Chapter 3: Reality and the Problem of Unobservables
Key concepts in this chapter
Hover a term to preview its definition, or click to open the concept page: Critical Realism · The Three Domains · Monism · Unobservables · Retroduction · Emergence
I Introduction: From Philosophical Foundations to Ontological Practice
Chapters 1 and 2 established the stakes of ontological commitment in social research and the conceptual architecture that adequate ontological analysis requires. The present chapter addresses the specific features of the ontological position that frames the rest of this book: the claim that social reality is stratified, that it contains entities and mechanisms not directly observable, and that the relationship between the observable and the unobservable is a central problem for social research rather than a difficulty to be dissolved by restricting claims to what can be measured.
The chapter proceeds in four stages. Section 3.2 develops the account of stratified reality on which the book depends, moving from the rejection of both dualism and flat empiricism toward a monist position that preserves the reality of the unobservable through stratification. Section 3.3 addresses the problem of unobservables directly, establishing why unobservable mechanisms and structures matter for social explanation and how stratified ontology handles them without collapsing into either metaphysical speculation or epistemological restriction. Section 3.4 examines the literature review as a research practice and the ontological function it serves, distinguishing the approaches available on different ontological positions. Section 3.5 introduces the four social ontological debates that organise Part II of the book, establishing the analytical framework through which positions are examined. Section 3.6 concludes by making the ontological commitments of this book explicit, preparing the ground for the practical analytical work that follows.
This chapter develops what ontology is and how it enables adequate social research. Rather than treating ontology as an abstract philosophical concern detached from empirical inquiry, the chapter shows how ontological analysis allows researchers to engage systematically with the complexity, instability, and apparent messiness of social reality. This messiness does not weaken scientific investigation; it is what makes explicit ontological reflection necessary. Chapter 1 traced philosophy’s wrestling with a fundamental problem: empiricism attempted to restrict science to observables, but Hume showed this generated scepticism about causation and external objects. Kant tried to rescue knowledge by arguing the mind actively structures experience, but this left things-in-themselves (noumena) forever unknowable behind the veil of perception (or the phenomena). Bhaskar’s critical realism appears to have found an approach to these problems that distinguishes what exists independently (intransitive dimension) from our changing knowledge about it (transitive dimension), showing how we can investigate reality through a stratified ontology without collapsing into either empiricist reduction or Kantian dualism. That approach is what will be discussed in depth in this chapter.
Building on that philosophical resolution, we now examine how ontological analysis becomes both possible and necessary for social research. Understanding ontology, the study of what exists and how different aspects of reality relate, proves essential because methodological choices cannot be separated from ontological commitments. But developing social ontology faces distinctive challenges: the social world is complex, constantly changing, conceptually mediated, and exists only through ongoing human activity. This complexity might suggest abandoning systematic ontological analysis in favour of purely epistemological or methodological approaches. However, the opposite proves true: precisely because the social world is complex and our knowledge of it is fallible, we need ontological frameworks that can distinguish what exists from what we currently know about it, preventing a collapse into relativism where competing theories become incommensurable or into positivism where existence is reduced to observability (Bhaskar, 1979; Sayer, 2000).
The chapter proceeds through five interconnected stages. It builds on the conceptual architecture of Chapter 2, which develops the definitions of ontology, epistemology, methodology, and methods as distinct but related dimensions, and which introduces the intransitive-transitive distinction as a solution to the epistemic fallacy. The present chapter takes up where that architecture leaves off. Section 3.2 examines stratified reality through the three domains of the Empirical, Actual, and Real, showing how this framework avoids both empiricist reduction and Kantian dualism through monist realist ontology. Section 3.3 addresses the problem of unobservables in social research directly, demonstrating how stratified ontology enables the investigation of mechanisms, structures, and emergent properties essential for adequate social explanation. Section 3.4 examines how the different traditions of literature review encode distinct ontological assumptions, and establishes ontological situating as the logical next step beyond those traditions. Section 3.5 introduces the four social ontological debates that structure social research and organise Part II of this book. Section 3.6 concludes by making the ontological commitments of this book explicit, preparing the ground for the practical analytical work that follows.
II Stratified Reality: The Three Domains and Monism
II.i From Dualism to Monism Through Stratification
This is for example one of the sections where critical realism is very important in understanding the conditions for a social science to be possible. Bhaskar’s critical realism resolves the Cartesian and Kantian problems through a distinctive move: rather than accepting dualism between knowable phenomena and unknowable noumena (Kant), or between mind and body (Descartes), critical realism proposes a monist realist ontology where reality is one substance (monism) but exists in stratified layers with different properties (stratification). This framework enables systematic investigation of unobservable mechanisms essential for social research while avoiding both empiricist reduction (collapsing reality to the observable) and Kantian dualism (treating noumena as forever inaccessible).
Monism means there is one reality, not two separate realms of phenomena and noumena, or mind and matter, that somehow must be connected. Cartesian dualism created the interaction problem: if mind and body are fundamentally different substances, how do they interact? Kantian dualism created the knowledge problem: if we can only know phenomena structured by our categories while things-in-themselves remain forever beyond reach, how can we claim scientific knowledge of reality? Historically, Spinoza has offered the primary monist position, which posits existence not as two substances of thought and extension/matter, but instead suggests that there is only one substance (monism) or reality, that includes both mind and matter as attributes or expressions of its existence. Similarly, critical realism avoids both problems through maintaining that all aspects of reality, including observable events, unobservable mechanisms, experienced phenomena, and underlying structures, are part of one ontological realm operating according to the same fundamental principles, though manifesting at different levels of reality, something we will go into detail in the next section.
