, this is done by designing out bias and positionality in the first
instance (as much as is possible) and then undergoing analyses of
reflexivity when examining the data. This is all well and good and, in
the main, works for the progression of science. Imagine though, if you
will, that if bias can be carried through to results via question design
and method construction what bias can be carried through if there is a
poor understanding, or the adoption of a simplistic notion, of causa-
tion? How does this thinking impact on what method we even use and
then, how we analyse our data and produce our results? If we as
researchers and scientists think only in if x then y terms we are at
best limiting our own research and at worst imposing an inherent bias.
If we think in these ways we are damaging the very discourses we are
attempting to further.
NOTE
1. There is also something called the Meta-Rosenthal Effect, which is where
every attempt to write out the Rosenthal Effect creates an extra level of
Rosenthal Effect and so on. This however leads to a point of reduction ad
absurdum and thus shall be ignored here.
CHAPTER 9
Conclusion
Abstract This final chapter summarises the argument contained in the

book and offers some advice to readers on how the book should be
utilised. The chapter highlights what this book has tried to do and explains
the threefold aim: firstly, to highlight more directly than has been done so
before in our field how causal reasoning has shaped and influenced crim-
inological thought; secondly, to introduce my learned and esteemed
colleagues, as well as our eager students, to the complexity involved in,
and the problems that exist when, formulating a theory of causation; and,
lastly, how these problems manifest and how they may be overcome by
thinking more clearly about the logic utilised when theorising in this way.
Keywords Causal Conclusion Logic
This book has been concerned with introducing and exploring not only
the issues and problems raised by causation for criminological theorising
but also the consequences of these issues and problems on the history,
present and future of criminological discourse. In order to construct the
argument that lies herein, the book has been divided into three broad
sections, each dealing with a different aspect of the argument.
The first section explored three central themes. The first of these was
the explication of why a poor understanding of causal mechanisms is a
problem for criminological theorising. This involved a number of argu-
ments covering such notions as how an inadequate understanding of

of Causation, DOI 10.1007/978-3-319-47446-5_9
causal mechanisms may cause theories to fail as causal explanations, how

the adoption of a simplistic causal framework renders a theory open to
damaging criticism, how confusion arises between causes and correlates
and how it can adversely affect the manner in which criminologists
approach specific explananda. The second theme identified the implicit
causal mechanism employed in much of the history of criminology as the
Humean chain model and showed why this is an inadequate model of
causation. Lastly, a number of examples, from the history of criminology,
Classical theory, Biological Positivism, Strain theory and Labelling theory,
were explored highlighting both the manner in which the Humean model
has shaped them and the impact upon the validity of those theories by
adopting this inadequate casual mechanism.
The second section focused upon the apparent ‘paradigm’ shift which
has occurred, and continues to occur, within the theoretical discourse of
criminology. It was argued that the reason for this ‘shift’ was the dissa-
tisfaction and rejection of the simplistic and inadequate Humean chain
mechanism implicitly employed in previous theorising. It was also argued
that though this rejection has occurred, and that explanations mentioned
in this context have achieved a more profound level of integration, there
has still neither been any identification nor adoption of an adequate model
of causal explanation.
The third and final section was divided into two parts: the first identi-
fied an adaptation of Mackie’s INUS condition model as an adequate
mechanism of causation. It was shown that this particular model works
by explicitly defining not only that which can be counted as causal but also
how it must fit into the structure of the mechanism. There was also some
explication of not only how this model is more suited to contemporary
integrated forms of theorising but also how the INUS conditions solve
many of the problems, such as the confusion between correlates and
causes, that besets extant theorising. However, as no system is perfect
there was also a brief exploration of some of the criticisms and problems
that this causal explanation may face. Lastly, there was an exploration of
some of the potential consequences of explicitly adopting the INUS
conditions model as the new paradigm of causal explanation. The conclu-
sion thus drawn was that as the INUS model solves the problems that
haunt the old Humean conditional model of explanation and better
reflected contemporary theorising it would not only be fit for criminolo-
gical purposes but also be suitable for explicit adoption as the method of
causal explanation.
9 CONCLUSION 79
So, what does this all mean? Am I arguing that all criminologists
involved in causal theorising should explicitly adopt the INUS model in
order to construct their theory? No, of course not. I would not be so
arrogant. Instead what this book has tried to do is threefold: firstly, to
highlight more directly than has been done so before in our field how
causal reasoning has shaped and influenced criminological thought; sec-
ondly, to introduce my learned and esteemed colleagues, as well as our
eager students, to the complexity involved in, and the problems that exist
when, formulating a theory of causation; and, lastly, how these problems
manifest and how they may be overcome by thinking more clearly about
the logic utilised when theorising in this way. My intention is that this
book may act as a guide to some of my fellow criminologists (and indeed
myself – I do not want to have to spend too much time in the future
revisiting, and wrestling with, these issues!) when thinking about crime
and its causes. The book, I am hoping, will also act as a guide for students
who are interested in our field and enable them to look at the history of
criminology and understand some of the shifts that have occurred and the
reasons for them. I am hoping that this book will give some context to
what can be a bewildering array, and progression of, theory.
Thank you.
APPENDIX A
FORMAL LOGIC AND TRUTH TABLES

