Let Them Theorize
Normalizing Subjective Priors in Management Science
Management science has spent decades locked in an unresolved debate about whether its devotion to theory helps or harms the field. The “theory fetish” stifles discovery (Hambrick, 2007). Bad theories actively destroy good management practice (Ghoshal, 2005). Organizational theories may not progress at all (Davis, 2015). Abandoning theory means surrendering to “dustbowl empiricism,” data without meaning (Suddaby, 2014). Bayesian epistemology might reconcile the warring camps by formalizing speculative theorizing as rigorous, transparent, and cumulative science.
The stakes extend well beyond the academy. Management theories performatively shape organizational reality. Agency theory’s assumption that managers act as self-interested opportunists produces the very behavior it predicts (Ferraro et al., 2005). The question of what kind of theorizing management science should pursue, therefore, determines which organizations and managers society will produce. Every major management theory arrived theory-first, built on conceptual speculation and subjective priors rather than data-driven discovery. That historical fact sits uncomfortably alongside a contemporary publishing regime that demands both novel theory and rigorous empirics in every paper, a double bind that suppresses the very kind of ambitious, speculative work that built the field.
The argument here does not reduce rich qualitative theorizing to statistical formalism. The Bayesian framework proposed below operates at the level of epistemological orientation, not at the level of methodological prescription. It offers a logic of cumulative knowledge-building that accommodates both quantitative and qualitative evidence. Nor does it propose Bayesianism as a solution to the institutional politics of peer review. Institutional barriers require institutional remedies. What Bayesian epistemology provides is a vocabulary and a formal architecture for making speculative theoretical commitments transparent, comparable, and revisable, thereby removing the most common intellectual objection to ambitious theorizing: that it amounts to mere opinion.
The Theory Fetish and Its Discontents
In 2005, 100% of articles in AMJ, ASQ, and Organization Science contained the word “theory,” averaging 18 mentions per article, compared to only 78% in finance, marketing, and accounting journals. The blanket requirement was counterproductive: “Our field’s theory fetish…prevents the reporting of rich detail about interesting phenomena for which no theory yet exists. And it bans the reporting of facts, no matter how important or competently generated, that lack explanation” (Hambrick, 2007, p. 1346).
Important empirical patterns cannot reach publication without theoretical frameworks, meaning that a 1930s epidemiologist who discovered the smoking-cancer link would face rejection from management journals for lacking theory. The requirement forces scholars to “latch onto” ill-fitting frameworks, producing contorted prose. Only 9% of theoretical presentations in AMR receive empirical testing, yielding an absurdly high ratio of ideas to tests. A simpler standard would ask: “Does this paper have a high likelihood of stimulating future research that will substantially alter managerial theory and/or practice?” (Hambrick, 2007, p. 1350).
Academic theories have also exerted “very significant and negative influences on the practice of management” (Ghoshal, 2005, p. 75). The first problem was the “pretense of knowledge” (borrowing Hayek’s phrase), importing methods from natural science that demand “partialization of analysis, the exclusion of any role for human intentionality or choice, and the use of sharp assumptions” (Ghoshal, 2005, p. 77). The second was an “ideology-based gloomy vision” in which the most influential theories embed negative assumptions about human nature. Agency theory posits that managers cannot be trusted. Transaction cost theory assumes opportunistic behavior as the default. “By propagating ideologically inspired amoral theories, business schools have actively freed their students from any sense of moral responsibility” (Ghoshal, 2005, p. 76). The 2008 financial crisis seemed to validate these warnings (Khurana, 2007; Pfeffer, 2005).
Foundational paradigms (population ecology, institutional theory, resource dependence, transaction costs) were all formulated in the 1970s and have accumulated increasing citations without replacement. “In soft psychology, theories rise and decline, come and go, more as a function of baffled boredom than anything else” (Meehl, 1978, as cited in Davis, 2015, p. 181). The scholarly career incentive system “can promote novelty rather than truth and impact rather than coherence” (Davis, 2015, p. 192), and big data combined with current incentives will yield a high volume of novel papers with sophisticated econometrics and no obvious prospect of cumulative knowledge development.
The Defense of Theory and What Constitutes a Legitimate Contribution
A legitimate theory requires four building blocks: What (which factors merit consideration), How (the relationships among them), Why (the underlying psychological, economic, or social dynamics justifying the model, “the most critical and fruitful element”), and Who/Where/When (boundary conditions) (Whetten, 1989, p. 491). Legitimate theory must explain causal mechanisms, not merely describe patterns.