This monism does not collapse everything to one level (as empiricism collapses everything to observables, or as some idealisms collapse everything to ideas). Rather, critical realism argues that reality is stratified. This means it is organized into levels or domains that are ontologically continuous (part of one reality) but analytically distinguishable (requiring different investigative approaches). The three domains, the Empirical, Actual, and Real, constitute analytical distinctions within unified reality rather than separate ontological realms requiring mysterious bridges between them (Bhaskar, 1975, 1979).
II.ii The Three Domains: Empirical, Actual, Real
Understanding stratified reality requires careful examination of how the three domains relate while remaining analytically distinct. The domains are presented here from most accessible (Empirical) to most fundamental (Real), moving from what we can directly observe to what must exist for observations to be possible. This movement from Empirical to Real also maps onto the architecture of social research introduced in section 3.2: the entities and mechanisms that social ontology identifies as existing in the social domain inhabit the Real, the events and processes that meta-theory and theory attempt to explain unfold in the Actual, and the empirical material that methods generate and analyse belongs to the Empirical domain. The three domains are therefore not a separate framework but a description of the depth structure of the intransitive dimension itself.
The Empirical domain encompasses everything we can observe, experience, or measure, constituting the realm of direct perception and documented evidence. When researchers collect data, they operate in the Empirical domain: survey responses documenting attitudes and behaviours, interview transcripts capturing expressed meanings, or ethnographic observations recording interactions. The Empirical consists of observed and experienced events that provide evidence requiring explanation rather than being explanations themselves.
Crucially, the Empirical domain is real: these are actual observations and experiences, not illusions or mere constructions. When we measure educational outcomes and find systematic gaps between social groups, these patterns are real features of the Empirical domain requiring explanation. But Empirical evidence alone cannot provide adequate explanation because it documents patterns without identifying the mechanisms generating them. In other words, seeing what is happening does not necessarily tell us why it is happening. So, knowing that students from different backgrounds attain certain grades at different rates does not tell us why this happens or what mechanisms produce these differences (Bhaskar, 1975).
The Actual domain encompasses everything that happens, whether or not we observe it. It includes all events, processes, and occurrences that take place in reality. The Actual includes observed events (which are part of the Empirical domain) but extends to unobserved events that occur regardless of whether anyone documents them. When institutional selection systems stream students into different educational pathways, these sorting processes occur in the Actual domain whether or not researchers observe them. When cultural advantages operate through everyday interactions that nobody documents, these interactions happen in the Actual domain despite leaving no systematic evidence.
The Actual domain is crucial because it reminds us that reality extends beyond what we happen to observe. Social processes operate continuously whether or not researchers study them, and many important events leave minimal Empirical traces while still having real effects. Discrimination, for example, operates through subtle patterns that may be invisible to casual observation. Understanding social phenomena requires recognising that Actual events generating observable patterns often remain partially hidden from direct investigation (Bhaskar, 1979).
The Real domain encompasses the underlying structures, mechanisms, and causal powers that generate events in the Actual domain, including both observed and unobserved events. The Real includes not just what happens but also what could happen given certain enabling conditions (so the possessed powers and tendencies of social arrangements that may or may not be activated). Institutional structures, cultural systems, power relations, and material conditions all exist in the Real domain. They possess causal powers that shape what becomes possible, even when those powers are not currently being exercised.
Understanding the Real domain requires recognising that mechanisms and structures exist independently of their manifestation in particular events. Educational institutions possess systematic properties such as selection criteria, tracking systems, and resource allocation patterns that persist as organisational arrangements even when they are not currently generating observable outcomes. These institutional properties constitute Real mechanisms: they have causal powers to shape opportunities, channel development, and generate systematic advantages and disadvantages. The mechanisms exist whether or not their effects are currently visible, and they may operate differently under different conditions while retaining their essential properties as causal structures (Bhaskar, 1975; Sayer, 2000).
It is important to be precise about the ontological status of each domain. All three domains are real in the sense that observations genuinely occur, events genuinely unfold, and mechanisms genuinely operate: none is illusory or merely constructed. However, they are not equally real in the same sense. Causal efficacy resides in the Real domain: it is the structures and mechanisms located there that possess the causal powers to generate events in the Actual domain and to make the observations of the Empirical domain possible. The Actual and Empirical domains are the products of causal powers exercised in the Real, not independent sources of causation. The causal arrow therefore runs from Real to Actual to Empirical, which is why adequate social explanation must move in the opposite direction, from observed patterns through actual events to the real mechanisms that generated them. This does not mean the relationship between domains is strictly one-directional in all theoretical frameworks. Depending on one’s social ontology and meta-theoretical commitments, feedback loops from the Actual or Empirical back into the Real are possible: social events and accumulated empirical patterns can under certain conditions transform the very structures and mechanisms that generated them. A strictly structuralist position, by contrast, would not admit such feedback, treating structures as the singular and unidirectional source of social causation. How one theorises these relationships between domains is therefore itself an ontological and meta-theoretical commitment, not a purely empirical question.
II.iii Relationships Between Domains
The three domains are nested rather than separate. Real mechanisms generate Actual events (whether observed or not), and Actual events provide the phenomena we observe in the Empirical domain. But the relationship is not mechanical nor deterministic. Real mechanisms possess causal powers operating as tendencies that make certain outcomes possible without guaranteeing specific results. Whether these mechanisms actually generate particular events depends on contingent conditions, on whether mechanisms are activated, and on how different mechanisms interact when multiple causal powers operate simultaneously.
