Critical Realism and Qualitative Research: An introductory Overview.

Steve Vincent and Joe O’Mahoney

Abstract

This chapter introduces a critical realist approach to qualitative research. Critical realist
research methods are primarily focused on understanding, rather than merely describing,
social reality. To this end, critical realist research is ecumenical in its approach to research,
fitting designs and methods to the problem in hand. Crucially, in addition to deduction and
induction, critical realism uses retroduction and abduction to approximate the causal
mechanisms which generate social events, and distinguishes these events and mechanisms
from our perceptions and talk about them.

Key Words

Critical realism; research methods; abduction; retroduction; causal mechanisms; Bhaskar.

Authors

Steve Vincent is Chair in Work and Organization at Newcastle University Business School.
He has co-authored a book on critical realist research methods, and uses critical realism to
inform his research, which touches on skills and soft-skills at work, self-employed workers,
work organization and the labour process.

Joe O’Mahoney is a Reader in Organisational Studies in Cardiff Business School. He has co-
authored a book on critical realist research methods, and researches realist approaches to
organisation theory, especially in the area of management ideas and worker identities.

Introduction

Many of the methodology textbooks that business and management students are encouraged
to use employ a simple and, in our view, flawed dichotomy between objectivist (positivist,
deductive, and empiricist) approaches, which are typically aligned with quantitative empirical

1
methods, and subjectivist (social constructionist, inductive, and interpretive) approaches,
which are typically aligned with qualitative empirical methods. Whilst this dichotomy
certainly simplifies approaches to research, it is based on the odd assumption that some data
should, a priori, be ruled out simply on the basis of whether it is appropriate and consonant
with the general orientation of the approach. Such an assumption is not only, as we discuss
later, theoretically flawed, but is also out of kilter with the day to day experiences of
individuals: counting can often give us insights into meaning or values (for example when
suicide rates rise), and interpreting sights and sounds can give us greater understanding of an
external and objective reality (think about all the signs drivers respond to when obeying "the
rules of the road", for example).

Critical realist scholars seek to overcome this odd dualism (objectivism or subjectivism) by
distinguishing between ontology (what is real) and epistemology (what we know). Critical
realist scholars assume the existence of an objective (‘intransitive’) world that has powers
and properties that can be more accurately known as a consequence of scientific endeavour
but recognise that knowledge is a subjective, discursively bound (i.e. transitive) and
constantly changing social construction. Crucially, for science, social science, and for the
purposes of this chapter, careful methodological practices form a bridge between our
epistemological knowledge and ontological reality: good research means we can understand
the world better. This marriage of epistemological relativism and ontological realism,
however, presents novel challenges to qualitative social scientists, not least of which is the
question of how researchers make discoveries about any social formation from qualitative
datasets when participants may be unaware of or (consciously or otherwise) misrepresent the
social formations of which they are a part.

In our view, dealing with this problem requires qualitative researchers to think about the
research act in theoretically grounded ways (see also, Edwards et al. 2014), which we attempt
to distil in this chapter. We begin by offering a brief introduction to critical realism, as a
particular philosophy of science, before then sketching out the implications of this position
for theorising with qualitative data. The overall argument is that critical realist researchers
should be led by ontological concerns and, consequentially, they should embrace all
qualitative (and many quantitative) research techniques. However, their approach to data
collection and analysis is likely to be different from those who are committed to alternative
ontological positions because their approaches to theorising, which we sketch below, is also
distinctive.

2
Critical Realist Ontology

The origin of critical realism, as a philosophy of science, is attributed to a series of books by

Roy Bhaskar (1975, 1979, 1994). He argues that the universe, including the social world, is a
stratified and open system of emergent entities. This section breaks this phrase down into
simple parts in the effort to clarify its meaning, so if you are already familiar with this
ontological position you might want to skip forwards to the next section.

Entities, Powers and Systems

Entities are things which ‘make a difference’ in their own right, rather than as mere sums of
their parts (Fleetwood 2005: 199). Molecules, individuals and organizations are all entities,
even though one may be (partially) comprised of others. Entities may be material (e.g. water),
immaterial (e.g. class) or both (e.g. a contract). Bhaskar differentiates between ‘real essences’
which are the ‘structures or constitutions in virtue of which [a] thing or substance tends to
behave the way it does’ (Bhaskar, 2008: 209), and ‘nominal essences’ which are ‘those
properties the manifestation of which are necessary for the thing to be correctly identified as
one of a certain type’ (Bhaskar, 2008: 279). Such a distinction enables a separation of the
essential generative or enabling processes of an entity and its epiphenomenal features. We
may thus identify a book nominally, as paper bound together in a particular way, or in a more
sophisticated way in terms of its language, genre, intended audience or purpose, to realise the
real essence of the specific book observed.

Entities may also be real in different ways and at different levels. For example, the bogeyman
is not materially, but ideationally real: the discourses about him have real effects, not least on
the bedtime activities of children and on the writers of children's fiction. The bogeyman thus
exists as a conceptual/cultural schema in some societies, which affects social and material
events (such as the production of books and toys). Entities are thus organized hierarchically,
in that they exist and have relations at different levels. For example, the entity ‘organization’
is made up from (among other things) people and resources (broadly defined); people are
(partially) constructed of tissue and organs; which are, in turn, made up of cells, and so on.

The relationship between these levels can be usefully conceived in terms of emergence
(Elder-Vass 2010), wherein one level is dependent upon but irreducible to the level below.
For example, we cannot understand the wetness of water in reference to its components of
3
hydrogen and oxygen, nor can we explain the functioning of an organisation with sole
reference to its separate roles, routines and resources. This because each entity has causal
powers that depend on the relational properties of its parts. These relational properties can be
usefully conceived of as essences. An essence is ‘what makes something that thing and not
something else’ (O’Mahoney 2011:726). Water, is H 2O and has the power to soak,
organizations have directors and the power to employ; money has legal status and has the
power to purchase1.

