Introduction

Philosophy of science issues are under-appreciated in entrepreneurship and small business (ESB) research (Kenworthy & McMullen, 2014). Occasional calls are made for researchers to provide a stronger philosophical footing for studies but few heed them (Grant & Perren, 2002). The inference might easily be drawn that philosophical standpoints are arbitrary or of little consequence in ESB studies. Yet ESB researchers – like all other social investigators – necessarily make assumptions about the nature of the world they study (ontology) and how it can be known (epistemology) (Bhaskar, 1979; Fleetwood, 2005); the sum total of these presuppositions constitutes a philosophy of science (Porpora, 2015). This chapter seeks to increase appreciation of ontological issues to show how a better understanding of them can contribute to ESB studies. Ontological reflection can support ESB research, encouraging researchers to propose clear conceptions of the social objects¹ they believe exist and their properties. Critical realism is therefore intended as a philosophy for, as well as of, ESB research as it directs attention to what needs to be included in a powerful explanation of entrepreneurial action and, in particular, to the conditions that make such action possible.

Three aims motivate the chapter. First, to demonstrate that ontological presuppositions cannot be avoided in ESB research. Such commitments are inevitable in all research and are important for the kinds of explanation that can be made (Archer, 1995; Fleetwood, 2005). To be clear, therefore, researchers ought to be as explicit about these commitments as possible. Second, to argue for a particular standpoint, critical realism, and, to show why I believe it is superior to currently available alternatives (Bhaskar, 1978, 1979; Collier, 1994; Lawson, 1997; Sayer, 2000; Groff, 2004). Critical realism focuses on ontological questions, with what exists; such issues are logically prior to the epistemological question of what we can know. Third, to show that, in practice, ESB researchers often make assumptions close to those of critical realism, whatever their declared commitments. Fleetwood and Ackroyd (2004) have made a similar point in relation to the organisation and management literature. Where ESB researchers implicitly make ontological assumptions similar to those of critical realism, studies might be improved by applying critical realist thinking systematically in making conceptual, methodological and analytical choices.

The chapter is organised as follows. Section 2 elaborates why researchers inevitably make ontological commitments, either intentionally or inadvertently, in their studies. Section 3 traces the origins and development of critical realism through immanent critiques of the main philosophical alternatives – positivism, social constructionism and pragmatism. Section 4 sets out the main features of critical realist social ontology, drawing principally on the early work of Bhaskar. To illustrate the benefits, section 5 brings a critical realist-inspired framework to bear on one particular topic, entrepreneurial identity, where most studies have adopted a constructionist approach. Section 6 concludes.

The necessity for ontological commitments in entrepreneurship and small business research

Every researcher makes ontological assumptions in their theoretical and empirical work (Bhaskar, 1978; Archer, 1995; Fleetwood, 2005). Ontological assumptions refer to the kinds of social objects that exist and might be studied, or to generic features of the world. ESB researchers assume the existence of entrepreneurs, businesses, products, opportunities and markets, for instance. Where researchers refer to these objects in their analyses, they take them to be real (Archer, 1995); to argue otherwise is to commit a theory/practice contradiction. Moreover, ESB researchers frequently make causal claims – that X influences, produces or impacts Y. They thereby assume that objects – firms and markets, for instance – stand in a causal relation to one another. Such claims presuppose that causality is a feature of the social world (Groff, 2016). Researchers may state ontological commitments explicitly or, more commonly, adopt them implicitly in their research practice. Ontological presuppositions are therefore non-optional (Fleetwood, 2005); failure to be explicit about them only leads to their unacknowledged reintroduction into data collection, analysis and explanation. Such assumptions necessarily influence research practice, shaping conceptions of the social objects alleged to exist and the methods used to study and analyse them.

Bhaskar (1978) distinguishes between philosophical and scientific (regional or domain-specific) ontologies. Scientific ontologies are those presupposed by theories in a particular field positing the existence of particular entities. Philosophical ontologies specify generic features of reality with which scientific ontologies must be consistent; critical realism is a philosophical ontology. It does not posit the existence of particular entities as a scientific theory would, so it does not say anything about specific objects such as new ventures, business practices or market relations. The value of critical realism lies in supporting robust, though fallible, causal explanations of ESB objects such as venture creation, product development or business performance by providing strong accounts of agency, causality, social structure and emergent properties. Critical realism cannot adjudicate between rival theories that are consistent with its ontological assumptions; these are substantive matters for ESB researchers to debate.