Consider how institutional selection mechanisms (Real domain) operate. These mechanisms possess causal powers to systematically channel students from different backgrounds toward different outcomes. But whether these powers actually generate particular streaming patterns (Actual events) depends on many factors: how criteria are interpreted by decision-makers, how students and families respond to institutional signals, what resources different groups can mobilise, how cultural capital operates in specific contexts. And whether we observe these streaming processes (Empirical domain) depends on what gets measured, documented, and studied. Many selection events occur without leaving systematic traces that researchers can analyse.
This layered understanding reveals why adequate social research must investigate all three domains. Empirical patterns document that inequality exists and provide crucial evidence requiring explanation. Actual events reveal how mechanisms operate in specific instances, showing the concrete processes through which structures generate effects. Real mechanisms provide explanations for why patterns persist, identifying the causal powers that generate systematic outcomes across varying circumstances. No single domain is more important than the others, and adequate understanding requires analysing how all three relate (Danermark et al., 2002).
II.iv How Stratification Avoids Both Empiricism and Kantian Dualism
This stratified ontology resolves fundamental problems that plagued both empiricism and Kantian philosophy. Against empiricism, stratification shows why reducing science to observables proves inadequate: the Real domain contains unobservable structures and mechanisms that are nonetheless causally efficacious. Educational inequality cannot be explained simply by correlating observable variables. We need to identify underlying institutional mechanisms, cultural processes, and power relations that generate observable patterns. These mechanisms are real (they exist and have causal powers) even though we cannot observe them directly, and adequate explanation requires investigating them through their effects rather than limiting science to observable correlations (Bhaskar, 1975).
Against Kantian dualism, stratification demonstrates how we can investigate things-in-themselves rather than being trapped behind phenomena. Kant’s problem was that if our categories structure all experience, then we can never know whether our structured experience corresponds to reality-in-itself. Critical realism resolves this through recognising that structures and mechanisms operate in the Real domain whether or not we experience them, and we can investigate these realities through multiple forms of evidence that do not depend only on direct observation.
When we identify institutional mechanisms through comparing how they generate different patterns under different conditions, or when we trace cultural transmission through examining how advantages persist across generations, we are investigating Real structures rather than merely analysing our phenomenal experience. The crucial point is that reality provides resistance and constraint: our theories can be wrong, our predictions can fail, our interventions can produce unexpected results, because there are Real structures operating independently of how we conceptualise them. This constraint proves we are investigating reality-in-itself rather than simply studying our own mental constructions (Bhaskar, 1979; Collier, 1994).
III The Problem of Unobservables in Social Research
III.i Why Unobservables Matter for Social Explanation
Scientific explanation has often been defined in terms of what can be directly observed, measured, and verified. Social phenomena, however, operate through entities and processes that cannot always be directly observed but are nonetheless real and causally significant for understanding how social life works. Institutional structures, cultural meanings, power relations, and collective identities all involve unobservable features that shape observable outcomes in ways that cannot be captured by focusing only on what we can directly measure. The problem of unobservables thus arises not because such entities are obscure or marginal, but because empiricist standards of explanation systematically exclude them. This creates a fundamental challenge: how can social research investigate entities and mechanisms that cannot be directly observed without abandoning scientific rigour or collapsing into unfalsifiable speculation?
Consider organisational culture as an illustrative case. We can observe many aspects of organisational life: formal policies documented in handbooks, productivity rates measured through performance metrics, employee satisfaction captured through surveys, interaction patterns recorded through observation. But organisational culture itself, comprising the shared assumptions, tacit norms, and collective meanings that shape how people understand “how we do things here”, cannot be directly observed in the way we observe behaviours or measure outcomes.
Yet organisational culture clearly has causal effects. It influences how new employees learn organisational expectations, how decisions get made through unstated assumptions, how resistance to change persists despite rational arguments for new approaches. Understanding organisational performance therefore requires analysing cultural mechanisms that operate through but cannot be reduced to individual behaviours. Some organisations consistently outperform others despite similar resources and formal structures, suggesting that emergent cultural properties generate systematic advantages that individual-level analysis cannot capture. In this case, focusing only on observables misses the cultural mechanisms generating these persistent differences (Schein, 2010; Martin, 2002).
This example illustrates why unobservables prove essential for adequate social explanation. Many of the most important aspects of social reality, such as institutional arrangements, power structures, cultural systems, and collective meanings, operate through mechanisms and properties that we cannot directly observe but that nonetheless shape social outcomes in systematic ways. Excluding unobservables from scientific investigation because they resist direct observation would eliminate precisely the phenomena that social research needs to understand. The question becomes not whether to study unobservables, but rather how to investigate them rigorously (Bhaskar, 1979; Sayer, 2000).
This illustration of the blind-folded people and the elephant exemplifies the problem of unobservables. In the image, each person touches a different part of the elephant and, based on that limited contact, forms a different conclusion about what the object is. One feels the trunk and thinks it is a rope; another touches the leg and believes it is a pillar; a third feels the side and concludes it is a wall. Each description is based on real contact with the object, yet none captures the elephant as a whole. In social research, a similar problem arises when explanation is restricted to what can be directly observed. Researchers may accurately describe specific behaviours, patterns, or outcomes, but still miss the underlying structures that connect them. Social reality, like the elephant, cannot be grasped all at once through isolated observations. Understanding it requires moving beyond partial perspectives toward an account of the unobservable structures and mechanisms that organise what we see.
III.ii How Stratified Ontology Handles Unobservables
Critical realism’s stratified ontology provides a framework for investigating unobservables systematically by distinguishing what exists (Real domain) from what happens (Actual domain) and what we observe (Empirical domain). This distinction clarifies that unobservables are not unreal or mystical; instead they constitute aspects of reality that must be investigated through their effects rather than through direct observation. Understanding how this works requires examining the relationships between domains and the forms of inference that enable scientific investigation of unobservable mechanisms.