Ideas about the causal powers of entities are important because change often occurs when the
powers of one entity interact with another: water can be heated by fire, teams may elect a
leader, and organizations might be bought by other organizations. Such mechanisms often
transform entities: water might turn into steam, a new leader might change the organisation,
and a purchased organization might be asset-stripped. So, entities emerge and may be
transformed (i.e. manifest with new properties and powers) as a result of the interactions of
entities and causal powers.

We should note here the different ways in which a power may exist: An entity may possess a
power simply due to its properties (a government may have some power to increase
employment); this power may be exercised by the power being triggered (the government
takes on more public sector workers); yet this power may or may not be actualized (i.e. come
to be) because there may be countervailing powers, (e.g. private sector employment may have
collapsed). The social world is full of powers, the actualization of which is often retarded by
other powers within the open systems in which they are located.

It follows that social mechanisms cannot be empirically isolated, as happens in laboratory

experiments. As a consequence, critical realists theorise social mechanisms as ‘laminated
systems’ (Bhaskar 1993; Elder-Vass 2010) ‘whose internal elements are necessarily ‘bonded’
in a multiplicity of structures’ (Bhaskar 1993: 25). Examples might include financialised
capitalism, the Russian legal system, or the British public sector: whilst these cannot be
separated from the rest of society, they comprise systems, mechanisms and entities which can
be usefully considered and conceptualised together.
1
‘Power’ here does not concern the sociological concepts of power (such as that detailed by Lukes, S. 2005. Power: a

radical view. London: Palgrave Macmillan.) but instead the metatheoretical specification of the (potential) of one entity to

act upon another (e.g. heat on water; wages on labour markets; competition on innovation). Although we need to be careful

when using terms like ‘essence’ and ‘power’, it is important to note that without some idea that an entity has one or more

essences or powers it is indistinguishable from anything else.

4
We can now see that any power necessitates the actualisation (manifestation, existence) of at
least one mechanism for the potential exercise of that power. For example, the possessed
power of an employer to sack a worker implies a mechanism (legal termination of a contract)
by which this occurs. Powers thus depend on mechanisms (such as employment laws) that
relate one entity to another. The exercising of powers (the employer's ability to dismiss
employees) results in tendencies (attempts at dismissing employees in the event of
absenteeism) which, in the social world, may be manifest in observed patterns of events
(employees leaving). However, such event regularities are not constant conjunctions
(sometimes workers quit before they are dismissed and the termination of an employment
contract may not be legal), and so observation (employees leaving) does not ‘prove’ one
thing or another. As we have seen powers are often unexercised or unactualised, and so
investigation is required to explain the specific manifestation of events.

This framing is important because it allows a conceptualisation of what could happen, what
should happen and what isn’t happening, which gives the researcher prompts for further
investigation. Thus, when we observe the unexpected (women dominating in highly paid
jobs, global temperatures decreasing) it indicates that there are (perhaps) deficiencies in our
current theorisations or methods which need attention. Moreover, as Bhaskar (2008) argues,
this reading of events also allows a conceptualisation of absences which accords causal
powers: the absence of legislation on worker rights, for example, has real effects (such as on
employer behaviour) which empiricist and constructionist accounts of the world might miss.
However, knowing such things requires a commitment to a depth ontology, which we now
turn to.

The empirical, the actual, the real

In the introduction, we saw how critical realism commits to both ontological realism and
epistemological relativism, yet the picture is more complicated than this, as critical realism,
whilst accepting that actual events do occur, also proposes that these events are caused by
real mechanisms that are often invisible to the researcher. This stratified or ‘depth ontology’
makes a distinction between the ‘empirical’, the 'actual' and the 'real'. The empirical is what
we perceive to be the case: human sensory experiences and perceptions (we observe a driver
approaching a speed camera and slowing down). The ‘actual’ is the events that occur in space
and time, which may be different to what we perceive to be the case (we may assume the

5
driver has slowed down before the speed camera when this is not what happened). The real or
deep is constituted of the mechanisms and structures which generate (and explain) events, a
point which we shall illustrate below.

This stratified conception of causation facilitates a more adequate understanding of how

(material and social) powers, which operate in different times and/or locations and/or at
different hierarchical levels, are related. As reality is ‘multiply determined’, with no single
mechanism determining events (Bhaskar 1975), multiple causes must be teased out from
detailed explorations of the setting. It follows that a key commitment of critical realist
research is that there are deeper levels awaiting discovery. Thus, and beyond direct
observation, it is also possible to posit various other potential mechanisms that may be (in
part or whole) neither manifest nor readily observable, but that still have an effect.

To clarify this critical realist jargon, we might continue with the example of the fixed
roadside speed camera, which routinely slows down the traffic regardless of whether the
camera within it is actually working. A traditional positivist approach might measure the
incidence of new speed cameras (working or not) and the changes in speed of cars, and
undertake a regression analysis to show, for example, that a positive correlation which, they
claim, proves that cameras cause speed retardation. Whilst the numbers here can be (though
are not necessarily) useful, this analysis contains an inadequate or "thin" conceptualisation of
cause which focuses on description rather than explanation. Crucially, it misses out on the
most complex element in the processes: the human. An alternative, post-structural approach
might emphasise the self-disciplining Foucauldian effects of disciplinary surveillance, and
the role such technology has in producing compliant subjects (for example, Fyfe and
Bannister 1996). This perspective can also produce useful insights, but when discourse is
over-emphasised, and the self is seen as constructed, then resistance, social structure, and the
wider historical context can be missed.

A critical realist perspective, however, not only accepts the distinction between the empirical
(the appearance of a speed camera) and the actual (a speed camera with no film), but also
seeks to discover the (deep) causal mechanisms that relate the appearance of the camera with
the person, asking what variety of causal relations must exist in order for the empirical events
to occur. This not only opens up a wider variety of interesting phenomena, such as the mind
of the driver, the powers of the police, and the braking mechanisms of cars, but also helps us
understand why things might change. For example, when the camera gets vandalised because
people think cameras are simply a mechanism for extorting fees, passing drivers may change

6
their behaviour. The critical realist approach, therefore, can place the speed camera in a
number of related, stratified ‘laminated systems’ - which might include both the structural
relations of the citizen, the state and the police, as well as the psychological
institutionalisation of drivers and their responses (and resistance) to disciplinary techniques –
which rely upon the identification of deep causal relations which may be invisible to the
researcher focused merely on actual events.