What follows is a brief description of the logical rules governing validity
and the truth-values of the atomic declarative schemas: the conjunction,
the disjunction, the material conditional and the bi-conditional. Also,
there is a brief description of the universal and existential quantifiers.
This summary is informed by Wittgenstein’s (1921) infamously complex
Tractatus. What follows is a very basic explication of his truth tables in
order to provide the reader with a usable guide.
Firstly, logical validity is a matter of when:
An inference is valid if it is not possible for the conclusion to be false and (all)
the premises true at the same time.
Another way of thinking about this impossibility is that in a valid inference,

you would contradict yourself if you held that the conclusion was false and all
the premises true. However, in the case of an invalid inference you might be
wrong if you asserted that the conclusion was false while accepting the pre-
mises as true but you would not, in this case, be contradicting yourself. A good
way to think about what makes an inference invalid is that it is invalid if it is
possible for the conclusion to be false even though the premises are all true.
Following is a description of the rules governing and the truth tables for
the atomic declarative schemas: the conjunction, the disjunction, the
material conditional and the bi-conditional.

of Causation, DOI 10.1007/978-3-319-47446-5
82 APPENDIX A
Conjunction
The conjunction is a connective that denotes a relationship between two
atomic sentences. The most common use of the conjunction is the use of
‘and’. However, synonymous terms are: ‘coupled with’ and ‘conjoined’.
Validity does not depend at all on what the individual sentences are about.
The truth-value of the compound sentence can be determined once the
truth-values of the components are known. Any conjunction is true only if
both conjuncts are true, and is false otherwise (that is, the conjunction is
false if either conjunct is false [or both are]). If p and q are the individual
atomic sentences, then the truth-value of the sentence ‘p and q’ is ‘true’ if
and only if the truth-values of p and q are both ‘true’. Ludwig
Wittgenstein developed a graphic device, known as a truth table, to
demonstrate this. The symbol ‘^’ means ‘and’.
Truth Table for Conjunction
p q p^q
T T T
T F F
F T F
F F F
Disjunction
The disjunction is a compound sentence using the either/or construction,
for example either x is the case or y is the case. Any compound disjunction
is true in any circumstance where at least one of the atomic sentences is
true. The disjunction is only false if both atomic sentences are false.
A sentence such as either p or q will be true in all circumstances where
one of the atomic sentences is true. The shorthand symbol for ‘or’ in this
inclusive sense is ‘v’. So we have the following:
Truth Table for Disjunction
p q pvq
T T T
T F T
F T T
F F F
APPENDIX A 83
Conditionals
Another way in ordinary language of making a compound sentence out of
simple atomic ones is by the ‘if/then’ construction. This is called a condi-
tional sentence and, just to have some handy terminology, the sentence after
the ‘if’ is the antecedent of the conditional and the sentence after the ‘then’ is
called the consequent of the conditional. The truth of this conditional sen-
tence too is dependent on the truth-values of its components. So, symbolis-
ing any sentence of the form ‘if p then q’ as ‘p → q’, we have the following:
Truth Table for the Conditional
p q p→q
T T T
T F F
F T T
F F T
There are four lines in this truth table corresponding to each of the
different possible combinations of truth-values of the conjuncts. The final
column gives the overall truth-value of the compound for the correspond-
ing truth-values of the components.
Bi-conditionals
The final way of compounding atomic sentences that we shall consider is
found more often perhaps in scientific and mathematical contexts than in
ordinary discourse, but – maybe because of this – is another straightfor-
ward case like conjunction. This involves connecting sentences using the
phrase ‘if and only if’. (Synonymous phrases to ‘if and only if’ are ‘exactly
when’ or ‘in case of’.) The ‘if and only if’ connective (often abbreviated to
‘iff’) and symbolized as ‘↔’ is again clearly truth-functional: the truth-
value of the compound p ↔ q is dependent on the truth-values of p and q.
In fact, p ↔ q is true whenever p and q have the same truth-value and false
whenever they have different truth-values.
84 APPENDIX A
Truth Table for the Bi-conditional

p q p↔q
T T T
T F F
F T F
F F T
THE UNIVERSAL AND EXISTENTIAL QUANTIFIERS

The notions of quantifiers are taken from First Order Predicate Logic and
are designed to denote the extent the declarative statements capture the
elements within the domain referred to by the sentence. The domain,
usually represented by the indicator x, is the subject of which the pre-
dicates in the sentence relate.
The Universal Quantifier

The universal quantifier, symbolised by ∀x, represents every element in the
referred domain. As such a sentence that is quantified over by the universal
will read: for every instance of, or for all, x such and such is the case.
The Existential Quantifier