Arguably, all ESB researchers are committed to what might be termed a minimal ontological realism, the claim that the social world exists and is what it is largely independently of any particular observer (Searle, 2010). Such a ‘thin’ realism, however, may be of little help to researchers because it permits approaches that have arguably had a damaging effect such as positivism (Sayer, 2004), about which I say more in section 3. ‘Thicker’ versions of realism may be of more help to researchers because they set a higher bar for theory to meet. Not every realist philosophy can meet these more demanding standards with regards to issues such as causality, for instance. Critical realism supplies a thicker set of ontological commitments to support research and theorising (e.g. Bhaskar, 1979; Archer, 1995; Elder-Vass, 2010), licensing a range of theories that are consistent with them, while judging those that are inconsistent with such assumptions to be either false or incomplete (Bhaskar, 1978).

Critics of critical realism claim that Bhaskar makes ontologically foundationalist – i.e. infallible – claims that the world must possess particular characteristics if science is to be possible (Pleasants, 1999; Cruickshank, 2004). While Bhaskar’s comments on this issue have been seen as ambiguous, he has insisted that his approach was the only theory of science then consistent with the scientific data (Bhaskar, 1978). But, as others have noted, this cannot guarantee that another future ontological theory will not do better (Chalmers, 1988). Critical realists can accept that all knowledge claims, including ontological ones, are in principle fallible and revisable.

Following Bhaskar, therefore, I contend, first, that ontology cannot be avoided, and, second, that critical realism offers the best approach – but do not claim that no better ontological argument will ever be developed (Bhaskar, 1978; Lawson, 1997). The argument against critical realism that ontology should not regulate, and thereby limit, the conduct of social inquiry (Kemp, 2005) or form of explanation (Tsilipakos, 2015) needs to be balanced against Bhaskar’s (2016, p. 209) view that critical realism is ‘maximally inclusive’ and open to the discovery of new entities and layers of reality. Critical realism permits a wider range of social objects as potentially significant for scientific explanation than rival philosophical positions.

Immanent critiques of alternative philosophical standpoints

From a realist perspective, philosophy is wrong to start with the problem of epistemology, with how we know things. What we can know depends on what there is to know, that is, on the nature of reality (Potter & López, 2001). This is the problem of ontology. Epistemological questions presuppose answers to ontological ones. Hence Bhaskar’s (1978, 1979) interest in the ontological question: what must the world be like for science to be possible? The word possible is necessary here to make it clear that scientific explanations depend on the activities of researchers in the respective fields to produce them. Philosophy cannot provide such explanations; rather, it underlabours for the sciences, striving for conceptual clarity and specifying the conditions the sciences presuppose (Bhaskar, 1978).

Bhaskar (1978, 1979) developed critical realism out of immanent critiques of prior philosophical positions, notably positivism and hermeneutics. Immanent critique takes a particular argument and demonstrates theory/practice inconsistencies within it or shows how adopting the argument generates problems that are insoluble in its own terms (Hartwig, 2007). Bhaskar (2016) distinguishes his approach to the natural sciences, which he terms transcendental realism, from that of the human sciences, which he terms critical naturalism. The term ‘critical realism’ originated in the elision of these two separate approaches. Positivism and hermeneutics are now discussed in turn.