Critical Realist Research

An Underpinning Ontology

The example above highlights an obvious limitation of critical realism: it is a metaphysical

ontology and (contrary to some writing) does not imply anything about the existence of the
self, society, social structures and so on. To achieve a critical realist account of such things
we need to develop and use what Cruickshank (2003) refers to as ‘domain specific’ theories.
These emerge when general critical realism is drawn upon to develop a framework for
studying a specific empirical domain, and the consequence is that the theories that are
developed, whilst adhering to the principles and language of critical realism, are only
implicitly ontological. For example, those studying labour process theory might (and often
do) use critical realism as an ‘underlabourer’ to build theories concerning the exploitation or
alienation of workers. However, there is nothing in a critical realist ontology which demands
such propositions. Indeed, the epistemological relativism inherent in critical realism means
that one would expect such theories to be at best approximations of reality, not least because
all social theorising involves simplification.

Applied critical realism, therefore, provides an underpinning structure and language to guide
good research by asking what the entities, causal powers, dependencies and relations are, but
it does not specify what these should be. The task of the researcher, then, is to work out a
better and more causally accurate, correct, or reliable explanation for observed patterns of
events via the development of more adequate (and domain specific) accounts of the powers,
entities, mechanisms and relations which created them. However, as we saw with the
example of the speed camera, developing such an explanation can be difficult because the
powers of entities do not have to be present or even exerted in order to have an effect.
7
One consequence of both the ‘domain agnosticism’ and stratified nature of critical realism is
that it is compatible with a wide array of other theoretical (though not ontological) positions.
For example, critical realists are comfortable with the domain level concepts of discourse,
identity or materiality (Elder-Vass 2014, Mutch 2013), though clearly not with the
ontological positions that discourse is everything or that the material is entirely social (e.g.
Barad 2007). This provides critical realists with great opportunities to re-examine competing
ontological positions, which might make important theoretical insights but which are limited
by ontological structures. Such approaches have not only led to many critical realist re-
readings of concepts such as discourse (Fairclough 2005), identity (O'Mahoney 2011) and
socio-materiality (Mutch 2013), but also to claim theorists such as Foucault and Derrida are
more critical realist than post-structuralist (Wright 2004, Pearce and Woodiwiss 2001).

Mechanisms, Contexts, Outcomes

In seeking to understand and explain the mechanisms behind empirical and actual events,
critical realism is primarily interested in causal explanations – moving from the what to the
why. This challenges the researcher, the policy maker or the manager to develop deep
understandings of the worlds in which they inhabit: away from the simplistic certainty of
regression analyses, that might show implementing X ‘causes’ Y, and towards an
understanding of why different contexts, conditions, and aspects of X can cause Y. The key
to this enquiry is the ‘mechanism’ and the events that it produces, but the mechanism in an
open system cannot be isolated from its context, hence Pawson and Tilly’s (1997) equation,
Mechanism + Context = Outcome. In more detail:

 Mechanism: how do the properties of one or more entities affect those of others?

 Context: what conditions are needed for an entity’s causal mechanisms be to triggered?

 Outcomes: what are the empirical manifestations produced by causal mechanisms being
triggered in a given context?

The role of context means that, unlike the image promoted by positivists, there are no clear,
simple or easy answers in the social world (Fleetwood and Hesketh 2010). For example, we
might find that low-waged service workers working in a call centre can be happy when they
work on quality services, have high levels of discretion in their jobs, and positively identify
with the values of their work (Jenkins and Delbridge 2014). The mechanisms that manifested

8
here concerns the relational identifications of the workers with the values of their employer
and their work. However, Jenkins and Delbridge also make clear that the context of this
mechanism was a labour market in which there was limited choice, a family-owned, small
and successful business, and so the identified mechanisms may produce different outcomes in
different contexts or even within the same company at a different time.

CR Research Design

The centrality of identifying mechanisms to explain why things happen means that critical
realists put theory first. Thus, the aim of critical realist research design is to produce
explanations (theories) about the essences (properties) and exercise of transfactual, hidden,
and often universal mechanisms. This takes any potential generalisations from the empirical
to the theoretical, in contrast to the positivist approach where generalizations are only
concerned with an empirical population (Danermark 2002: 77). This distinction is important
because theoretical generalizations are more enduring and can be applied through time and
space. For example, a theoretical generalization that capitalism tends to commodify should
not only be applicable in all forms of capitalism (including future ones), but when one finds
empirical evidence to the contrary it should prompt the researcher to seek reasons which
prevent the mechanism being actualized. In contrast, a generalization that notes only the
empirical instances of capitalist commodification has limited explanatory value because it
simply identifies the empirical event and says little about why is happens, to what extent, and
in which circumstances. Moreover, evidence to the contrary simply modifies the
generalization and does not prompt a pursuit of counter-mechanisms.

The importance of CR emphasising ontological questions (what is X? how does it work?)

over epistemological questions (how can we know X?) means that CR is methodologically
ecumenical. Indeed, many realists would argue that the emergent stratified nature of social
reality means that a wide range of methodological approaches or ‘extended methods’ is
necessary for a richer conceptualisation of the mechanisms at work in the social world. For
this reason, CR scholars embrace a range of qualitative research techniques. Our own edited
collection (Edwards, O'Mahoney and Vincent, 2014), which looked at the implications of CR
for a range of methods within the field of organisation studies, included chapters on discourse
analysis (Sims-Schouten and Riley 2014), grounded theory (Kempster and Perry 2014),
interviewing (Smith and Elger 2014), ethnography (Rees and Gatenby 2014), case studies

9
(Vincent and Wapshott 2013), comparative case methods (Kessler and Bach 2014), action
research (Ram et al. 2014), historical and documentary methods (Mutch 2014). Elsewhere
CR scholars have used observational (Bøllingtoft 2007), diary-based (Næss and Jensen 2002)
and autoethnographical (Botterill 2003) methods.