The existential quantifier, symbolised by ∃x, represents the notion that
there is at least one instance within the domain. As such a sentence that is
quantified over by the existential will read: there is at least one instance of x
where such and such is the case.
Logical Equivalence
Logical equivalence, represented by the symbol ≡, refers to a situation
where one or a set of atomic compound sentence[s] is logically equivalent
to some other atomic compound sentence.
REFERENCES
Adams, D. (1987). Dirk Gently’s Holistic Detective Agency. London: William

Heinemann Ltd.
Agnew, R. (1995). Determinism, Indeterminism and Crime: An Empirical
Exploration. Criminology, 33(1), 83–109.
Agnew, R. (2000). Sources of Criminality: Strain and Subcultural Theories. In
J. F. Sheley (ed.), Criminology: A Contemporary Handbook (3rd edn.,
pp. 349–371). Belmont, CA: Wadsworth.
Anderson, C. (2008). The End of Theory: The Data Deluge makes the Scientific
Method Obsolete. Wired Magazine, http://www.uvm.edu/~cmplxsys/word
press/wp-content/uploads/reading-group/pdfs/2008/anderson2008.pdf.
Archer, M. S. (2007). The Ontological Status of Subjectivity: The Missing Link
Between Structure and Agency. In C. Lawson, J. Latsis, and N. Martins (eds.),
Contributions to Social Ontology. Routledge.
Aristotle. (1999). Nicomachean Ethics. Translated by T Irwin. Cambridge:
Hackett Publishing Company Ltd.
Barak, G. (1998). Integrating Criminologies. Boston: Allyn and Bacon.
Beauchamp, T. L. (1974). Philosophical Problems of Causation. Encino CA:
Dickenson.
Beauchamp, T. L. (1999). Introduction. In T. L. Beauchamp (ed.), David Hume: An
Enquiry Concerning Human Understanding (pp. 7–61). Oxford University Press.
Beauchamp, T. L. and Rosenberg, A. (1981). Hume and the Problem of Causation.
Clarendon: New York and Oxford.
Beccaria, C. (1775). An Essay on Crimes and Punishments. Translated by David
Young (1986). Cambridge: Hackett Publishing.
Becker, H. S. (1963). Outsiders: Studies in the Sociology of Deviance. New York:
The Free Press.

of Causation, DOI 10.1007/978-3-319-47446-5
86 REFERENCES
Bernard, T. J. and Snipes, J. B. (1996). Theoretical Integration in Criminology. In

M. Tonry (ed.), Crime and Justice (Vol. 20, pp. 301–348). Chicago: University
of Chicago Press.
Blackburn, S. (1990). Hume and Thick Connexions. Philosophy and
Phenomenological Research, 50(Supplement), 237–250.
Brush, S. G. (1996). Dynamics of Theory Change in the Social Sciences: Relative
Deprivation and Collective Violence. Sage Publications Inc. The Journal of
Conflict Resolution, 40(4), 523–545.
Burgess, J. P. (1998). Occam’s Razor and Scientific Method. In M Schirn (ed.),
The Philosophy of Mathematics Today. Oxford: Clarendon Press.
Burke, R. H. (2001). An Introduction to Criminological Theory. London: Willan
Publishing.
Burton Jr., V. S., Cullen F. T., Evans, D. T., and Dunaway, G. R. (1994).
Reconsidering Strain Theory: Operationalisation, Rival Theories and Adult
Criminality. The Journal of Quantitative Criminology, 10(3), 213–239.
Chalmers, A. F. (2013). What Is This Thing Called Science? (4th edn.).
Cambridge: Hackett Publishing Company.
Clark, A. (1998). Twisted Tales: Causal Complexity and Cognitive Scientific
Explanation. In F. Keil and R. Wilson (eds.), Explanation and Cognition.
Cambridge: The MIT Press.
Cohen, A. K. 1955. Delinquent Boys: The Culture of the Gang. New York: The Free
Press a division of Simon & Schuster.
Cohen, L. E. and Felson, M. (1979). Social Change and Crime Rate Trends:
A Routine Activity Approach. American Sociological Review, 44, 588–608.
Conan Doyle, A. (1890). The Sign of Four. London: Spencer Blackett.
Cowls, J. and Schroeder, R. (2015). Causation, Correlation and Big Data in Social
Science Research. Policy and Internet, 7(4), 447–472.
Cullen, F. T. and Agnew, R. (2006). Criminological Theory Past to Present:
Essential Readings. (3rd edn.). Los Angeles: Roxbury Publishing Company.
Cummins D. D. (1995). Naive Theories and Causal Deduction. Memory &
Cognition, 23(5), 646–658.
Davidson, D. (1994). Actions, Reasons and Causes. In M. Martin and L. C.
McIntyre (eds.), Readings in the Philosophy of Social Science (pp. 675–686).
Cambridge: MIT Press.
Ehring, D. (1997). Causation and Persistence: A Theory of Causation. New York:
Oxford University Press.
Einstadter, W. J. and Henry, S. (2006). Criminological Theory: An Analysis of it
Underlying Assumptions. (2nd edn.). Oxford: Rowman and Littlefield
Publishers Inc.
Empiricus, S. (2000). Sextus Empiricus: Outlines of Scepticism. Translated and
edited by J. Annas and J. Barnes. Cambridge: Cambridge University Press.
REFERENCES 87
Epicurus. (1993). Principal Doctrines’. In E O’Connor (ed. Translation.), The