Unobservable mechanisms exist in the Real domain, where they possess causal powers that generate events in the Actual domain (whether observed or not). These Actual events leave traces, patterns, and manifestations in the Empirical domain that provide evidence for the operation of underlying mechanisms. Scientific investigation works backward from Empirical patterns through Actual events to Real mechanisms, asking what must exist for documented phenomena to be possible. This process (called retroduction) moves from effects to necessary conditions rather than from causes to predicted effects, enabling systematic investigation of unobservables through their manifestations (Danermark et al., 2002).
Consider how this works for studying institutional mechanisms generating educational inequality. Empirical evidence documents systematic patterns: students from different social backgrounds enter different educational pathways at different rates, achievement gaps persist across cohorts and institutions, and resource distributions systematically favour particular groups. These patterns require explanation: they are too systematic and persistent to result from random variation or individual choices alone.
Actual investigation reveals concrete processes through which sorting occurs: admissions decisions made using particular criteria, or tracking systems channelling students into differentiated programmes. Observing these actual events shows mechanisms operating in specific instances but raises questions about what institutional properties enable such systematic sorting.
A real analysis would theorise the underlying mechanisms that make these processes possible: for example selection systems with particular organisational properties, cultural capital operating through shared understandings, and institutional discrimination working through accumulated micro-decisions. These mechanisms possess causal powers explaining why systematic patterns persist, as they are not simply collections of individual acts but emergent institutional properties that shape what becomes possible and generate systematic effects across varying circumstances (Lareau, 2011; Bourdieu and Passeron, 1977).
IV Literature Review and Its Role in Science
Writing to Robert Hooke in 1675, Isaac Newton observed that if he had seen further than others it was by standing on the shoulders of giants. The remark has become so familiar as to risk losing its analytical force, but it contains a genuine epistemological claim: that knowledge is cumulative, that each generation of inquiry builds on what preceded it, and that understanding the present state of knowledge requires understanding the structure of what has been inherited. Every serious inquiry into social phenomena must do precisely this: situate itself in relation to what has already been established, understand what existing research has accomplished, and account for what justifies a new investigation in view of what already exists.
But Newton’s metaphor also conceals a problem that becomes acute in the social sciences. Standing on the shoulders of giants is not a neutral act. How one stands, which giants one has climbed, how one’s weight is distributed, what one can see from that particular vantage, is already shaped by commitments one brings to the ascent. Two researchers may engage the same body of existing work and reach systematically different conclusions about what it has established, where it falls short, and what kind of new inquiry is called for. This is not simply a matter of reading more carefully or more comprehensively. It reflects the fact that what counts as a finding, what counts as a gap, and what counts as an adequate response to that gap are not self-evident features of a literature but depend on the philosophical commitments through which one reads it. The literature review is the institutional form through which engagement with inherited knowledge is organized in social research. What that form requires, and what it can and cannot accomplish, has been understood very differently across the major traditions of social inquiry, and understanding those differences is a prerequisite for understanding what ontological situating adds and why it is necessary.
The major traditions of literature engagement in social science did not emerge simultaneously or in isolation from broader intellectual contexts. Each crystallized in response to particular pressures, practical, philosophical, and political, and each carries the marks of the historical moment in which it took shape. Recognizing this does not relativize the traditions or reduce them to mere period pieces; their analytical limitations are structural, not historical. But it does help to explain why each tradition has the particular shape it does, and why each found itself confronted with limitations it could not resolve on its own terms.
IV.i The Positivist Literature Review
The dominant tradition of systematic literature review in social science emerged from the positivist epistemological programme that shaped the discipline through much of the nineteenth and twentieth centuries. Positivism was itself a response to two developments: the remarkable success of the natural sciences in generating reliable, generalizable knowledge through systematic observation and experiment, and the aspiration of social theorists from Comte to Durkheim to place the study of society on an equally rigorous footing. If the natural world could be understood through the disciplined accumulation of observed regularities, then so, the argument went, could the social world. The literature review that followed from this aspiration treats existing research as a cumulative archive of empirical findings, each study contributing an increment of documented knowledge. The purpose of engagement is to synthesize those increments, identify where they converge on stable conclusions, and locate the gaps where sufficient evidence has not yet been accumulated. Meta-analysis represents the most technically refined expression of this approach, pooling results across studies through statistical aggregation to produce estimates of effect size with greater precision than any individual investigation could achieve (Cooper, 1998; Petticrew and Roberts, 2006). The underlying assumption is that with sufficient methodological rigour and enough studies, research converges progressively toward an accurate picture of how social phenomena work.
This approach handles certain tasks with genuine effectiveness. It is well suited to synthesizing empirical regularities, identifying where findings are robust across populations and contexts, and establishing a principled basis for determining where more evidence is needed. What it handles less well is research that asks not how often X co-occurs with Y but what mechanism makes X and Y tend to co-occur at all. When the question concerns underlying generative processes and causes rather than observable patterns, the positivist accumulation model lacks a clear procedure for synthesis or gap identification. More consequentially, when two studies reach contradictory conclusions not because one has been conducted better but because they operate from different ontological assumptions about what kind of thing they are investigating, the positivist review has no principled way to address the disagreement. It treats the contradiction as an empirical problem requiring more evidence rather than as a philosophical one in which the studies are examining different objects.
In short, the positivist literature review is a powerful instrument for synthesis within a shared ontological framework, but it has no resources for identifying or addressing disagreements that originate at the level of ontological commitment rather than empirical evidence. When the standing point is already settled, it can tell you how high you have climbed; it cannot tell you whether you are standing in the right place.