Table 1 Critical Realist Research Designs (amended from Ackroyd and Karlsson
2014)

Intensive   Extensive

What is the How do context & How do context & What is the
mechanism?
mechanism typically mechanism historically context?
interact? interact?

Detached Case-study Comparative case- Institutional / Surveys

study historical analysis

Engaged Action Intensive realist Barefoot research Extensive

research literature evaluations realist
evaluation

As Ackroyd and Karlsson (2014) have argued, the choice of research design for critical
realists very much depends on the position of one’s research purpose on two dimensions. The
first, relates to the focus of research, and ranges from intensive research to extensive research.
Intensive research prioritises qualitative research designs, such as case-studies or action
research, where the context is known and the mechanism is unknown (for example, why are
workers happy in this factory?). Extensive research examines the effect of different contexts
on a mechanism, for example by using surveys (see also, Sayer 1997). Quantitative data can
be used to generate taxonomies and is useful more where mechanisms are known (or at least
inferred) but the context varies (for example, in what contexts do 'sophisticated' approaches
to HRM result in happier workers?). Between these two extremes are examples of where the
focus of study is on the interaction of mechanisms and contexts. A second scale, concerns the
extent of detachment of the researcher, whether they are simply diagnosing from a distance or
whether they are trying to manipulate the mechanism or context under study. In Table 1, we

10
amend Ackroyd and Karlsson’s categories to re-present eight CR research designs. This
provides us with eight research designs that are described and illustrated briefly below.

Table 2 CR Research Strategies, Explanations and Examples.

Research Explanation Examples

Design

Case-study The most common, and arguably most useful, Gouldner (1964); Beynon
form of CR research. In-depth exploration of (1979) (1973); Burawoy
a case to abduct causal mechanisms from (1979)
their empirical manifestations. Cases may
range from people to companies to whole
economies.

Action Intervention by researchers to explore the Cassell and Johnson

research workings of a mechanism by triggering it or (2006); Friedman and
changing its context. Rogers (2009); Morgan
and Olsen (2008)

Comparative Exploring how similar mechanisms operate in Delbridge (1998); Taylor

case-study different contexts. and Bain (2003);
Kirkpatrick et al. (2005)

Intensive Builds theories to explain what mechanisms Marchal et al. (2012);

realist work in different contexts by reviewing the
literature extant literature.
evaluations.

Institutional / Examining causal sequences over time to Smith and Meiksins

historical explore how mechanisms and contexts (1995); Clark (2012),
analysis interact over time, and the conditions for such Mutch (2007).
interaction.

Barefoot Training / encouraging participants / Lindqvist (1979)

research employees to do their own research.

11
 Surveys Primarily focuses on descriptive statistics Cully et al. (1999)
(e.g. sampling or population data) to illustrate
the empirical consequences or conditions of
mechanisms. Occasionally used to prompt
explanatory investigations, but in conjunction
with other methods.

Extensive Mixed methods: stage one uses qualitative (Kazi et al. 2002, Kazi
realist work to identify causal mechanisms. Stage 2 2003)
evaluation uses statistical techniques to examine how
different contexts affect a mechanism.

CR Data Collection Methods

There is little to say about CR methods, because they do not exist. As detailed earlier, CR has
a highly ecumenical approach to data collection and holds that methodological choices should
‘depend on the nature of the object of study and what one wants to learn about it’ (Sayer
2000: 19). Thus, whilst CR work is often based on case-study research using methods such as
interviews and ethnography, it has also involved observation, focus groups, literature reviews
and surveys. The designs detailed in Table 2 will often share methods, and the distinction
between methods and designs is not always unambiguous.

One caveat that is rarely explored with either research designs or methods is the extent to
which the phenomenon being studied can be isolated. It is true that in the social world this is
probably futile, but many disciplines which are increasingly turning to critical realism, such
as psychology (Ponterotto 2005) the focus of study may be mechanisms which are more
capable of being isolated, at least at one level, than others. This raises interesting questions
about the extent to which experimental design may be appropriate, perhaps as part of a mixed
methods study. As critical realism is still a relatively nascent topic, we hope to see
developments on this and other questions by other researchers.

Analytical method: Abstraction and Retroduction

As far as analysis is concerned, whilst both induction and deduction are used in critical realist
research, the need to move from the empirical to the real (causal mechanisms) means that the

12
emphasis tends to be on abstraction and retroduction. Abstraction, as in "taken from" the
concrete or empirical, involves one-sidedly representing the different elements of a
phenomenon in order to synthesize how they work in combination to affect events (Sayer
1998). The analytical process involves re-description of findings as a causal mechanism or
process between related elements which serves to explain them (see also, O'Mahoney and
Vincent 2014). This involves re-describing that which is observed (interviews, observation,
documents) in terms of (the abstracted) theory in order to describe the sequence of causation,
or the relations between things, that gives rise to observed regularities in the pattern of
events. It involves combining observations, often although not inevitably in tandem with
theory, to produce the most plausible explanation of the mechanisms that caused the events.
For example, a researcher looking at HR practices might interview a group of workers about
their appraisal and seek to explain these data in terms of theory about disciplinary power of
managerial discourses (what is said in the appraisal situation) in the employment relationship,
wherein it is theoretically assumed that the bargaining position of the appraisee (worker) is
weaker than their appraiser (manager). Here, and if the explanation of the mechanisms is
successful, theory and data will be consistently and effectively ‘fitted together’ in such a way
as to render the nature of the mechanism clearer.

Retroduction involves imagining a mechanism, which, if it were real, would account for the
phenomena in question. In other words, it seeks to ascertain what the world (i.e. the broader
context) must be like in order for the phenomena we observe to be as they are and not
otherwise (Sayer 2010). This often involves first identifying patterns over periods of time and
in different contexts to creatively asking ‘what if?’ to identify often hidden causal
mechanisms. For example, extending the example used above, we might observe (1) that
some managerial discourses are more persuasive than others, and (2) that in certain
organizational contexts (opportunity structures, labour markets, etc.) workers are more
inclined to accept the managerial version of ‘the truth’, however this might be constructed.
This, in turn, suggests a number of other causal processes are also at play to affect the
mechanism observed, suggesting the opportunity to understand more about the relationship
between the mechanisms we observed and the contexts in which it operates.