Five things are commonly mistaken for theory: references, data, lists of variables, diagrams, and hypotheses. None qualify unless a logical argument explains why the empirical relationships exist (Sutton & Staw, 1995). But the products that fail as finished theory (lists, diagrams, approximations) represent legitimate “interim struggles” in theorizing and should not be confused with lazy work (Weick, 1995). The distinction between theory (a finished product) and theorizing (an ongoing process) reframes the debate: most work labeled “theory” in organizational studies constitutes approximation, and that approximation is acceptable so long as scholars acknowledge it. The broader tradition of sensemaking, enactment, and loose coupling exemplified a style of theorizing that was processual, interpretive, and resistant to physics-style formalization (Weick, 1995).
Theory serves three value propositions: as knowledge accumulation (capturing and summarizing phenomena to enable progressive science), as knowledge abstraction (providing perceptual lenses that structure experience, without which “humans cannot organize experience”), and as normative vision (allowing scholars to see the world as it might be, challenging dismal assumptions) (Suddaby, 2014). Abstract knowledge also protects jurisdictional legitimacy. “To cede theory means to give up legitimacy” (Suddaby, 2014, p. 408). The warning against the anti-theory camp was pointed: “Dustbowl empiricism is, of course, doomed to fail. Knowledge accumulation cannot occur without a conceptual framework. When explicit frameworks are pushed into the background, theory becomes implicit. Implicit theories are inherently dangerous” (Suddaby, 2014, p. 411).
Theoretical contribution can also be evaluated along two dimensions: originality (incremental to revelatory) crossed with utility (scientific to practical). Scholars have historically over-privileged originality at the expense of utility, and scientific utility at the expense of practical utility. The highest aspiration is “prescience,” an orientation toward anticipating what organizations and societies will need (Corley & Gioia, 2011, p. 26). The question is not “more or less theory” but “what kind of theory.”
The Epistemological Difference: Management Science and Its Object
Management science differs epistemologically from natural science in ways that fundamentally alter what theorizing can and should accomplish (Giddens, 1984; Flyvbjerg, 2001; Bhaskar, 1975).
Social science studies phenomena (Giddens, 1984). Management and economic theories do not merely describe reality but actively constitute it. When the Chicago Board Options Exchange opened in 1973, the Black-Scholes formula did not accurately predict option prices (deviations of 30-40% were standard). As traders adopted the formula and the exchange institutionalized it through automated systems, actual prices converged toward the model’s predictions, with deviations dropping to roughly 2% by 1978 (MacKenzie, 2006). The theory became true because people acted as if it were true. Three mechanisms drive this self-fulfilling dynamic in management: institutional design (theories shape organizational structures and reward systems), social norms (theoretical assumptions become normative expectations), and language (merely naming the same prisoner’s dilemma game “Wall Street Game” versus “Community Game” dramatically altered cooperation rates). “Theories can ‘win’ in the marketplace for ideas independently of their empirical validity to the extent that their assumptions and language become taken for granted” (Ferraro et al., 2005, p. 8).
Reality also cannot be reduced to observable correlations. A stratified ontology distinguishes three domains: the empirical (what we observe), the actual (what occurs, whether or not observed), and the real (the underlying generative mechanisms that produce events). Social structures, unlike natural structures, are activity-dependent, concept-dependent, and transient. Reducing reality to observation commits the “epistemic fallacy” (Bhaskar, 1975). This ontology has since been translated into practical methodology for organizational research (Fleetwood, 2005; Edwards et al., 2014).
Social science has tried and failed to produce episteme (universal, context-independent knowledge): “No predictive theories have been arrived at in social science, despite centuries of trying” (Flyvbjerg, 2001, p. 4). Social science’s distinctive strength lies instead in phronesis, the practical wisdom necessary for deliberating about values, power, and how societies ought to organize themselves. Abduction complements this orientation: “the process of forming explanatory hypotheses. It is the only logical operation which introduces any new idea” (Peirce, 1903/1998, p. 216). Unlike deduction (which tests) or induction (which generalizes), abduction generates novel explanations when existing theories fail. Incorporating abduction within the hypothetico-deductive tradition is particularly informative for researchers grappling with complex, evolving phenomena that rarely fit cleanly into prespecified models (Sætre & Van de Ven, 2021; Wickert et al., 2025).