IV.ii The Interpretivist Review
The interpretivist tradition crystallized in the latter half of the twentieth century, partly as a direct response to positivism’s limitations and partly as an expression of broader intellectual currents, including hermeneutics, phenomenology, and symbolic interactionism, that had long insisted on the meaningfulness and irreducible particularity of social life. Where positivism sought to model social inquiry on the natural sciences, interpretivism argued that the social world is fundamentally different in character: its objects are meaning-laden, its phenomena are constituted through actors’ interpretations, and adequate understanding requires grasping the frameworks through which participants make sense of their situation. The interpretivist review is concerned not with what studies have established as fact but with how different researchers have made sense of social phenomena through particular theoretical frameworks, cultural positions, and interpretive commitments. Rather than synthesizing findings toward a convergent picture, it maps the diversity of perspectives and the range of meanings that have been constructed around a phenomenon, attending to how knowledge is situated and how the frameworks researchers use shape what becomes visible (Lincoln and Guba, 1985; Schwandt, 2000).
Interpretivist engagement is well suited to tracing how the same phenomenon has been understood differently across traditions and research communities, and it provides resources for recognizing when apparent disagreements reflect different interpretive frameworks rather than contradictory empirical claims. What it handles less well is the question of what independently existing social conditions make varying interpretations possible, or constrain which meanings can be sustained. Because it locates knowledge at the level of interpretation and meaning-making, it tends to remain at the level of what different researchers believe, construct, and argue, without raising the question of the independently existing structures and conditions that those interpretations are about. This is not a failure on the approach’s own terms, but it means that interpretivist engagement cannot assess whether different interpretive frameworks are better or worse suited to capturing aspects of a social reality that exists independently of any particular framework.
To return to Newton’s image: the interpretivist review is well suited to describing the different positions from which different researchers have chosen to stand, and to taking seriously the insight that what one sees depends on where one stands. What it cannot do is assess whether any of those positions offers a more adequate perspective on what is actually there to be seen.
IV.iii Immanent Critique
Immanent critique, rooted in the Hegelian and Marxist traditions, represents a different kind of engagement with existing work, one that arose not from dissatisfaction with positivism’s methods but from a philosophical commitment to the internal development of thought as the primary mode of critique. Rather than synthesizing findings or mapping interpretive diversity, immanent critique evaluates a theoretical framework against its own stated commitments, demonstrating internal contradictions that the framework cannot resolve without abandoning its own premises. Marx’s engagement with classical political economy is the paradigmatic case: by accepting the categories of classical economics and tracing their internal logic rigorously, Marx showed that the framework generates conclusions it cannot accommodate on its own terms (Marx, 1867/1976; Hegel, 1807/1977; Bhaskar, 1993). The strength of immanent critique is that it does not impose external evaluative standards but holds a framework to account for what it has itself committed to. This internal orientation is also, however, its limit from a realist standpoint: a framework evaluated only against its own commitments may be internally coherent while remaining entirely insulated from the question of whether those commitments correspond to anything in an independently existing social reality. Immanent critique can expose what a framework contradicts within its own terms; it cannot establish what that framework fails to reach beyond them. It is also well suited to identifying the ideological functions that theoretical frameworks can serve, namely how inconsistencies or silences within a framework correspond to interests it protects.
What immanent critique handles less well is precisely the situation that most concerns social ontological analysis: two research traditions operating from genuinely incommensurable ontological commitments, where there is no shared internal logic against which both can be measured. It also does not, by itself, provide a basis for identifying what aspects of social reality a framework structurally cannot access, as distinct from what internal contradictions it generates. The gap immanent critique reveals is normative and dialectical: the framework falls short of its own standard. This is a productive form of critique, but it is not the same as identifying what the framework cannot reach by virtue of what it ontologically presupposes.
Put differently, immanent critique can tell you when a framework has failed to live up to what it promised. It cannot tell you what the framework was constitutively unable to see, regardless of how faithfully it was applied. A framework may be internally consistent and still constitutively blind to significant dimensions of what exists.
IV.iv Genealogy
Genealogy, associated most closely with the work of Foucault in the 1970s and 1980s, emerged from a quite different intellectual context: the post-structuralist critique of both Enlightenment progressivism and Marxist teleology, and the diagnosis of knowledge itself as a site of power. Rather than asking how well a framework has accomplished its stated aims, genealogy asks how the framework came to define its aims in the way it did, and what historical contingencies, institutional arrangements, and relations of power produced it as authoritative. The genealogical method traces how concepts, categories, and forms of knowledge came to be established as legitimate, showing that what appears natural or necessary is the outcome of specific historical processes involving exclusions and contestations that could have developed otherwise (Foucault, 1977; 1980). The strength of genealogy lies in exposing how research traditions normalize particular questions while rendering others unaskable, and in destabilizing the apparent naturalness of prevailing categories.
Because genealogy deliberately suspends the question of whether any knowledge claim corresponds more or less adequately to an independently existing reality, it does not provide a basis for preferring one account of social phenomena over another on grounds other than the political or strategic. It also operates without a concept of the research gap in any conventional sense: its orientation is toward exposing and destabilizing what has been established rather than toward supplementing existing knowledge with what is missing.
The genealogical tradition is, in this sense, the most radical departure from Newton’s image. It does not ask whether we are standing in the right place or seeing further than our predecessors; it questions whether the very idea of a right place or a further view is anything other than a contingent historical effect. This is a powerful critical move, but it comes at a cost: by suspending the question of correspondence to an independent reality, genealogy also suspends the capacity to assess what any given ontological position enables and forecloses with respect to that reality. It can show how a framework came to occupy the ground it stands on; it cannot assess whether that ground gives access to what actually exists.