In order to build better explanations of the interconnections between strata which might
explain such variance within a known mechanism, we must either bring in or develop other
theoretical resources. For example, we might use comparative analysis to demonstrate that it
is those appraisers who are able to transpose the dominant cultural tropes of their appraisees

13
into the appraisal meetings who prove to be the most persuasive. As a result, we might
develop a new theory that connects the organizational outcomes with broader cultural
contexts. Alternatively, and drawing on existing theory, we might consider the nature of
opportunity structures within the broader labour market in order to explain why it is that, say,
managerial trainees and those with an interest in opportunities for further technical training
have a greater interest in being complicit with managerial discourses than, say, those with
little opportunity for career advancement. In either case, new lines can be drawn between the
operation of a mechanism, at one level, with the context(s) within which the mechanism
resides.

As different theories emphasize different aspects of mechanisms that are simultaneously

implicated in a pattern of events, retroduction implies a commitment to theoretical pluralism,
at least at the outset of an investigation. Multiple theoretical lenses can be considered for
what they tell us about the various and stratified influences that are affecting the things we
observe.

Abstraction and retroduction can also be used as part of an immanent critique of a competing
position in order to develop theorising in a field. This provides a critique from within a
theoretical position and identifies contradictions, ambiguities or inconsistencies. It seeks the
‘Achilles heel’ (Bhaskar and Hartwig 2010) of an existing theoretical position in order to
identify theoretical weaknesses that require further investigation. For example, Archer
undertakes an immanent critique of conceptions of the social agent within economics and
social-constructionists sociology to demonstrate the need for a new theory about the ways in
which humans are conditioned by, but irreducible to, social norms and structures. This then
spurs Archer’s (2000) investigation and development of a theory of humans which
incorporates the notion of morphogenetic cycle and the subjects ‘inner dialogue’. In
undertaking an immanent critique a critical realist research may seek to ameliorate the flaws
they discover by asking what the theory could be like in order to overcome its inconsistencies
and re-describing the theoretical tools of the position within an alternative (in this case,
critical realist) theory. For example, O’Mahoney (2011) shows that the anti-essentialist
positions of many anti-essentialist scholars are inconsistent with their own practices and
theorising. He uses this flaw to redescribe their main theoretical commitments (for example,
concerning identity and discourse) within a critical realist ontology.

Thus, when abstraction and retroduction succeed they offer a new and often unanticipated
view of things: what was hitherto unobserved becomes the basis of new understanding. By

14
postulating a new view of the object of study in the light of a new or existing theorization, a
successful realist study, therefore, re-conceptualizes the subject and the processes to which it
is connected. Finally, it should be noted that many researchers, for simplicity, treat
abstraction and retroduction as one movement, often from qualitative data to the best theory
that explains the data (Mingers 2006, Ketokivi et al. 2010). As we believe that abstraction
necessitates some form of retroduction, and vice versa, we do not disagree with this
combination.

An exemplar analysis

One qualitative CR study that uses abstraction, retroduction and multiple theoretical lenses is
Vincent (2008). This study wanted to understand more about the way in which public sector
organisations procured complex goods from private sector suppliers, and so it came to focus
on an exemplar case: a ‘strategic partnership’ between Govco (a large and bureaucratic
government department) and Futuretech (a multinational software development specialist).
This "Partnership" became the focus of a case study and a range of qualitative data collection
tools, including interviews, observation and documentary analysis were used to abduct the
Partnership as an inter-organisational mechanisms.

Abstraction was used to work out how the organisation functioned, as a mechanisms in itself.
The interviews indicated that the Partnership was created because Govco was interested in
accessing Futuretech stock of in-house technology and expertise. Futuretech, on the other
hand, wanted to make a profit from the arrangement. A contract was agreed, which stipulated
(amongst other things) that the cost of IT systems development should halve over five years,
with Futuretech benefitting (financially) where performance exceeded these expectations.
The analysis came to focus on how this contract was managed, with subsequent interviews,
observation and documentary analysis focused on this contractual mechanism.

The data indicated that a group of senior IT experts and managers who constituted the
Partnership were employed, by both organisations, to manage the contract. These managers
knew a great deal about the technological possibilities, capabilities, and susceptibilities of
both organisations. They met regularly to define the work Futuretech would deliver and,
subsequently, they were responsible for ensuring that it was delivered to contractual targets.
As such, the Partnership had a good deal of latitude to decide what ‘good performance’
looked like (even if most respondents didn't say so). Some more junior managers admitted

15
that, as there was no easy yardstick with which to measure time spent developing IT systems,
some technologies were much easier to deliver than others. Others hinted that, where
performance dipped below contractual expectations, any underperformance could be
reconciled against easier-to-deliver work. In the end, contractual targets were consistently
delivered, but there was little evidence to confirm whether performance was "good" or "bad"
(and thus no way to easily work out what the Partnership delivered as a whole)

In some ways the Partnership appeared to be effective: it hit contractual targets and IT user
surveys showed improved perceptions of performance. In interviews, senior personnel boasts
about the cordiality of relations. All were committed to Futuretech making a "reasonable"
profit and had a strong interest in the contract being a "success" (future careers depended on
it!). However, there was also evidence that performance was less than effective. Some
(knowledgeable) respondents complained about the quality of the technologies delivered and
suggested that, over time, Futuretech delivered increasingly ‘off-the-peg’ rather than more
complex and challenging ‘bespoke’ technologies. Ultimately, it is likely to have been more
generally recognised that the Partnership was failing to deliver as effectively as it might (the
contract was not renewed after its five-year term ended).