Management theories are not neutral descriptions awaiting confirmation. They are interventions in the reality they study. Any epistemological framework adequate to management science must account for this reflexivity rather than treating it as a methodological nuisance.
Grand Theories Emerged from Speculation, Not Data
Every major management theory arrived overwhelmingly theory-first, built from conceptual reasoning, interdisciplinary synthesis, and subjective theoretical priors rather than systematic empirical observation.
Transaction cost economics emerged from PhD training at Carnegie Mellon (1960 to 1963), absorbing bounded rationality, organizational routines, and the architecture of complexity from Simon, March, and Cyert. The “Carnegie Triple” was: be disciplined, be interdisciplinary, have an active mind (Williamson, 2010, p. 456). TCE’s core concepts (opportunism, asset specificity, governance structures) derive from conceptual reasoning and interdisciplinary synthesis. Empirical testing followed the establishment of the framework (Williamson, 1975, 1985).
“The Iron Cage Revisited” ran 14 pages, contained no original empirical data, and proposed 12 testable hypotheses derived entirely from conceptual reasoning (DiMaggio & Powell, 1983). The paper synthesized three mechanisms (coercive, mimetic, and normative isomorphism) from conversations at Yale’s Institution for Social and Policy Studies, influenced by Weber, Meyer, and Rowan (1977), and Zucker (1977). It has since accumulated over 68,000 citations, making it one of the most cited papers in all of social science, entirely on the strength of its conceptual architecture (Powell & DiMaggio, 2023).
The resource-based view followed an even more extended trajectory of pure theorizing. The Theory of the Growth of the Firm (Penrose, 1959) was empirically informed but primarily theoretical. “A Resource-Based View of the Firm” (Wernerfelt, 1984) was a conceptual paper. The 1991 paper crystallizing the VRIN criteria (valuable, rare, inimitable, non-substitutable) contained no original empirical data (Barney, 1991). Meta-analyses confirming the RBV’s tenets arrived 15 to 20 years later. The tautological risk was noted early (Priem & Butler, 2001), but the theory’s influence was established long before empirical validation.
Population ecology began as an explicit theoretical transplant from biological ecology (Hannan & Freeman, 1977). Resource dependence theory was closer to a mixed model, but the core framework was still driven by conceptual synthesis of exchange theory, political economy, and open systems thinking (Pfeffer & Salancik, 1978).
TCE (1975), agency theory (Jensen & Meckling, 1976), population ecology (1977), institutional theory (1977/1983), and RDT (1978) all emerged within a single compressed period. The most influential theories emerged at disciplinary boundaries, through interdisciplinary synthesis rather than narrow empirical specialization. The contemporary publishing system, which demands rigorous empirics alongside theory, would likely have rejected every one of them.
A Thought Experiment: Institutional Theory Under Two Review Regimes
Consider how “The Iron Cage Revisited” would fare under two evaluation logics. Under the prevailing frequentist regime, an editor receives a 14-page paper with no original data, 12 hypotheses, and a theoretical argument synthesized from Weber, Meyer, and Rowan (1977), and Zucker (1977). Reviewer 1 notes the absence of any empirical test. Reviewer 2 acknowledges the elegance but demands at least a preliminary dataset confirming one of the isomorphism mechanisms. The paper received a rejection with encouragement to collect data and resubmit. The most consequential theoretical contribution in the history of organizational studies dies in the review process.
Now consider the same paper under a Bayesian epistemological framework. The three mechanisms function as structured priors over organizational behavior. Coercive isomorphism encodes a prior expectation: organizations subject to regulatory mandates will converge in form. The strength of that prior reflects accumulated theoretical reasoning about state power, legal compliance, and resource dependence on government funding. Mimetic isomorphism encodes a priori about uncertainty: organizations facing ambiguous environments will model themselves on perceived successful peers. Normative isomorphism encodes a priori about professionalization: common training and credentialing among personnel will homogenize organizational practices.
Each of these priors can be formalized not as a single parameter estimate but as a family of predictions over observable organizational characteristics, specifying the direction, magnitude, and boundary conditions of expected convergence. The paper’s theoretical architecture then functions as a pre-registered set of Bayesian priors, openly specifiable, subject to scrutiny, and structured in a way that subsequent empirical work can update. The 12 original hypotheses become testable predictions with explicit expected effect sizes derived from the strength of the theoretical reasoning supporting each mechanism. Bold priors (firm theoretical commitments about the dominance of mimetic isomorphism under high uncertainty, for instance) yield sharper predictions that gain more evidential support when confirmed and lose more when disconfirmed, exactly the incentive structure that rewards ambitious theorizing over incremental hedging.