IV.v Ontological Situating as the Next Step
What this survey reveals, taken together, is a progressive expansion in what engagement with existing research is understood to accomplish. Each tradition can do something important that the others cannot: the positivist review at synthesizing empirical regularities, the interpretivist review at mapping theoretical diversity, immanent critique at exposing internal contradictions, genealogy at revealing the historical and political conditions of knowledge production. The recurring limitation across all four is that none provides a systematic procedure for identifying what any given research position cannot reach by virtue of its ontological commitments. When existing research produces contradictory findings, maps diverse interpretations, generates internal tensions, or normalizes certain questions, all of these may reflect underlying ontological commitments about what kind of thing is being studied, what relationship holds between observable phenomena and their conditions of existence, and what kinds of causal claims are appropriate. None of the four traditions, on its own terms, has a procedure for making those commitments the primary object of analysis. That is what ontological situating provides.
Ontological situating is the systematic practice of making explicit the ontological commitments embedded in research, both in existing scholarship and in the self-positioning of researchers embarking on a new investigation. The word systematic matters: the method does not rely on researchers announcing their philosophical positions, which they rarely do, but provides analytical procedures for identifying those positions from how research is actually conducted. What entities are treated as real and causally efficacious? What relationship is posited between observable patterns and whatever underlies or generates them? What kinds of causal claims are advanced? These questions can be asked of any piece of research, regardless of whether it has addressed them explicitly, and the answers reveal the ontological structure that governs what the research can see and what it cannot (Blaikie, 2007; Sayer, 2000; Danermark et al., 2002).
The method takes what is productive from each tradition surveyed above. From positivism it inherits the commitment to systematic and replicable procedure: ontological situating follows explicit analytical steps applicable across different studies and traditions. From interpretivism it takes the recognition that knowledge is always produced from a positioned perspective and that understanding what research claims requires understanding the framework through which those claims were produced. From immanent critique it takes the practice of evaluating frameworks against their own internal commitments, asking whether the ontological assumptions a study operates from are consistent with the conclusions it reaches. From genealogy it takes attentiveness to what has been excluded and what conditions have made particular forms of inquiry possible while foreclosing others. To all of these it adds a systematic focus on ontological structure: on what a framework posits to exist, how it understands the relationship between what exists and what can be known, and what causal architecture it assumes (Bhaskar, 1979; Collier, 1994; Lawson, 1997).
Ontological situating applies to both existing research and the self-positioning of researchers embarking on a new investigation, and the two applications are mutually reinforcing. When applied to existing research, it reveals what ontological commitments make a study’s findings possible and what they foreclose, specifically what questions can be asked from within the framework, what entities can be investigated, and what aspects of the phenomenon under study remain systematically out of reach. When applied to a new investigation, it enables deliberate and coherent positioning: rather than inheriting ontological assumptions from disciplinary convention or theoretical fashion, the researcher can make commitments that are appropriate to the phenomena being investigated and the questions being asked, and can anticipate in advance what those commitments will enable and what they will constrain. The concept of the research gap takes a different form within this framework: an ontological gap is not an absence of evidence or of interpretive perspective, but a structural foreclosure built into the research position itself, one that cannot be corrected by accumulating more evidence or adopting a more nuanced perspective, but only by repositioning.
V Social Ontology and the Four Ontological Debates
V.i The Distinctive Challenge of Social Ontology
Social ontology examines what exists in social reality and how different levels relate. But it also faces distinctive challenges because social reality is complex, contested, conceptually mediated, and exists only through ongoing human activity. Social ontology needs sophisticated frameworks precisely because of social reality’s complexity, which establishes why the ontological analysis developed through this book proves essential for adequate social research. The following chapters develop detailed examination of the four fundamental debates that emerge from core features of social ontology requiring systematic analysis (Bhaskar, 1979; Archer, 1995).
The social world manifests complexity through multiple interconnected dimensions that together create what we might call the ontological problem of social research, namely the challenge of developing systematic frameworks for investigating reality that is simultaneously structured and emergent, constraining and enabling, reproduced and transformed. Several core features of social reality create this challenge.
First, social reality is conceptually mediated: structures operate partly through actors’ understandings, cultural meanings shape behaviour, and institutional arrangements depend on normative interpretations. How inequality is understood affects how it operates; how institutions are interpreted shapes their effects; how power is recognised influences its effectivity. In this sense, social reality is partially mind-dependent: meanings and interpretations are not merely subjective representations but constitutive elements of how social mechanisms operate. This conceptual mediation is ontological (part of how social reality works), not merely epistemological (about how we study it). Institutional systems work partly through participants’ understanding of what criteria mean, through cultural assumptions about merit and appropriate futures, through normative frameworks about fair selection. These interpretations are themselves components of institutional mechanisms rather than merely our representations of mechanisms.
Second, social reality is historically developed: current arrangements emerge from past processes, and inherited patterns shape what is possible in the present. Path dependencies constrain present possibilities, structural legacies shape what becomes possible. In this respect, social reality exhibits a form of mind-independence: past social structures continue to shape present action regardless of contemporary actors’ intentions or awareness. Contemporary institutional systems embed historical traditions of social differentiation, accumulated assumptions about preparing different populations for different roles, and norms that have become institutionalised over time. The past remains present in several ways: through existential constitution (historical processes become embedded in current identities), co-presence (different temporal layers coexist), lagged efficacy (past causes have delayed effects), agentive perspectivity (all action occurs within inherited contexts). These forms of temporal embedding make temporal analysis essential for understanding social reality, more on this in Chapter 3.