In this case, abstraction from qualitative data revealed the Partnership, as a set of inter-
organisational meetings that negotiated contractual details. In this study access to the
partnership was good and so abstraction of the basic mechanisms (the Partnership) from the
data required little prior theoretical engagement. However, the realists' movement, from the
description of the mechanism to an account of the specific outcomes associated with also
assumes a context in which deeper levels were awaiting discovery. It is explaining thiese
deeper levels which necessitates the use of novel theoretical resources. For example, and
from the data, we might have tracked the career interests of the people who constituted the
partnership and abstracted novel theory about these interests resulted in self-serving
outcomes. However, such data-abstracted domain specific theories can always be deepened.
Why did the Partnership have so much latitude to determine its own ‘successes’? Why could
Futuretech get away with delivering apparently shoddier and less customer-focused
technology? Answering these questions involves taking a retroductive steps ‘backwards’ to
ask why the Partnership manifested itself as it did and not otherwise.

At this point existing theory can be an invaluable tool in developing a deeper understanding
of the mechanism observed. As the data were collected the researchers also trawled the
literature to see how others theorised cooperative inter-organisational relationships. Various

16
theories were considered and combined in the effort to provide a more effective insight into
particular outcomes observed in the Partnership. Three theories, in particular, appeared to be
useful. These were transaction cost economics (TCE) which suggests inter-organisational
relations are affected by the type of product developed and exchanged; resource dependency
theory (RDT) which suggests inter-organisational relations are affected by each partner’s
relative dependency on the other’s resources and abilities; and, institutional theory, which
suggests inter-organisational relations can be affected by broader ideological trends and
norm-enforcing mechanisms. There is not space here to do justice to the complexities of these
theoretical frameworks or why they were selected. Instead, we explain how these theories
were incorporated within an explanatory framework for the Partnership.

Each of these theory resonated with different patterns in the data. For example, TCE suggests
that where contractual mechanisms govern complex, uncertain, changeable, and idiosyncratic
tasks and technologies the actors involved will necessarily have greater autonomy in
determining their own ends. Documentary analysis confirmed that technologies were
developing quickly within the market and interviews confirmed that benchmarks for good
performance were ineffective. TCE, a theory about how market conditions affect contracts,
thus helped explain the Partnership's apparent autonomy and rooted this explanation in a
specific cause (technological uncertainty within a broader social formation).

Suffice to say, a similar story can be told for RDT, which appeared to account for the
increasingly 'off the peg' technologies Futuretech apparently delivered. Specifically, as
Futuretech assumed responsibilities for developed technologies, knowledge accrued on
Futuretech's side of the contract rather than Govco's. As a result, Futuretech’s agents became
increasingly powerful in asserting their own technological imperative and interests, even if
these did not meet Govco’s needs exactly. RDT thus contributed to the theoretical model
developed, with data suggesting Govco (as one entity) became increasingly dependent on
Futuretech (as another) over time, and that this may have altered the balance of power
negotiations about technologies developed. Again, this theory fits the data.

Finally institutional theory, which suggests (institutional) mechanisms are conditioned by

dominant organising logics that operate across broader social formations, contributed to the
explanation of the Partnership’s tendency to extol its own successes. At the time of the
Partnership’s inception private sector provision was prioritized over public sector provision
as a matter of policy prescription - owing to a generalized ideological faith in the relative
efficiency of private sector providers (even where transactional considerations suggested this

17
may not actually be the case). In these circumstances, it is unsurprising that few were
shouting about the Partnership’s failures.

Arguably, this synthesis of existing theory and the qualitative case data generated a better and
theoretically informed model of the generative mechanism that more accurately accounted for
the specific empirical tendencies observed within the case. It added richness to the
explanation of the Partnership by rendering more explicit the causal dynamics that existed
between the Partnership and its antecedents, including the material (e.g. technology/resource)
and ideational (norms, discourses) phenomena to which it related.

Theoretical models, such as this combination of TCE, RDT and institutionalism, are
transferable generalisations that can be reapplied in analyses of similar institutional settings.
More specifically, the combination of insights about idiosyncratic technical conditions,
mutual resource dependencies, and supporting institutional rule systems, which inhered
within the Partnership, are likely to be present in cooperative inter-organizational forms more
generally. This model can, then, be used as a basis for building alternative explanations of the
particular configuration of contextual determinants that patinas other similar structures.
Knowledge obtained about a single case study is thus not confined to the boundaries of the
case itself but is theoretically transferable across a class of cases. Thus, the theoretical models
we develop through our case studies, as we explain their peculiarities, also help articulate the
specific conditions that makes a class of cases classifiable in terms of their common
antecedents. It is this form of theoretical generalization that qualitative CR researchers should
seek to extend and develop.

Conclusions

Critical realism provides, in our view, a more serious, consistent and credible alternative to
positivism, interactionism, sociomaterialism, and constructivism, and allows researchers to
move beyond the cultural and moral relativism inherent in many of these accounts. Moreover,
through its commitment to stratification and emergence critical realism is entirely capable of
incorporating epistemological insights from these competing positions without accepting
their ontological flaws. This is because critical realism overcomes the objectivist/subjectivist
and qualitative/quantitative dichotomies, because it is methodologically pluralist and
inclusive, and because it provides a philosophically informed methodologies (abduction and
retroduction) for generate new insights.

18
Yet, critical realism is not without its difficulties. First, as we have shown, it does not (and
cannot) generate any specific domain level theorising, other than by ensuring that studies
commit to a stratified, emergent account of the social world. Second, it does not prescribe
which methods are suitable for investigating which problems. This again, is down to the
researcher’s experience and intuition. Third, in its ontological guise, critical realism is not
actually particularly critical. Its commitment to truth does mean that false beliefs and
ideologies can be labelled as such, but, in order to really emancipate, we must do more than
identify the disempowered. As Collier explains:

When it is just a set of false beliefs that enslaves, their replacement by true beliefs is
liberation. But the vast bulk of human bondage, misery and repression is not like that.
The extension of emancipatory critique from cognitive error to unsatisfied needs
makes it clear that false belief is not the only chain that binds us . . . unemployed
workers, homeless families, bullied wives, tortured prisoners, may all know exactly
what would make them free, but lack the power to get it . . . Hence cognitive
enlightenment is a necessary, though not a sufficient, condition of their emancipation.
(Collier 1998: 461)

Yet some critical realist are working to develop this limitation. Sayer (2011) and others have
sought to detail the properties humans possess which imply ethical practices for how they
should be treated (see also Smith 2010; Nussbaum 2006). CR researchers should therefore be
challenged to consider the means required to act on and change the world (see Ram et al.,
2014).