Essential Epicurus: Letters, Principal Doctrines, Vatican Sayings and Fragments.
New York: Prometheus Books.
Farrington, D. (2005). The Integrated Cognitive Antisocial Potential (ICAP)
Theory. In D. P. Farrington (ed.), Integrated Developmental and Life-Course
Theories of Offending, Advances in Criminological Theory (Vol. 14, pp. 73–92).
New Brunswick, NJ: Transaction.
Gauderis, T. (2013). Modelling Abduction in Science by Means of a Modal
Adaptive Logic. Foundations of Science, 18(4), 611–624.
Gerring, J. (2005). Causation: A Unified Framework for the Social Sciences.
Journal of Theoretical Politics, 17(2), 163–198.
Gibson, M. (2002). Born to Crime: Cesare Lombroso and the Origins of Biological
Criminology. Westport, CT: Praeger.
Garland, D. and Sparks, R. (2000). Criminology, Social Theory and the Challenge
of our Times. British Journal of Criminology, 40(2), 189–204.
Glueck, S. and Glueck, E. (1950). Unravelling Juvenile Delinquency, published
by the Commonwealth Fund.
Gurr, T. R. (1970). Why Men Rebel. New Jersey: Princeton University Publishing.
Habermas, J. (1988). On the Logic of the Social Sciences. Translated by S. W. Nicholson
and J. A. Stark. Oxford: Polity Press.
Harcourt, B. E. (2013). Beccaria’s On Crimes and Punishments: A Mirror on the
History of the Foundations of Modern Criminal Law. In M Dubbers (ed.),
Foundational Texts in Modern Criminal Law. Oxford: Oxford University Press.
Harding, S. G. (1976). Introduction. In S G Harding (ed.), Can Theories be
Refuted? Essays on the Duhem-Quine Thesis. State University of New York:
D Reidel Publishing Company.
Hahn, U. (2011). The Problem of Circularity in Evidence, Argument, and
Explanation. Perspectives on Psychological Science, 6(2), 172–182.
Hausman, D. M. (1986). Causation and Experimentation. American Philosophical
Quarterly, 23(2), 143–154.
Hempel, C. G. (1965). Aspects of Scientific Explanation and Other Essays in the
Philosophy of Science. New York: Free Press New York.
Hirschi, T. (1969). Causes of Delinquency. Berkeley: University of California Press.
Hitchcock, C. (2007). Three Concepts of Causation. Philosophy Compass, (3),
508–516.
Hobbes, T. (1991). Leviathan. In R. Tuck (ed.), Cambridge Texts in the History
of Political Thought. Cambridge: Cambridge University Press.
Hoerl, C., McCormack, T., and Beck, S. R. (2011). Introduction: Understanding
Counterfactuals and Causation. In C. Hoerl, T. McCormack and S. R. Beck (eds.),
Understanding Counterfactuals, Understanding Causation: Issues in Philosophy and
Psychology (pp. 1–15). Oxford: Oxford University Press.
88 REFERENCES
Hollis, M. (1994). The Philosophy of Social Science: An Introduction. Cambridge:

Cambridge University Press.
Honderich, T. ed. (1995). The Oxford Companion to Philosophy. New York:
Howson, C. (1997). Logic with Trees: An Introduction to Symbolic Logic. London
and New York: Routledge.
Hume, D. (1999). An Enquiry Concerning Human Understanding. Tom
L. Beauchamp (ed.). Oxford: Oxford University Press.
Hume, D. (2003). A Treatise of Human Nature. J. P. Wright, R. Stecker, and
G. Fuller (eds.) London: Everyman Paperbacks.
Humphries, P. (2014). The Chances of Explanation: Causal Explanation in the
Social, Medical and Physical Sciences. Princeton: Princeton University Press.
Jacquette, D. (1993). Logical Dimensions of Question-Begging Argument.
American Philosophical Quarterly, 30(4), 317–327.
Kenny, D. A. (1979). Correlation and Causality. New York: A Wiley-Interscience
Publication, Wiley and Sons.
Kim, J. (1971). Causes and Events: Mackie on Causation. The Journal of
Philosophy, 68(14), 426–441.
Kuhn, T. (1962). The Nature and Necessity of Scientific Revolutions. In M. Curd and
J. A. Cover (Eds.), 1998 Philosophy of Science: The Central Issues (pp. 86–101). New
York and London: W. W. Norton.
Kuhn, T. (1977). Objectivity, Value Judgement and Theory Choice. In M. Curd and
J. A. Cover (Eds.), 1998 Philosophy of Science: The Central Issues (pp. 102–118).
New York and London: W. W. Norton.
Laub, J. H. and Sampson, R. J. (1991). The Sutherland-Glueck Debate: On the
Sociology of Criminological Knowledge. American Journal of Sociology, 96,
1402–1440.
Laub, J. H. and Sampson, R. J. (2003). Shared Beginnings, Divergent Lives:
Delinquent Boys to Age 70. Cambridge: Harvard University Press.
Lemert, E. M. (1951). Social Pathology: A Systematic Approach to the Theory of
Sociopathic Behaviour. New York: McGraw-Hill.
Lemert, E. M. (1972). Human Deviance, Social Problems and Social Control.
(2nd edn.). Englewood Cliffs, NJ: Prentice-Hall.
Lombroso, C. (1910). The Criminal Man. New York: G. P. Putnam and Sons.
Mackie, J. L. (1965). Causes and Conditions. American Philosophical Quarterly,
2(4), 245–264.
Mackie, J. L. (1966). The Direction of Causation. The Philosophical Review, 75(4),
441–466.
Mackie, J. L. (1974). The Cement of the Universe: A study of Causation. New York:
Magnani, L. (2001). Abduction, Reason and Science: Processes of Discovery and
Explanation. New York: Kluwer Academic/Plenum Publishers.
REFERENCES 89
Marini, M. M. and Singer, B. (1988). Causality in the Social Sciences. Sociological