Third, social reality is emergent: emergence refers to the fact that some effects or capabilities emerge from certain groups or combinations of elements that cannot be reduced to their components. This means that the effect is greater than the sum of its parts, the way a baked cake is more than just the sum of its ingredients. Similarly, collective properties arise from individual interactions but possess irreducible causal powers, institutional arrangements exercise effects that cannot be reduced to individual characteristics, cultural systems shape action in ways that transcend individual psychology. Although emergent properties depend on individual actions for their reproduction, they are not reducible to individual mental states or intentions. Institutional systems are not reducible to individual decision-makers’ choices even though they operate through such decisions. They possess organisational-level properties that shape what individual actors can do while being reproduced through their ongoing practices.
Fourth, social reality is reflexive: social research affects what is being studied, theoretical frameworks shape how actors understand situations, and knowledge about arrangements influences how they operate. When researchers document how institutional systems reproduce inequality, this knowledge can transform systems’ operation: administrators may reform procedures, decision-makers may resist sorting. This reflexivity means social science cannot achieve the fully observer-independence characteristic of natural science, making social ontology inherently more complex (Bhaskar, 1979; Sayer, 2000; Archer, 1995).
These features might appear to make ontological foundations unstable. Indeed, social arrangements change, actors reinterpret situations, new practices emerge, meanings shift, structures transform. If we start from epistemology rather than ontology, this instability seems to make systematic analysis impossible: we cannot get fixed knowledge of a moving target, as theories become relative to historical moments, frameworks succeed each other without cumulative progress. But this commits the epistemic fallacy by assuming that epistemic instability (our knowledge keeps changing) implies ontological instability (reality has no enduring structures).
Critical realism resolves this by distinguishing the intransitive dimension (social structures exist and operate with causal powers providing ontological referent) from the transitive dimension (our knowledge is historically developing and fallible). Social structures exist in the intransitive dimension. They operated before being theorised, they will continue operating (in some form) after current theories are superseded, their operation does not depend on theoretical adequacy. Institutional systems existed before sociologists analysed them, generated inequality before researchers documented effects, and operated according to institutional logic before being theorised.
But investigation proceeds through the transitive dimension where theories develop historically: we progressively understand institutional mechanisms better as research accumulates, earlier frameworks get refined or replaced, new insights emerge from ongoing investigation. The complexity of social reality makes ontological analysis harder but more necessary: without ontology distinguishing what exists from what we know, research collapses into relativism where competing theories become incommensurable (no way to compare or evaluate them) or into instrumentalism where theoretical concepts are merely useful fictions (having no referent in independent reality) (Bhaskar, 1979).
V.ii Four Ontological Debates in Social Research
The features of social ontology identified in section 3.5.1 generate four fundamental debates that structure the analysis developed in this textbook. These debates are not merely theoretical disagreements about how best to explain particular phenomena; they are ontological disagreements about what kinds of things exist in the social domain, what properties they possess, and how they relate to one another. Each debate arises because the complexity, emergence, conceptual mediation, and reflexivity of social reality make it genuinely unclear how certain foundational questions should be resolved, and different answers carry different implications for every level of the architecture introduced in section 3.2. The debates are introduced briefly here and developed in full in Chapters 5 through 8.
Individuals and groups. The first debate concerns the most basic question of social ontology: what are the primary constituents of social reality? Do social phenomena consist ultimately of individuals, their attributes, dispositions, and interactions, such that groups, institutions, and collective entities are nothing more than aggregates of individual properties? Or do supra-individual entities such as groups, classes, and communities exist in their own right, with properties and causal powers that are irreducible to the individuals who compose them? This debate connects directly to the emergence identified in section 3.5.1: if social reality is genuinely emergent, then collective properties cannot be reduced to individual-level characteristics, and the ontological status of groups becomes a substantive question rather than a merely analytical convenience. How researchers position themselves in this debate shapes what kinds of entities their theories can identify, what levels of reality their methods need to access, and what kinds of explanation count as adequate. This debate is examined in Chapter 5.
Structure and agency. The second debate concerns how enduring social arrangements relate to the actions of the individuals who inhabit them. Do social structures, institutions, norms, and cultural systems determine the actions of agents, such that individual behaviour is best understood as an effect of structural position? Or do agents possess genuine causal powers to act otherwise, to reproduce or transform the structures they encounter, such that structure and agency must be treated as genuinely distinct and mutually conditioning? This is perhaps the central debate in social ontology because it bears directly on how causality works in the social domain: whether explanation should move from structure to action, from action to structure, or whether both must be analytically separated and their interaction traced over time. The conceptual mediation and historical development of social reality identified in section 3.5.1 are both implicated here, since structures operate partly through agents’ interpretive frameworks while also persisting across generations independently of any particular agent’s intentions. This debate is examined in Chapter 6.
The material and the ideational. The third debate concerns what kinds of things exercise causal power in the social domain. Do material conditions, resource distributions, physical arrangements, and biological constraints provide the primary causal substrate of social life, with ideas, meanings, and cultural frameworks being either derivative of or reducible to material factors? Or do ideas, meanings, norms, and cultural systems possess independent causal powers that shape social outcomes in ways that cannot be reduced to material conditions? This debate is directly connected to the conceptual mediation of social reality: if meanings and interpretations are constitutive elements of how social mechanisms operate rather than merely representations of material processes, then the ideational cannot be treated as simply epiphenomenal. At the same time, the claim that ideas have independent causal power must be reconciled with the recognition that social life is always also materially embedded and that material constraints are genuinely real. This debate is examined in Chapter 7.