Positivism claims that scientific laws are grounded in exceptionless empirical regularities – ‘whenever X, then Y’ – which, in turn, presuppose an ontology of constant conjunctions of events. Theoretical claims are confirmed or falsified by their empirical instantiation (Bhaskar, 1978). To critique positivism in the natural sciences, Bhaskar takes experimentation as his starting point because it is widely regarded by scientists as a successful knowledge-generating practice. His analysis shows brilliantly that experiments presuppose an ontological distinction between scientific laws and patterns of events. For Bhaskar, laws are simply the ways that entities act, not their empirical manifestations in regular (or irregular) sequences. Through experimental activity, scientists produce experimental events but not the causal powers/mechanisms that such events enable them to identify. Only by assuming the real ontological independence of powers from the events they generate can we assume they endure and continue acting outside the experimentally closed conditions that allow scientists to identify them. Positivism, presupposing an ontology of invariant events, cannot explain what governs events in open systems where exceptionless regularities do not obtain. To be consistent, positivists would have to say that scientific laws apply only in the laboratory where the regularities occur or that events outside the laboratory are not governed by laws!

Bhaskar (1978) refers to event-based ontologies like positivism as guilty of actualism, that observable events exhaust the real. Actualism restricts science to that which can be observed. Positivism has no place for ‘unobservables’ as possible objects of scientific investigation. What cannot be perceived cannot be known and is therefore inadmissible to science (Fleetwood, 1996). This rules out reference to businesses, markets, institutions and discourses, none of which can be observed. Realist explanations, conversely, permit a causal criterion for reality as well as an experiential one. Theory can posit entities which cannot be observed directly and whose existence can only be inferred from the effects they cause (Sayer, 1992; Danermark, Ekstrom, Jakobsen, & Karlsson, 2001). Critical realism rejects positivism as a philosophy of both the natural and the human sciences on the grounds that exceptionless regularities presuppose closed systems, that is, social objects and their contexts of operation remain invariant over time. But both the natural and social worlds are open systems involving the interaction of multiple entities that generate novel events that we (sometimes) experience.

In the human sciences, a long tradition of research has pointed to their important differences from the natural sciences, specifically that researchers study meaningful action and seek interpretive understanding rather than causal explanation (Winch, 1958). The meaning of action is argued to be constructed through interaction between people (Gergen, 2009). Constructionist accounts typically emphasise the role of language or discourse in the constitution of social reality. Four points might be made by critical realists in relation to such arguments. First, construction is a misleading metaphor and often confuses ontological and epistemological talk (Porpora, 2015). It confuses in two ways by evading the question of the relationship between social constructions and their referents (Sayer, 1997): first, it elides the material construction of a practice or event with the naming, categorising, interpreting or construal of it and, second, it conflates agents’ and researchers’ interpretations/construals. Hacking (2002) poses the question, ‘the social construction of what?’ to capture these ambiguities: just what is being constructed when constructionists make knowledge claims?

Second, social life is inherently conceptual (meaningful) but it is not exhausted by the conceptuality it depends upon (Bhaskar, 2016). Meanings are real (Maxwell, 2012) – they exist independently of the researcher studying them – but are not all there is. Although semiosis (meaning-making) is an aspect of any social practice, no practice is reducible to semiosis alone (Fairclough, Jessop, & Sayer, 2002). Meaning-making necessarily occurs in a social context of structured relations that pre-date meaning-making (Fairclough et al., 2002). Material relations of resource inequality, autonomy and dependence enable and constrain activities, even where agents consciously draw on cultural rules to act (Porpora, 1993). Moreover, making sense of events, even recent ones, is inevitably a historical act, that is, an act taking place at a particular time; agents give meaning to a world that pre-dates, and therefore exists (or existed) apart from, their interpretations/construals.

Third, people are not only meaning-endowing creatures attributing meanings to relations, events, human-made artefacts and nature; they are also causal agents capable of intervening in, and transforming, the social and natural worlds (Bhaskar, 1979; Archer, 1995). Persons are proactive emergent bio-psycho-social realities, conscious, reflexive, embodied, self-transcending centres of subjective experience and durable identity (Smith, 2010). Bhaskar (2016, p. 88) notes how conceptuality is both causally conditioned by, and causally efficacious on, the material, extra-conceptual, extra-discursive, extra-linguistic dimensions of human life. Not just any meanings will do; to argue otherwise is to insist that agents’ descriptions/construals can never be mistaken. Agents’ accounts are limited by the existence of unacknowledged conditions, unintended consequences, tacit skills and unconscious motivations, and therefore corrigible (Bhaskar, 1998). Researchers need to investigate the extra-semiotic conditions that make semiosis possible and secure its efficacy. We cannot, in most cases, change reality merely by redescribing it (Bhaskar, 1991).