This translation is not easy. Stating that “normative isomorphism encodes a priori about professionalization” compresses an enormous amount of conceptual work into a single sentence. Operationalizing that prior demands specifying what “professionalization” means observationally (credentialing rates, standard curricula, professional association membership), determining a baseline convergence rate against which to measure isomorphic effects, and eliciting from domain experts a plausible distribution over effect magnitudes. The field currently lacks the infrastructure to perform this translation routinely. Few doctoral programs teach prior elicitation methods. No established conventions govern how to report the justification linking a theoretical mechanism to a specific distributional form. Building this infrastructure will require the same kind of sustained methodological investment that the fields of structural equation modeling in the 1980s and hierarchical linear modeling in the 1990s devoted (Kruschke et al., 2012; McKee & Miller, 2015). The difficulty of the translation does not invalidate the epistemological argument. It identifies a capability gap that the field must close if it wants to make cumulative knowledge-building more than an aspiration.
The epistemological logic of the original contribution already mirrors the Bayesian structure: firm theoretical commitments, derived from synthesis and reasoning, offered to the field for subsequent empirical updating. What changed between 1983 and 2025 is not the quality of speculative theorizing but the institutional willingness to publish it without accompanying data. A Bayesian epistemological framework provides a formal vocabulary for defending exactly the kind of work that built the field.
An honest reckoning with the Bayesian proposal must also confront what might be called the relocation risk. The current theory fetish demands that scholars justify their hypotheses through elaborate theoretical argumentation. A Bayesian regime relocates that demand: scholars must now justify their priors. Reviewer 2 will not stop demanding 15 pages of prose; the prose will move from the “Hypothesis Development” section to the “Prior Specification” section. The theory fetish does not disappear. It migrates.
The advantage, however, is that in the Bayesian regime, the theory must pay rent. A theoretical justification that produces no distributional commitment, that refuses to specify direction, magnitude, or uncertainty, fails on its own terms. Under the current regime, a scholar can write pages of dense theoretical argumentation connecting constructs through elaborate causal chains, arrive at a hypothesis that says only “X is positively related to Y,” and submit the paper with no commitment about how significant the effect should be, under what conditions it should strengthen or weaken, or what pattern of data would cause the scholar to abandon the theory entirely. The Bayesian framework forbids this evasion. The prior is a commitment. It has mathematical consequences. The debate over whether the prior is well-justified is a more productive debate than the one management science currently conducts, because it forces theories to make predictions specific enough to be wrong.
Bayesian Epistemology as a Framework for Rigorous Speculation
The Core Logic
In Bayesian reasoning, probability represents the degree of belief rather than long-run frequency. The updating process (prior belief multiplied by the likelihood of data yields a posterior belief) maps directly onto the scientific process: existing theory (prior) encounters new evidence (data) to produce a revised understanding (posterior). The posterior from one study becomes the prior for the next, creating a formal mechanism for cumulative science, precisely what management research lacks (Davis, 2015).
In formal models, parameters correspond to meaningful theoretical variables, and “knowledge about the values parameters are likely to take is often available before data are observed, based on previous experience or on theoretical expectations” (Vanpaemel & Lee, 2012, p. 1031). Informative priors simultaneously decrease model flexibility and increase empirical content, making theories more falsifiable, not less. When priors encode theory, Bayesian model selection (via Bayes factors) directly evaluates competing theoretical frameworks (Vanpaemel & Lee, 2012).
Bridging Conceptual Priors and Statistical Priors
A legitimate objection arises here: the “priors” that animated transaction cost economics or institutional theory were nomological and mechanistic beliefs about human nature and social structure (bounded rationality, isomorphism), not probability distributions over parameter spaces. Conflating qualitative causal assertions with stochastic expectations risks a category error.
The bridge requires two moves. The first is recognizing that every theoretical commitment generates empirical predictions, and every empirical prediction has a corresponding probability structure. The assertion that asset-specific investments increase the likelihood of hierarchical governance was a directional claim with an implicit magnitude. A Bayesian prior does not replace the qualitative reasoning; it formalizes the downstream empirical expectations that the qualitative reasoning generates. The prior distribution does not capture “mimetic isomorphism” as a concept. It captures the observable predictions that follow from the concept: convergence rates, adoption timelines, and variance in organizational form within and across fields. The semantic content of the theory lives in the justification for the prior, not in the prior itself.