Methodology, 18, 347–409.
Martin, M. (1994). The Philosophical Importance of the Rosenthal Effect. In
M Martin and L. C. McIntyre (eds.), Readings in the Philosophy of Social Science
(pp. 585–596). Oxford: MIT Press.
Matza, D. (1969). Becoming Deviant. Englewood Cliffs, NJ: Prentice-Hall.
McIntyre, L. C. (1994). Complexity and Social Scientific Laws. In M. Martin
and L. C. McIntyre (eds.), Readings in the Philosophy of Social Science
(pp. 131–144). Oxford: MIT Press.
Mele, A. R. (2007). Decisions, Intentions, Urges and Free Will: Why Libet has not
shown what he says he has. In J. K. Campbell, M. O’Rourke, and H. Silverstein
(eds.), Causation and Explanation (pp. 241–264). A Bradford Book,
Cambridge: MIT Press.
Merton, R. K. (1938). Social Structure and Anomie. American Sociological
Review, 3(5), 672–682.
Millican, P. (2009). Hume, Causal Realism and Causal Science. Mind, 118(471),
647–712.
Moffitt, T. (1993). Adolescent-Limited and Life-Course-Persistent Antisocial
Behaviour: A Developmental Taxonomy. Psychological Review, 100, 674–701.
Moffitt, T. (2006). A Review of Research on the Taxonomy of Life-Course-
Persistent Versus Adolescent-Limited Antisocial Behaviour. In F. T. Cullen,
J. P. Wright and K. R. Blevins (eds.), Taking Stock: The Status of Criminological
Theory – Advances in Criminological Theory (Vol. 15, pp. 277–311). New
Brunswick, NJ: Transaction Publishers.
Morgan, S. L. and Winship, C. (2015). Counterfactuals and Causal Inference:
Methods and Principles for Social Research. (2nd edn.). New York: Cambridge
University Press.
Musgrave, A. E. (1973). Falsification and its Critics. Studies in Logic and the
Foundations of Mathematics, 74, 393–406.
Pearl, J. (2000). Causality: Models, Reasoning and Inference. Cambridge:
Cambridge University Press.
Plato. (1994). Republic. Translated by R Waterfield. Oxford: Oxford University
Press, Oxford World Classics.
Popper, K. (1968). Conjectures and Refutations: The Growth of Scientific
Knowledge. New York: HarperCollins Publishers.
Psillos, S. (2002). Causation and Explanation. Chesham: Acumen.
Rosenberg, A. (2012). Philosophy of Science: A Contemporary Introduction.
(3rd edn.). London and NY: Routledge.
Russell, B. (1913). On the Notion of Cause. Proceedings of the Aristotelian Society,
13, 1–26.
Salmon, W. C. (1984). Scientific Explanation and the Causal Structure of the
World. Princeton: Princeton University Press.
90 REFERENCES
Salmon, W. C. (1994). Causality without Counterfactuals. Philosophy of Science,