Fourth, agents may construe practices, events and relations in certain ways – but it is always possible to conceptualise these social objects in different ways for different purposes. Multiple descriptions may be ‘practically adequate’ (Sayer, 1992). For example, individuals going to work every day might construe their activity as earning a livelihood. A third party observer might, in contrast, describe it as reproducing a market economy and that this is the case whether the individuals involved are aware of, or intend, this. So we do not have to rely solely on agents’ own construals to provide an adequate description. Critical realists are comfortable with ‘moderate’ forms of constructionism that acknowledge the epistemic relativity of meanings and the socio-historical variability of knowledge claims, while denying ‘radical’ constructionist claims that that the reality of social processes is exhausted by agents’ culturally shaped descriptions (or constructions) of them (Elder-Vass, 2012).

A newer entrant to philosophical debates in the ESB field is pragmatism (Sarasvathy, 2008; Berglund & Wennberg, 2016). Pragmatists try to stand clear of ontology (Powell, 2003), avoiding ontological questions as being impossible to settle (e.g. Rorty, 1982; Baert, 2005). Talking about ontology is treated simply as a language game played by a particular breed of philosopher (Kivinen & Piiroinen, 2004). Pragmatists prefer to emphasise the usefulness of ideas and beliefs as guides to action. Against this kind of thinking, two points might be made. First, as argued in section 2, ontological claims cannot be avoided. There is no way to think or talk about social reality without committing oneself, metaphysically; it is a myth to believe otherwise (Groff, 2016). Pragmatists, like everyone else, make ontological assumptions and represent the world through their descriptions. Terms like ‘ontology’, ‘breed of philosopher’, ‘language game’ and ‘usefulness’ must be intended to refer to something in order to be meaningful and thereby useful. Otherwise, entirely different words might be used and be expected to be equally useful. Second, one might ask: what is it about a particular belief that makes it a useful guide to action? To engage in a practice successfully, one might argue that one’s beliefs must not be inconsistent with the way the world actually is, although one cannot take the further step and assume that one’s beliefs mirror the world. Pragmatists are therefore caught in theory/practice inconsistencies where they assume their analyses do not assume certain ontological features or represent the world in particular ways.

To end this section, it is important to distinguish philosophical accounts of social science from researchers’ actual practices. It is often thought that quantitative research is underpinned by positivism, and qualitative approaches are supported by some form of interpretivism/constructionism. I cannot recall, however, a single instance of a quantitative ESB researcher making an explicit argument for positivism to support their work. Rather, the approach has become entirely conventional, with authors perceiving no need to argue why probabilistic empirical regularities – exceptionless ones hardly ever occur, if at all – constitute good grounds for knowledge claims. Probabilistic associations do not, however, satisfy the standard for positivist knowledge claims because there are unexplained exceptions. It seems more plausible to argue instead that, in the absence of invariant event regularities, quantitative researchers work, implicitly, with a different set of ontological assumptions.

Similarly, constructionist approaches often refer to non-experiential entities in their analyses. Alvarez and Barney (2007), for instance, in their constructionist approach to opportunity creation, make reference to market imperfections but these appear to be treated as real independently of anyone perceiving them or construing their existence. Gergen (2009, p. 161) reports that realist talk is essential to carry out life in a particular tradition but that we cannot describe the world as it really is. Such claims assume the existence of different traditions of life, but to be consistent a constructionist would have to limit reference to people’s perceptions/constructions of traditions of life. Many constructionists thereby assume, unwittingly, the existence of non-experiential objects independent of agents’ constructions of them.

In both cases, researchers refer to a wider range of objects in explanations than is strictly permitted by either a positivist or a constructionist ontology. This suggests that quantitative and qualitative research practices might be better conceptualised within a more encompassing ontology, one allowing experiences and events but also the powers that generate them. Such powers should include purposive human agents and the relations that influence action. A robust social ontology should allow for both agents and social relations without depending on event regularities for their identification or restricting what exists to what the agents involved in particular practices say exists.