Finally, critical realism is relatively difficult to operationalise: many critical realist pieces in
sociology and organisation studies are compatible with critical realism, but do not end up
detailing a clear set of entities, properties, causal mechanisms, triggers and so on. Others use
theorists, such as Archer (2003), to provide an applied critical realist framing concerned with
social structure, agency and reflexivity, which is more suited for social research than a pure
critical realist ontology. Others, including many Marxists, and Labour Process Theorists,
undertake studies which do not mention, but are entirely compatible with critical realism (see
Edwards, 2005). As one prominent theorist recently retorted to one of the authors ‘I don’t see
why you need to mention critical realism, surely it’s just good sociological research!’.

And yet, we would retort, good sociological research benefits from having clear analytical
targets (indeed, this what makes it "an ology" in the first place!). As such, critical realism

19
offers a common language for the identification of such targets that does more than appeal to
"common sense". However, as critical realism is a relatively new position, it is still
developing: making new connections, re-interpreting alternative perspectives, and responding
to challenges. The work of Bhaskar (Bhaskar 2008) on dialectical critical realism, for
example, certainly offers opportunities for novel research which have not yet been
incorporated into mainstream social research. As we have seen, this nascence presents
challenges to researchers, but also offers them opportunities to contribute meaningfully to
one of the most important developments in social theory in recent times.

References
Archer, M. S. 2000. Being Human: The Problem of Agency. Cambridge: Cambridge University Press.
Archer, M. S. 2003. Structure, Agency and the Internal Conversation. Cambridge: Cambridge
University Press.
Barad, K. 2007. Meeting the Universe Halfway: Quantum Physics and the Entanglement of Matter
and Meaning. Duke University Press.
Beynon, H. 1979. Working for Ford. EP Publ.
Bhaskar, R. 1975. A Realist Theory of Science. Routledge.
Bhaskar, R. 1979. The possibility of naturalism : a philosophical critique of the contemporary human
sciences. Atlantic Highlands, N.J.: Humanities Press.
Bhaskar, R. 1993. Dialectic : the pulse of freedom. London ; New York: Verso.
Bhaskar, R. 1994. Plato etc. : the problems of philosophy and their resolution. London ; New York:
Verso.
Bhaskar, R. 2008. Dialectic : the pulse of freedom. London ; New York: Routledge.
Bhaskar, R. & Hartwig, M. 2010. The formation of critical realism : a personal perspective. New York:
Routledge.
Bøllingtoft, A. 2007. 'A critical realist approach to quality in observation studies.' In H. Neergaard & J.
Parm (Eds.) Handbook of Qualitative Research Methods in Entrepreneurship: 23-49.
Cheltenham, UK: Edward Elgar.
Botterill, D. 2003. 'An autoethnographic narrative on tourism research epistemologies.' Loisir et
société/Society and Leisure, 26:1, 97-110.
Burawoy, M. 1979. Manufacturing Consent: Changes in the Labor Process Under Monopoly
Capitalism. University of Chicago Press.
Cassell, C. & Johnson, P. 2006. 'Action research: Explaining the diversity.' HUMAN RELATIONS, 59:6,
783-814.
Clark, P. 2012. Organizations in Action: Competition Between Contexts. Taylor & Francis.
Cruickshank, J. 2003. Critical Realism: The Difference it Makes. London: Taylor & Francis.
Cully, M., Oreilly, A. & Dix, G. 1999. Britain At Work: As Depicted by the 1998 Workplace Employee
Relations Survey. Taylor & Francis.
Danermark, B. 2002. Explaining Society: Critical Realism in the Social Sciences. Routledge.
Delbridge, R. 1998. Life on the Line in Contemporary Manufacturing: The Workplace Experience of
Lean Production and the "Japanese" Model. Oxford University Press.
Edwards, P. K., O'Mahoney, J. & Vincent, S. 2014. Studying Organizations Using Critical Realism: A
Practical Guide. OUP Oxford.
Elder-Vass, D. 2010. The Causal Power of Social Structures: Emergence, Structure and Agency.
Cambridge: Cambridge University Press.