61(2), 297–312.
Sampson, R. J. and Laub, J. H. (1993). Crime in the Making: Pathways and
Turning Points through Life. Cambridge, MA: Harvard University Press.
Schaffer, J. (2000). Causation by Disconnection. Philosophy of Science, 67(2),
285–300.
Sober, E. (1981). The Principle of Parsimony. The British Journal for the Philosophy
of Science, 32(2), 145–156.
Song, J. J. (2013). Causatives and Causation: A Universal-Typological Perspective.
London & New York: Routledge.
Sowel, T. (1995). The Vision of the Anointed: Self-Congratulation as a Basis for
Social Policy. New York: Basic Books.
Steffensmeier, D. and Allan, E. (2000). Looking for Patterns: Gender, Age and
Crime. In Joseph F. Sheley (ed.), Criminology: A Contemporary Handbook
(2nd edn., pp. 84–127). Belmont, CA: Wadsworth.
Strawson, G. (2014). The Secret Connexion: Causation, Realism and David Hume.
(Revised Edn.) Oxford: Oxford University Press.
Suppes, P. (1999). Introduction to Logic. New York: Courier Dover Publications.
Tennant, N. (2005). Rule Circularity and the Justification of Deduction.
The Philosophical Quarterly, 55(221), 625–648.
Walton, D. (2005). Abductive Reasoning. Alabama: The University of Alabama Press.
Warburton, N. (2000). Thinking from A to Z. (2nd edn.). London and New York:
Routledge.
Weisburd, D. and Piquero, A. R. (2008). How Well Do Criminologists Explain Crime?
Statistical Modelling in Published Studies. Crime and Justice, 37(1), 453–502.
Wikström, P.-O. H. (2004). Crime as Alternative: Towards a Cross-Level
Situational Action Theory of Crime Causation. In Beyond Empiricism:
Institutions and Intentions in the Study of Crime J. McCord (ed.). New
Brunswick, NJ: Transaction. Advances in Criminological Theory 13.
Wikström, P.-O. H. (2006). Individuals, Settings and Acts of Crime: Situational
Mechanisms and the Explanation of Crime’. In P.-O. H. Wikström and R. J.
Sampson (eds.), The Explanation of Crime: Context, Mechanisms and
Development (pp. 61–107). Cambridge: Cambridge University Press.
Wilson, J. Q. and Kelling, G. L. (1982). The Police and Neighborhood Safety.
Broken windows. Atlantic Monthly, 127(2). http://www.theatlantic.com/
ideastour/archive/windows.html.
Wittgenstein, L. (1921). Tractatus Logico-Philosophicus. New York: Start
Publishing LLC.
Woodward, J. (2003). Making Things Happen: A Theory of Causal Explanation.
New York: Oxford University Press.
Young, J. (1986). The Failure of Criminology: The Need for Radical Realism. In
R. Matthews and J. Young (eds.), Confronting Crime. London: Sage.
INDEX
A Positivism, 5, 30, 32, 33, 78

Abductive, 9, 25, 26, 57 Blackburn, S., 8, 19n2
Abduction, 21, 25 Bonds, 17, 52
Adams, Douglas, 4 Brush, S. G., 44
Aetiological, 2, 3, 5, 9–12, 58, 72 Burgess, J. P., 26
extensivity, test of, 9, 10, 12, 58, 72 Burke, R. H., 35
Aetiology, 9 Burton jr., V S., 35
Agnew, R., 30, 33, 35, 36, 38
Anderson, C., 11
Anomie, 35 C
Archer, M S., 45 Causal, 2–18, 21–27, 29–30, 32–37,
Argumentation, 3, 5, 6, 9, 10, 35, 37, 40, 43–53, 55–65, 67–68, 70–75,
40, 49, 51, 59, 61, 68, 73, 74 77–79
Aristotle, 30, 32 Causality, 1, 4, 7, 8, 14, 16, 26, 56,
Axiomatic, 16, 22, 47, 48, 59, 63, 72 67, 72
Axioms, 15, 24, 47–49 Causation, 1–19, 21–22, 24, 26,
29–41, 43, 46–49, 51, 53, 56–60,
65–68, 71, 72, 74, 75, 77–79
B Cause and effect, 4
Barak, G, 51 Chalmers, A F., 44
Beauchamp, T L., 13, 15, 16, 26 Circularity, 18, 23, 40, 57, 68–69
and Rosenberg, A., 13, 16 Clark, A., 4
Beccaria, C., 30–32 Classical, 8, 11, 30, 32, 33, 35, 78
Becker, H. S., 38 Classicism, 5
Bernard, T. J., 51 Cohen, A. K., 35–37
and Snipes, J. B., 51 Cohen, L. E., 47, 48
Big Data, 8, 11 and Felson, M., 47, 48
Biological, 5, 30, 32–34, 51, 58, 78 Conan Doyle, A., 25