Critical realism: an ontological framework for the sciences?

Critical realism is intended, ambitiously, to provide an overall philosophical framework for the natural and human sciences (Bhaskar & Hartwig, 2010; Bhaskar, 2016). In contrast to both positivism and constructionism, critical realism prioritises ontology rather than epistemology, and within ontology, shifts the emphasis from events and experiences to the generative powers that produce events and experiences (Bhaskar, 1978, 1979). An adequate philosophy of science must be able to sustain both a transitive dimension and an intransitive dimension (Bhaskar, 1978). The transitive dimension refers to the existing knowledge that researchers use to generate new knowledge. No-one starts with a blank page; all start with preunderstanding, derived either from prior practical activity or scientific learning (Alvesson & Sköldberg, 2009). The intransitive dimension refers to the real entities that exist and act independently of the conditions that allow observers epistemic access to them. The world can only be known under particular descriptions, in terms of available conceptual resources (Bhaskar, 1978; Sayer, 2000) but our knowledge is always of or about something that exists (or existed) apart from the observation.

Realism does not entail claims that we can achieve foundational, infallible knowledge. On the contrary, realism and fallibilism presuppose one another (Sayer, 1992). To be a fallibilist about knowledge, it is necessary to be a realist about things (Bhaskar, 1978). Beliefs cannot be fallible unless there is something independent of the believing of them that would make them true or false (Collier, 2003).² It is the evident fallibility of our knowledge – the experience of getting things wrong – that justifies us in believing that the world exists regardless of what we happen to think about it (Sayer, 2000). Contrary to the constructionists, agents’ definitions of the situation are fallible. If the world is necessarily just as agents say, they could never be wrong.

Critical realism is a philosophical ontology that offers an answer to the question ‘what must the world be like for science to be possible?’ The centrepiece of critical realism is the transcendental deduction of the stratified, open nature of the world arising from analysis of the experimental activity of scientists. Bhaskar’s (1978) ontological stratification between the three domains of the real, the actual and the empirical is central to critical realist thinking:

• the real – the powers (or mechanisms) that generate events, including the capacity (liability) to undergo certain kinds of change;
• the actual – the events that occur; and
• the empirical – experience.

This ‘deep’, or stratified, ontology can be distinguished from the ‘flat’, actualist ontologies of positivism and constructionism that centre on experience and events but ignore powers, mechanisms and capacities. Causation is simply the activation of powers to produce effects, although the existence of a power does not guarantee the occurrence of particular events. A few examples should suffice to illustrate the difference between powers and events. Barcelona footballer, Lionel Messi, possesses the power to play even when he’s at home asleep in bed! Businesses retain the power to produce goods and services, even when they are closed for the weekend and not actually producing anything. Markets possess the power to transform the terms upon which businesses conduct trade and to eliminate some from competition, even as new firms enter the marketplace and incumbent firms persist with existing practices.

These three ontological domains of the real, the actual and the empirical are commonly out of phase (Bhaskar, 1978). The existence of a power does not determine events or entail that someone experiences those events. Powers may be possessed unexercised, exercised and actualised in a variety of events, and events may go unobserved. Contrary to positivism, event regularities are therefore neither sufficient nor necessary to identify a power. Given that society is an open system with multiple powers operating simultaneously, invariant event regularities are unlikely to occur. Powers, when exercised, often counteract the effects of one another. Ontologically speaking, powers must therefore be distinct from the pattern of events they generate, and from our observation of events. This moves us towards a science concerned with possibilities not actualities – there is more to reality than what actually happens. The task of the sciences is therefore to penetrate deeper levels of reality, showing how new powers emerge from the interaction of existing ones, making novel events and experiences possible.

Turning to the human sciences, Bhaskar (1979) starts from the widely held belief that what makes them different is intentional agency, the capacity for human beings to act on reasons.³ Critical realists, along with constructionist thinkers, accept the meaningful character of social life but make stronger arguments in relation to the intransitivity of reasons, their capacity to be causally efficacious, and the fallibility of the beliefs upon which reasons depend. Social life is not, however, exhausted by intentional agency. Critical realism incorporates meaningful activity within wider causal explanations of the reproduction and transformation of social practices and settings. The social world is also constituted by the relations between agents whose interaction produces events that nobody specifically willed. Society is the unintended outcome of agents’ intentional actions (Archer, 1995).