20
Elder-Vass, D. 2014. 'Disassembling actor-network theory.' Philosophy of the Social Sciences, 45:1,
100-45.
Fairclough, N. 2005. 'Peripheral vision: discourse analysis in organization studies: the case for critical
realism.' Organization Studies, 26, 915-39.
Fleetwood, S. 2005. 'Ontology in organization and management studies: a critical realist perspective.'
Organization, 12, 197-222.
Fleetwood, S. & Hesketh, A. 2010. Explaining the Performance of Human Resource Management.
Cambridge University Press.
Friedman, V. J. & Rogers, T. 2009. 'There is nothing so theoretical as good action research.' Action
Research, 7:1, 31-47.
Fyfe, N. R. & Bannister, J. 1996. 'City watching: closed circuit television surveillance in public spaces.'
Area, 37-46.
Gouldner, A. W. 1964. Patterns of Industrial Bureaucracy. Free Press.
Jenkins, S. & Delbridge, R. 2014. 'In pursuit of happiness: A sociological examination of employee
identifications amongst a ‘happy’ call-centre workforce.' Organization, 21:6, 867-87.
Kazi, M., Blom, B., Moren, S., Perdal, A.-L. & Rostila, I. 2002. 'Realist evaluation for practice in
Sweden, Finland and Britain.' Journal of Social Work Research and Evaluation, 3:2, 171-86.
Kazi, M. A. F. 2003. Realist Evaluation in Practice: Health and Social Work. SAGE Publications.
Kempster, S. & Perry, K. 2014. 'Critical Realism and Grounded Theory.' In P. Edwards, J. O'Mahoney
& S. Vincent (Eds.) Studying Organisations Using Critical Realism: A Practical Guide: 86-108.
Oxford: Oxford University Press.
Kessler, I. & Bach, S. 2014. 'Comparing Cases.' In P. Edwards, J. O'Mahoney & S. Vincent (Eds.)
Studying Organisations Using Critical Realism: A Practical Guide: 168-84. Oxford: Oxford
University Press.
Ketokivi, M., Mantere, S., Reasoning, I., Practical, T. H. E. & Dilemma, R. 2010. 'TWO STRATEGIES FOR
INDUCTIVE REASONING IN ORGANIZATIONAL RESEARCH.' Academy of Management Review,
35, 315-33.
Kirkpatrick, I., Ackroyd, S. & Walker, R. 2005. The New Managerialism and Public Service Professions:
Change in Health, Social Services and Housing. Palgrave Macmillan.
Lindqvist, S. 1979. 'Dig where you stand.' Oral History, 24-30.
Lukes, S. 2005. Power: a radical view. London: Palgrave Macmillan.
Marchal, B., van Belle, S., van Olmen, J., Hoerée, T. & Kegels, G. 2012. 'Is realist evaluation keeping
its promise? A review of published empirical studies in the field of health systems research.'
Evaluation, 18:2, 192-212.
Mingers, J. 2006. Realising Systems Thinking: Knowledge and Action in Management Science.
Springer US.
Morgan, J. & Olsen, W. 2008. 'Defining Objectivity in Realist Terms: Objectivity as a Second-Order
‘Bridging’Concept Part II: Bridging to Praxis.' Journal of Critical Realism, 7:1, 107-32.
Mutch, A. 2007. 'Reflexivity and the institutional entrepreneur: a historical exploration.'
Organization Studies, 28, 1123-40.
Mutch, A. 2013. 'Sociomateriality—Taking the wrong turning?' Information and organization, 23:1,
28-40.
Mutch, A. 2014. 'History and Documents in Critical Realism.' In P. Edwards, J. O'Mahoney & S.
Vincent (Eds.) Studying Organisations Using Critical Realism: A Practical Guide: 223-40.
Oxford: Oxford University Press.
Næss, P. & Jensen, O. B. 2002. 'Urban land use, mobility and theory of science: Exploring the
potential for critical realism in empirical research.' Journal of environmental policy &
planning, 4:4, 295-311.
O'Mahoney, J. 2011. 'Embracing essentialism: A realist critique of resistance to discursive power.'
Organization, 19:6, 723-41.

21
O'Mahoney, J. & Vincent, S. 2014. 'Critical realism as an empirical project: a beginner's guide.' In P.
Edwards, J. O'Mahoney & S. Vincent (Eds.) Putting Critical Realism into Practice: A Guide to
Research Methods in Organization Studies. London: Oxford University Press.
Pawson, R. & Tilley, N. 1997. Realistic Evaluation. SAGE Publications.
Pearce, F. & Woodiwiss, T. 2001. 'Reading Foucault as a Realist.' In J. López & G. Potter (Eds.) After
Postmodernism: An Introduction to Critical Realism. London: Athlone.
Ponterotto, J. G. 2005. 'Qualitative research in counseling psychology: A primer on research
paradigms and philosophy of science.' Journal of counseling psychology, 52:2, 126.
Ram, M., Edwards, P., Jones, T., Kiselinchev, A. & Muchenje, L. 2014. 'Pulling the Leavers of Agency:
Implementing Critical Realist Action Research.' In P. Edwards, J. O'Mahoney & S. Vincent
(Eds.) Studying Organisations Using Critical Realism: A Practical Guide: 205-22. Oxford:
Oxford University Press.
Rees, C. & Gatenby, M. 2014. 'Critical Realism and Ethnography.' In P. Edwards, J. O'Mahoney & S.
Vincent (Eds.) Studying Organisations Using Critical Realism: A Practical Guide: 132-47.
Oxford: Oxford: Oxford University Press.
Sayer, A. 1997. 'Essentialism, social constructionism, and beyond.' Sociological Review, 45, 453-87.
Sayer, A. 1998. 'Abstraction: A realist interpretation.' In M. Archer, R. Bhaskar, A. Collier, T. Lawson &
A. Norrie (Eds.) Critical Realism: Essential Readings: 120-43. London: Routledge.
Sayer, A. 2000. Realism and Social Science. Sage
Sayer, A. 2010. Method in Social Science: Revised 2nd Edition. Taylor & Francis.
Sayer, A. 2011. Why Things Matter to People: Social Science, Values and Ethical Life. Cambridge
University Press.
Sims-Schouten, W. & Riley, S. 2014. 'Employing a forms of Critical Realist Discourse Analysis for
Identity Research: An Example from Women's Talk of Motherhood, Childcare, and
Employment.' In P. Edwards, J. O'Mahoney & S. Vincent (Eds.) Studying Organisations Using
Critical Realism: A Practical Guide, Oxford: 46-65. Oxford: Oxford University Press.
Smith, C. & Elger, T. 2014. 'Critical Realism and Interviewing Subjects.' In P. Edwards, S. Vincent & J.
O'Mahoney (Eds.) Using critical realism to study organisations: 109-31. Oxford: Oxford
University Press.
Smith, C. & Meiksins, P. 1995. 'System, society and dominance effects in cross-national
organisational analysis.' Work, Employment & Society, 9:2, 241-67.
Taylor, P. & Bain, P. 2003. '‘Subterranean worksick blues’: humour as subversion in two call centres.'
Organization Studies, 24:9, 1487-509.
Vincent, S. 2008. 'A transmutation theory of inter-organizational exchange relations and networks:
Applying critical realism to analysis of collective agency.' HUMAN RELATIONS, 61:6, 875-99.
Vincent, S. & Wapshott, R. 2013. 'Critical realism and the organizational case study: A guide to
discovering institutional mechanisms.' In P. Edwards, J. O'Mahoney & S. Vincent (Eds.)
Putting Critical Realism into Practice: A Guide to Research Methods in Organization Studies.
Oxford: Oxford University Press.
Wright, C. 2004. 'Limited incorporation or sleeping with the enemy: reading Derrida as a critical
realist.' In J. Joseph & J. Roberts (Eds.) Realism Discourse and Deconstruction: 201-17.
London: Routledge.

22