of Causation, DOI 10.1007/978-3-319-47446-5
92 INDEX
Conditional, 14–18, 22, 23, 32, 61, Type 2, 23, 34, 37, 40, 63, 73
63, 73, 78 Disjunction, 14, 60
material, 14, 16, 17, 23, 63 Duhem-Quine Thesis, 10
Conjunction, 4, 14, 15, 23, 24, 33,
38, 50, 53, 57, 60–64, 66, 67, 73
Constant conjunction, 15, 23, 24, E
50, 53 Ehring, D., 4
Contiguity, in time, 15, 16 Einstadter, W J., 8, 14
Control Theory, 14, 18, 19, 22, 23, and Henry, S., 8, 14
57, 59 Einsteinian, 44
Copernicus, 44 Empiricus, Sextus, 18, 68
Correlates, 5, 6, 8, 10–12, 16, 26–27, Epicurus, 30
56, 68, 70, 72, 74, 78 Explananda, 5, 8, 12, 25, 45, 72, 74, 78
Correlation, 10, 11, 22, 26, 27
Counterfactuals, 56–59
Cowls, J., 8, 11
F
and Schroeder, R., 8, 11
Falsification, 69
Crime, 1–3, 8, 9, 12–19, 25, 26,
Farrington, D., 51
30–38, 45–49, 51, 53, 59, 61,
Folk attributive, 8, 72
65–69, 72, 75, 79
Criminological, 1–3, 5, 7, 8, 13, 14,
19n3, 21, 24, 29–30, 35, 39, 41,
43–53, 55–72, 77–79 G
Criminology, 1–3, 5, 7–12, 14, Gauderis, T., 25
25–26, 29–41, 43–45, 47, 51, 53, Generalisation, 9–10, 12, 16, 22, 23,
55, 56, 59, 70, 71, 78, 79 64, 74
Cullen, F. T., 30, 33, 35, 38 Geocentricity, 44
and Agnew, R., 30, 33, 35, 38 Gerring, J., 3, 12
Cummins D. D., 26 Gibson, M., 32, 33
Glueck, E., 46, 47
Glueck, S., 46, 47
Gurr, T R., 44
D
relative deprivation, 44
Davidson, D., 4
Deductive, 9, 26, 30, 31, 37, 47, 48,
57, 59, 64, 68, 73
Delinquency, 46, 51, 52, 61 H
Determinism, 34, 35 Habermas, J., 24
Developmental, 11, 50, 51, 58 Harcourt, B E., 30
Deviance, 9, 38–40, 46 Harding, S G., 10
Deviant Causal Chains Hausman, D M., 68
Type 1, 22, 26, 32, 37, 40, 49, 58, Heliocentricity, 44
63, 73 Hempel, C. G., 9, 26
INDEX 93
Hirschi, T., 17 and Sampson, R J, 46, 51–53

Hitchcock, C., 8 Lemert, E M., 38, 40
Hobbes, T., 30 Life Course, 5, 38, 49–52
Hoerl, C., 56, 58 Logic, 11, 14, 17, 18, 22, 25, 30, 35,
Hollis, M, 1 39–40, 47, 57, 79, 81, 84
Honderich, T., 11 Lombroso, C., 32–34
Howson, C., 19n1
Humean
chain, 5, 16, 78 M
ontology, 16, 48, 50, 51, 53, Mackie, J. L., 5, 56, 59–62, 65, 67,
56, 57 70, 71, 74, 78
regularity, 5, 13, 15, 22 Magnani, L, 25, 26
Hume, D., 3, 5, 15, 16, 19, 24 Marini, M. M., 1, 4, 8, 12, 14, 18, 26,
Humphries, P, 12 51, 57, 59, 72
and Singer, B., 1, 4, 8, 12, 14, 18,
26, 51, 57, 59, 72
I Martin, M., 59, 75
Inductive, 9, 15–16, 18, 23, 24, 26, Matza, D., 67
49, 53, 64, 73 McIntyre, L. C., 23
Integrated Theories, 56, 78 Mele, A R., 4
INUS Merton, R K., 35
conditions, 6, 55–70, 71, 74, 78 Method, 3, 7, 24, 30, 57–59, 74,
model, 5, 6, 56, 61–74, 78, 79 75, 78
Methodology, 44, 45, 47, 51
Millican, P., 27
J Moffitt, T., 49–53, 58
Jacquette, D., 18 Morgan, S L., 56
and Winship, C., 56
Musgrave, A E., 69
K
Kenny, D. A., 56
N
Kim, J, 74
Newtonian, 44
Knowing, 67
Knowledge, 5, 13, 75
Kuhn, T., 44
paradigm shift, 44 O
Occam’s Razor, 25, 57
L
Labelling Theory, 5, 30, 37, 38, P
40, 78 Paradigm Drift, 5, 43–53
Laub, J H., 46, 51–53 Pearl, J., 26
94 INDEX
Philosophy Schema, 14, 18, 19, 23, 26, 32, 37,