This brings us to the controversial issue of structure and agency. Societies are irreducible to people; from birth, individuals are enmeshed in a framework of social-structural relations that influence, but do not determine, what they can do (Archer, 1995). The social context is a necessary condition for any intentional act (Bhaskar, 1979). The pre-existence of social forms, such as businesses and markets, establishes their autonomy as possible objects of scientific investigation and their causal power establishes their reality (Bhaskar, 1979). Entrepreneurs establishing a company presuppose a legal system setting out rights and obligations; hiring employees presupposes a labour market; paying suppliers in 30 days presupposes a business norm regarding settling invoices; and cutting prices to increase sales presupposes a commodity market populated by rival producers and prospective buyers, themselves both motivated to act by their own socially structured positions.

The structure/agency approach seeks to avoid the twin problems of structural determinism (where large-scale social forces such as class, gender or nation determine events) and voluntarism (where events can be explained solely by reference to agent motivations). Much ESB research veers towards voluntarism, assuming that imaginative, creative and daring entrepreneurs can re-make the world at will while paying insufficient attention to the social context that enables, as well as constrains, resource acquisition and mobilisation (e.g. Sarasvathy, 2008; Chiles, Vultee, Gupta, Greening, & Tuggle, 2010). Archer (2000a) refers to this kind of voluntarist approach as a ‘stop-the-world-while-we-get-off-and-change-it’ condition, as if agents can simply choose to interpret their social relations differently independently of the influence of unequal resource distributions and agents’ vested interests in retaining, or changing, society. Social relations necessarily involve the exercise of power to reproduce or transform the social context, whether or not agents notice or conceptualise this (Porpora, 2015).

The pre-existence of social forms entails a transformational model of social activity, a relational conception of the subject matter of social science in which society is both the ever-present condition (material cause) and the continually reproduced outcome of human agency (Bhaskar, 1979). Social structures such as organisations and markets are relationally emergent products of human interaction that generate the material circumstances, positions and relations within which agents must act and which motivate them to act in certain ways (Bhaskar, 1979; Porpora, 1989; Sayer, 1992; Lawson, 1997). Social structures are constituted by relationships between internally related positions,⁴ possessing the power to influence the exercise of agency by those occupying particular positions. For example, the activities of entrepreneurs, as suppliers of goods and services, are influenced by relations with rival producers, consumers and suppliers. All action is positionally conditioned; agents’ activities are conditioned by their inherited structural and cultural context (Archer, 1995). Martinez Dy, Marlow, and Martin (2017), for example, note how individual positioning within intersecting class, gender and race hierarchies influences the social privileges and material resources available to invest in digital ventures and thereby constrain entrepreneurial potential.

We can separate, analytically, cycles of structural and cultural conditioning, agential interaction and structural and cultural elaboration (Archer, 1995). Agential interaction, drawing on prior structural and cultural conditioning, gives rise intentionally and unselfconsciously (Sayer, 2009; Akram, 2013) to structural and cultural emergent properties (Archer, 1995). Entrepreneurs forming new ventures, for instance, contribute to the modest or substantial transformation of a market economy whether or not they intend or understand this.

Emergent properties turn back to confront human actors with circumstances which are not fully of their own making. Changing market contexts causally shape the exercise of future agency; they enable and limit what entrepreneurs are able to do next. Such circumstances exist and exert autonomous effects, largely through the actions of important close and distant stakeholders, regardless of how entrepreneurs themselves perceive such circumstances. For example, an entrepreneur might experience a fall in demand for her firm’s product but be unaware whether this is due to the introduction of a direct competitor product, shifting consumer tastes, rising or falling disposable consumer incomes, or changing transport costs for overseas exporters. Constructionist accounts cannot simply wish such conditions away by encouraging agents to redescribe them (Gergen, 2009).