Transcendence and immanence. The fourth debate concerns the relationship between social reality as it is and the possibilities that exceed or transcend it. Does social reality contain within itself the generative principles of its own transformation, such that change is always immanent to existing structures and relations? Or are there principles, values, or potentials that transcend existing arrangements and provide standards for critique and transformation that cannot be derived from what already exists? This debate has a distinctive character among the four because it bears not only on ontological description but on the normative and critical dimensions of social research. It connects to the reflexivity identified in 3.5.1: if social research can contribute to the transformation of the arrangements it studies, the question of where the standards for that transformation come from is itself an ontological one. Unlike the other three debates, this one concerns an a priori assumption that researchers bring to their work rather than a question that can be resolved through empirical investigation, which makes it all the more important to make it explicit. This debate is examined in Chapter 8.
VI Conclusion: From Philosophical Foundations to Ontological Analysis
This chapter developed what ontology is and how it applies to social research by building on Chapter 1’s philosophical resolution of the subject-object problem. Chapter 1 established that Kant created an explicit dualism between unknowable noumena and knowable phenomena, that empiricism attempted resolution by eliminating unobservables but proved inadequate, and that Bhaskar’s critical realism resolved these problems through distinguishing intransitive reality from transitive knowledge.
This chapter developed those ontological foundations across four stages. It began by developing stratified ontology through the three domains of the Empirical, Actual, and Real, showing how this framework enables systematic investigation of unobservable mechanisms without collapsing into either empiricist reduction or Kantian dualism. It then addressed the challenge of unobservables in social research, demonstrating how stratified ontology enables the investigation of mechanisms, structures, and emergent properties that are essential for adequate social explanation. It examined the major traditions of literature review, from positivism through interpretivism and immanent critique to genealogy, showing that ontological situating follows from the cumulative limitations of those traditions as the logical next step. It then introduced the distinctive challenges of social ontology, including conceptual mediation, historical development, emergence, and reflexivity, and introduced the four ontological debates that these features generate.
The architecture developed in this chapter provides the analytical foundation for everything that follows. Each of the four ontological debates examined in Chapters 5 through 8 can now be understood not merely as a theoretical disagreement but as a question about what the social world contains in its Real domain, how those contents relate at the meta-theoretical level, and what their existence implies for the epistemological, methodological, and methods choices that flow from them. The quantitative and qualitative divide, which appeared in Chapter 2, section 2.1.1 as an apparently irresolvable methodological dispute, can now be seen for what it is: a symptom of unexamined ontological commitments operating at the wrong level of the architecture. Making those commitments explicit, subjecting them to critical scrutiny, and understanding how they constrain every subsequent dimension of research design is what ontological analysis enables. The chapters that follow develop the tools for doing this systematically.
References
Archer, M. S. (1995). Realist social theory: The morphogenetic approach. Cambridge University Press.
Archer, M. S. (1996). Culture and agency: The place of culture in social theory (Revised edition). Cambridge University Press.
Archer, M. S. (2003). Structure, agency and the internal conversation. Cambridge University Press.
Archer, M. S., Bhaskar, R., Collier, A., Lawson, T., & Norrie, A. (Eds.). (1998). Critical realism: Essential readings. Routledge. (online)
Berger, P. L., & Luckmann, T. (1966). The social construction of reality: A treatise in the sociology of knowledge. Anchor Books. (online)
Bhaskar, R. (1975). A realist theory of science. Leeds Books. (online)
Bhaskar, R. (1979). The possibility of naturalism: A philosophical critique of the contemporary human sciences. Harvester Press.
Bhaskar, R. (1993). Dialectic: The pulse of freedom. Verso.
Bourdieu, P., & Passeron, J. C. (1977). Reproduction in education, society and culture. Sage.
Carnap, R. (1928). Der logische Aufbau der Welt. Felix Meiner Verlag.
Cartwright, N. (1989). Nature’s capacities and their measurement. Oxford University Press.
Collier, A. (1994). Critical realism: An introduction to Roy Bhaskar’s philosophy. Verso. (online)
Danermark, B., Ekström, M., Jakobsen, L., & Karlsson, J. C. (2002). Explaining society: Critical realism in the social sciences. Routledge.
Dewey, J. (1938). Logic: The theory of inquiry. Henry Holt and Company.
Elder-Vass, D. (2010). The causal power of social structures: Emergence, structure and agency. Cambridge University Press.
Emirbayer, M., & Mische, A. (1998). What is agency? American Journal of Sociology, 103(4), 962-1023.
Giddens, A. (1984). The constitution of society: Outline of the theory of structuration. University of California Press.
Hayward, C., & Lukes, S. (2008). Nobody to shoot? Power, structure, and agency: A dialogue. Journal of Power, 1(1), 5-20.
Hempel, C. G. (1965). Aspects of scientific explanation and other essays in the philosophy of science. Free Press.
James, W. (1907). Pragmatism: A new name for some old ways of thinking. Longmans, Green, and Co.
Kuhn, T. S. (1962). The structure of scientific revolutions. University of Chicago Press. (online)
Lareau, A. (2011). Unequal childhoods: Class, race, and family life (2nd ed.). University of California Press.
Lawson, T. (1997). Economics and reality. Routledge.
Lukes, S. (2005). Power: A radical view (2nd ed.). Palgrave Macmillan.
Martin, J. (2002). Organizational culture: Mapping the terrain. Sage.
Pawson, R., & Tilley, N. (1997). Realistic evaluation. Sage.
Sayer, A. (2000). Realism and social science. Sage.
Schein, E. H. (2010). Organizational culture and leadership (4th ed.). Jossey-Bass.
Schwandt, T. A. (2000). Three epistemological stances for qualitative inquiry: Interpretivism, hermeneutics, and social constructionism. In N. K. Denzin & Y. S. Lincoln (Eds.), Handbook of qualitative research (2nd ed., pp. 189-213). Sage.
Suppe, F. (Ed.). (1977). The structure of scientific theories (2nd ed.). University of Illinois Press.