philosophical, 4 40, 48, 64, 65
of science, 3, 4, 44, 56 Science
Plato, 33 natural, 2, 3, 4, 7, 8, 14, 23, 32, 44, 56
Popper, K., 26, 69 social, 2, 3, 7, 23, 25, 45, 56
Positivism, 5, 30, 32, 33, 78 Sherlock Holmes, 25
Positivistic, 34–35 Situational Action Theory, 65, 66
post hoc ergo propter hoc, 10, 12, 74 Sober, E., 25
Priority, in time, 15, 16 Song, J. J., 4
Probabilistic, 9, 26, 73 Sowel, T., 10
Probability, 9, 24, 56, 59 Statistics., 10
Psillos, S., 68 Steffensmeier, D., 11
and Allan, E., 11
Strain Theory, 5, 30, 35, 36, 78
Q Strawson, G., 16
Quantifier Suppes, P., 18
existential, 14, 24, 81, 84
universal, 14, 24, 81, 84
Question Begging, 18–19, 40 T
Tennant, N., 18
Theory construction, 6, 30, 40, 47, 51
R
Rational Choice, 5, 30
Reasoning, 2, 7, 9, 10, 12, 18, 19, 21, U
23–25, 31, 39–40, 46–48, 53, Universalisation, 23–25, 35, 37, 53, 64
56–57, 59, 65, 70–74, 79
Reductio ad absurdum, 76n1
Refutation, 9, 10, 12, 21–27, 32, 34, W
37, 63, 64, 73, 74 Walton, D., 25, 26
Regularity, 5, 13, 15, 16, 18, 19n2, 22 Warburton, N., 10, 11, 74
humean, 5, 13, 16, 22 Weisburd, D., 2, 26
Rosenberg, A., 13, 16, 69 and Piquero, A R., 2, 26
Rosenthal Effect, 59, 75, 76n1 Wikström, P-O. H., 1, 8, 53, 65–67
Routine Activity, 5, 47, 48 Wilson, J.Q., 9
Russell, B., 8 and Kelling, G.L., 9
Wittgenstein, L., 10, 22, 81, 82
S Woodward, J., 9, 12, 26
Salmon, W C., 9, 26, 68, 73
Sampson, R. J., 46, 51–53
and Laub, J. H., 46, 51–53 Y
Schaffer, J., 16 Young, J., 2

An Introduction To Criminological Theory and The Problem of Causation PDFDrive

Uploaded by

Copyright:

Available Formats

You might also like

An Introduction To Criminological Theory and The Problem of Causation PDFDrive

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

An Introduction To Criminological Theory and The Problem of Causation PDFDrive

Uploaded by

Copyright:

Available Formats

An Introduction to Criminological

Theory and the Problem of Causation

ISBN 978-3-319-47445-8 ISBN 978-3-319-47446-5 (eBook)

Library of Congress Control Number: 2016954267

© The Editor(s) (if applicable) and The Author(s) 2016

Printed on acid-free paper

This Palgrave Macmillan imprint is published by Springer Nature

2 Criminology and the Problem of Causation 7

3 Humean Causation and Crime Theory 13

4 Deviant Causal Chains, Refutation and Other Problems 21

5 Humean Causation and the History of Criminology 29

6 Paradigm Drift and Criminological Theory 43

7 INUS Conditions and Criminological Theory 55

Keywords Aetiology Causation Criminology Theory

A spectre has haunted the history of theoretical criminology. As Wikström

© The Author(s) 2016 1

it is that we ought to do about the issue of crime. However, there is

Criminology and the Problem of Causation

Abstract The purpose of this chapter is to establish the problem with

Keywords Cause Correlation Criminological Problems

© The Author(s) 2016 7

supply answers to the following question: why is a poor understanding of

example here is the (rightfully) much-maligned Broken Windows theory

generalisation, they can be open to logical refutation through the means of

traditional models of scientiﬁc explanation. Also, if developing treatments

Humean Causation and Crime Theory

Keywords Causation Control Hume Logic Regularity Theory

This chapter is concerned with identifying the causal mechanism, or

© The Author(s) 2016 13

this has been done it shall be shown, by means of a somewhat simpliﬁed

• The conjunction: (p ^ q) which reads: both p and q

It is this that causes Hume’s account to be sometimes referred to as his

whenever a particular event occurs, such as Ball A striking Ball B, if we have

Other than Regularity theory, this notion of causality is often referred

Now that there is some indication of how Humean causation is perceived

Good and healthy family relations = P

Then there are two potential consequences, or outcomes, implicit to this

Argument 2: ð: Px^ : Qx^ :RxÞ ! Cx

Here, the ‘unless’ becomes expressed as a double bi-conditional (Suppes

Deviant Causal Chains, Refutation

Keywords Causal Complexity Deduction Humean Refutation

© The Author(s) 2016 21

Argument 1:2: ðPx^Qx^RxÞ ! Cx

Alternatively, it could be the case that there is an absence of good and

Argument 2:2: ð: Px^: Qx^: RxÞ ! :Cx

burden of deduction with its problematic inherent universalisation. In this

Humean Causation and the History

Abstract This chapter is concerned with tracing Humean causation

Keywords Classicism Causation Humean Labelling Positivism

© The Author(s) 2016 29

the underlying deductive Humean causal framework will be exposed and

necessarily need to be designed so as to ‘prevent the criminal from doing

1. That man is a rational actor

Now, let premise 1 be represented by the variable A and premise 2 be

1. That man is a rational actor

In order to maintain some continuity with the schema outlined previously,

ðAx^: BxÞ ! :Cx

in common with the essentialism of Plato (1994). The essential character-