Critics of critical realism argue that it is mistaken to grant social structures causal powers (Varela & Harré, 1996; King, 1999; Manicas, 2006) or to accord them emergent properties (Le Boutillier, 2013). For these critics, to ascribe causal powers to social structure is to mislocate them. Only people possess causal powers, it is argued; people coordinate their activities through their conversational practices. Structures, it is argued, are virtually real because they cannot exist independently of human activity; people’s changing practices are the only social reality. But critical realists might argue that while people are the only efficient causes of action in the social world, in that social events can only happen through their activities, social structures are sources of material causality (Lewis, 2000). So, although social structures can generate effects only through people exercising their agential powers, structures nevertheless shape how individuals exercise their agency (Mutch, 2004). Without reference to the structural and cultural context, explanations of entrepreneurial action are inevitably unanchored in real social settings characterised by unequal resource distributions, diverse vested interests and good or bad reasons for action that enable and constrain what entrepreneurs can do (Archer, 1995).

The open, emergent nature of the social world encourages researchers to focus on explanation not prediction (Sayer, 1992). Although, retrospectively, we can often give well-grounded explanations of past events, prospectively we are very rarely, if ever, able to do so because we do not know which of the myriad of possible sets of circumstances will actually materialise (Bhaskar, 1986). Causal explanation requires description of explanatory powers, narratives of the contingent conjunctures of powers and adjudication between rival explanations (Porpora, 2015). Critical realists contend that the social world can be studied in the same sense if not the same way as the natural world, in terms of an ontology of powers, events and experience (Bhaskar, 1978). But because the objects of the natural and human sciences are very different, methods of data collection and analysis also differ (Bhaskar, 1979; Sayer, 1992).

Why critical realism matters to entrepreneurship and small business research: revisiting entrepreneurial identity

This section briefly discusses studies of entrepreneurial identity, a stream of research dominated by constructionist approaches that define identity in terms of narrative or discourse. The purpose of the discussion is threefold: first, to propose an alternative critical realist-informed conception of entrepreneurial identity as a personal causal power rather than as a narrative performance; second, to show how such a framework can encompass and explain the findings of narrative-based studies while also directing research attention to the conditions that make successful narrative performances possible; and, third, to demonstrate how constructionist accounts inevitably presuppose a stratified ontology of powers, events and experience despite an actualist focus on practice.

Constructionist works in the entrepreneurial identity literature focus on self-narration (e.g. Down & Reveley, 2004; Essers & Benschop, 2007), often performed by appropriating elements of wider discourses of enterprise to tell particular stories about themselves (Cohen & Musson, 2000; Watson, 2009; Anderson & Warren, 2011).⁵ Hence entrepreneurial identities are described as situated, fluid and multiple (Hytti, 2005) to the extent that entrepreneurs perform, or construct, different identities in dialogical interaction with diverse stakeholders including financiers, customers, suppliers, family and community members. To the extent entrepreneurs succeed in accomplishing a desirable and credible identity with important stakeholders, this enables entrepreneurs to access resources and markets (Lounsbury & Glynn, 2001).

An alternative conception of entrepreneurial identity might define it as a personal causal power, one that motivates and enables action, rather than as a narrative performance (Kašperová & Kitching, 2014). Drawing on Archer (2000b), this conception distinguishes personal identity as the set of concerns that makes each of us a unique person, and social identity as the public roles in which we can invest ourselves and commit to (Archer, 2000b; Marks & O’Mahoney, 2015). Being an entrepreneur, or being a particular kind of entrepreneur, is one such social identity. If we conceptualise entrepreneurs as agents with particular properties and powers, pursuing particular projects to further their concerns, then the capacity to enact particular narrative performances is one of their powers. Such performances, however, are quite distinct from the concern to establish oneself as a credible, legitimate entrepreneur. Such concerns make diverse narrative performances possible but do not dictate that identity powers are exercised, or how they are exercised, or that, if exercised, that they will necessarily produce the desired effects.

