Professional Documents
Culture Documents
An Introduction To Criminological Theory and The Problem of Causation PDFDrive
An Introduction To Criminological Theory and The Problem of Causation PDFDrive
An Introduction To Criminological Theory and The Problem of Causation PDFDrive
An Introduction
to Criminological
Theory
and the Problem
of Causation
Jason Warr
University of Lincoln
Lincoln, United Kingdom
Cover illustration: Pattern adapted from an Indian cotton print produced in the 19th century
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
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.
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
4 AN INTRODUCTION TO CRIMINOLOGICAL THEORY AND THE PROBLEM OF . . .
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
6 AN INTRODUCTION TO CRIMINOLOGICAL THEORY AND THE PROBLEM OF . . .
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
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
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
12 AN INTRODUCTION TO CRIMINOLOGICAL THEORY AND THE PROBLEM OF . . .
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.
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).
p!q
ðp^q^rÞ ! c
3 HUMEAN CAUSATION AND CRIME THEORY 17
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
18 AN INTRODUCTION TO CRIMINOLOGICAL THEORY AND THE PROBLEM OF . . .
employment then that person will not be a criminal. The second way in
which this argument can be modelled is as follows:
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
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
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.
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 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:
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
24 AN INTRODUCTION TO CRIMINOLOGICAL THEORY AND THE PROBLEM OF . . .
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
4 DEVIANT CAUSAL CHAINS, REFUTATION AND OTHER PROBLEMS 25
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
4 DEVIANT CAUSAL CHAINS, REFUTATION AND OTHER PROBLEMS 27
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
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
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:
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
5 HUMEAN CAUSATION AND THE HISTORY OF CRIMINOLOGY 33
ðEx^Fx^GxÞ ! Cx
ðEx^Fx^:GxÞ ! Cx (1)
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:
½ðHx^IxÞ^ðJx^KÞ ! Cx
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.
½ðHx^IxÞ^ðJx^KÞ ! :Cx
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
5 HUMEAN CAUSATION AND THE HISTORY OF CRIMINOLOGY 39
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
5 HUMEAN CAUSATION AND THE HISTORY OF CRIMINOLOGY 41
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
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
6 PARADIGM DRIFT AND CRIMINOLOGICAL THEORY 47
ðWx^Yx^ZÞ ! C
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
6 PARADIGM DRIFT AND CRIMINOLOGICAL THEORY 51
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
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
58 AN INTRODUCTION TO CRIMINOLOGICAL THEORY AND THE PROBLEM OF . . .
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:
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
7 INUS CONDITIONS AND CRIMINOLOGICAL THEORY 61
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
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
7 INUS CONDITIONS AND CRIMINOLOGICAL THEORY 63
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
64 AN INTRODUCTION TO CRIMINOLOGICAL THEORY AND THE PROBLEM OF . . .
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:
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
7 INUS CONDITIONS AND CRIMINOLOGICAL THEORY 65
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
66 AN INTRODUCTION TO CRIMINOLOGICAL THEORY AND THE PROBLEM OF . . .
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:
7 INUS CONDITIONS AND CRIMINOLOGICAL THEORY 67
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.
7 INUS CONDITIONS AND CRIMINOLOGICAL THEORY 69
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
70 AN INTRODUCTION TO CRIMINOLOGICAL THEORY AND THE PROBLEM OF . . .
Consequences
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
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
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
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
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
An inference is valid if it is not possible for the conclusion to be false and (all)
the premises true at the same time.
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
T T T
T F F
F T F
F F T
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
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
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