There are a number of problems with the constructionist conception of entrepreneurial identity from a critical realist standpoint. First, accounts that reduce identity to narrative performance presuppose an actualist ontology. Entrepreneurs’ identities are constituted solely by narrative practices as they are performed and endure only for the duration of the performance. Consequently, practitioners do not have identities when they are not performing narratively. Such approaches ignore, or do not theorise adequately, the conditions of possibility of particular instances of narrative performance and thereby risk reducing the analysis of entrepreneurial identity to descriptions of discursive practices. Accounts of identity are incomplete without reference to a non-discursive reality (O’Mahoney, 2012). Accomplishing a successful social identity such as being an entrepreneur requires interaction with important stakeholders but is not reducible to self-narration.

Second, accounts that reduce identity to narrative performance accord a limited explanatory role to the conditions enabling such practices. These conditions include entrepreneurs themselves as embodied agents with particular properties and powers (Kašperová & Kitching, 2014), and relations with important stakeholders. Entrepreneurs perform identity work in specific social contexts that influence outcomes. Identity work might not succeed; the influence of narrative practices should be demonstrated not assumed (Alvesson & Kärreman, 2011). Prospective entrepreneurs may fail in their efforts to present themselves as credible and legitimate to stakeholders with the consequence they are unable to obtain finance from investors, win orders from customers, secure inputs from suppliers or receive emotional support from family and community. Entrepreneurs from disadvantaged groups, in particular, may confront severe constraints in convincing important stakeholders they are worthy of support. Conceptualising identity as self-narration can divert attention from the circumstances that might constrain individuals from accomplishing a successful entrepreneurial identity, even where discrimination and prejudice are recognised to be important features of the wider political context (Tedmanson & Essers, 2016). Critical realists would argue that identity work practices, and their effects, must be explained with reference to entrepreneurs’ positioning within particular social relations. The well-financed, well-educated and well-connected might be better able to accomplish a credible, legitimate entrepreneurial identity in dialogue with stakeholders than others.

Third, constructionist accounts must presuppose an entrepreneurial agent, one whose positioning in relation to various social hierarchies – class, gender, ethnicity, dis/ability – is likely to influence interaction with stakeholders, as well as a wider social context, one that may be characterised by discrimination and prejudice (Essers & Benschop, 2007). Entrepreneurial identity work does not therefore reduce to how entrepreneurs present themselves narratively in micro-level interactions with university researchers or even with business stakeholders. Entrepreneurs cannot control all of the signals they give off. Stakeholders endow entrepreneurs’ appearance and action with meaning, consciously and inadvertently; these meanings are likely to vary with the relative social positioning of the entrepreneur and the interacting stakeholder. Focusing research on the stories entrepreneurs tell about themselves may be insufficient to explain what particular entrepreneurs do, or are able to do, with regard to accessing resources and markets. This task requires attention to wider dimensions of context, reaching beyond a constructionist focus on narrative. Explaining identity work, and its effects, requires an ontology that goes beyond experience, one that treats causation as a feature of the social world and acknowledges a distinction between causal powers, events and entrepreneurs’ experiences. This points towards a more encompassing ontological framework than one reliant solely on agents’ narrative constructions.

Conclusion

Entrepreneurship/small business researchers cannot avoid making ontological commitments in their empirical and theoretical work. Few, however, make such commitments explicit and these have to be inferred from how researchers conceptualise and analyse the objects they study. ESB researchers conventionally assume the existence of entities such as firms, resources, products and markets – as well as causal relations between entrepreneurs and their stakeholders. This does not mean that such entities do, in fact, exist – but these constitute the ingredients of the analysis.

I have proposed critical realism as a superior ontology to the major current alternatives of positivism, constructionism and pragmatism which either implicitly acknowledge the existence of entities they formally deny, or place no limits on what can be said. Critical realism is inclusive and open to the discovery of new entities, or layers of social reality, in facilitating (fallible) explanations of ESB objects such as new venture creation and various forms of market activity. Critical realism is therefore a philosophy for, as well as of, ESB research as it directs attention to the conditions of possibility of particular business practices. Adopting such an ontology permits explanation that goes beyond positivism’s emphasis on empirical regularities and constructionism’s emphasis on agents’ descriptions. Explanation of ESB objects does not depend on the identification of empirical regularities; nor do agents’ descriptions exhaust social reality.