The second move treats models as tools for learning rather than claims about truth. Successful Bayesian practice is not inductivist but hypothetico-deductive: scholars propose models, check them against data, and revise them through iterative cycles of fitting, checking, and expansion. “Social-scientific data analysis is especially salient for our purposes because there is general agreement that, in this domain, all models in use are wrong, not merely falsifiable, but actually false” (Gelman & Shalizi, 2013, p. 10). The prior is a structured bet, not a photograph of reality. Its value lies in what it risks, not what it contains.
The distinction between subjective and arbitrary priors is therefore essential. The subjective/objective distinction is unhelpful; what matters is transparency, consensus, and correspondence to reality (Gelman & Hennig, 2017). Critics “strain on the gnat of the prior distribution while swallowing the camel that is the likelihood” (Gelman, 2008, p. 3). The choice of prior is no more subjective than the choice of statistical model. Well-formed priors can be constructed from meta-analyses, domain expertise, theoretical constraints, or previous empirical posteriors. Prior elicitation methods, including the SHELF framework, trial roulette, prior predictive checks, and regression-based elicitation, provide formal procedures for translating expert knowledge into probability distributions (O’Hagan et al., 2006).
The Performativity Problem: Does Bayesian Updating Become Circular?
Performativity poses a serious challenge to the Bayesian framework. If management theories shape the reality they describe (Ferraro et al., 2005; MacKenzie, 2006), then the data against which priors are updated may itself be an artifact of the theory. A Bayesian practitioner who holds a prior that managers are opportunists, designs a governance system reflecting that prior, and then observes opportunistic behavior has not learned anything about human nature. The posterior confirms the prior because the prior shaped the data-generating process.
This circularity, however, is not a Bayesian problem. It is a management science problem that afflicts frequentist inference equally. A null-hypothesis significance test on data generated by a performative theory is no less circular; it merely hides the circularity behind a veil of objectivity. The Bayesian framework makes the circularity visible. Because the prior is explicit and the updating process is transparent, a critic can identify precisely where the self-fulfilling dynamic enters the analysis. Frequentist inference buries theoretical assumptions in unstated model choices and treats the resulting p-value as theory-free evidence.
The Bayesian response to performativity, therefore, has two parts. First, prior predictive checks (simulating data from the prior and comparing them to observed patterns) can expose cases in which the theory generates data it expects, triggering model criticism rather than confirmation (Gelman & Shalizi, 2013). Second, the iterative Bayesian cycle of fit, check, and expand treats every posterior as provisional. When posterior predictive checks reveal that the model reproduces the data too well, or in the wrong ways, the practitioner revises the model rather than declaring victory. This discipline of self-suspicion does not eliminate performativity, but it provides a formal mechanism for detecting and correcting its distortions, something the prevailing frequentist regime does not offer.
Sensemaking and Bayesian Updating: The Synthesis
Theorizing as sensemaking and pragmatic Bayesianism describe the same epistemological process from different disciplinary vocabularies (Weick, 1995; Gelman & Shalizi, 2013).
Theorizing is an ongoing, provisional, self-correcting process in which scholars construct plausible accounts, test them against anomalies, and revise them through engagement with disconfirming evidence. “Interim struggles,” partial, approximate, deliberately incomplete frameworks offered to the field for collective refinement, are legitimate stages in this process (Weick, 1995). Pragmatic Bayesianism describes the same loop: the specification of a model (prior plus likelihood), its confrontation with data, the identification of misfit through posterior predictive checks, and the expansion of the model to address discrepancies (Gelman & Shalizi, 2013). Both accounts reject the notion of a final, settled theoretical product. Both treat the encounter between theoretical expectation and empirical evidence as generative rather than merely confirmatory.
The synthesis yields a specific contribution: Bayesian epistemology provides the formal architecture for sensemaking-as-theorizing. The structural correspondence between the two processes can be stated precisely:
Figure 1: Isomorphism Between Weickian Sensemaking and Bayesian Inference
Weickian SensemakingBayesian InferenceShared FunctionEnactment (constructing a plausible account of the environment)Prior Specification (formalizing theoretical expectations as a probability distribution)Committing to an initial interpretation of realitySelection (testing the account against experience and anomalies)Likelihood Evaluation (confronting the prior with observed data)Disciplined encounter with evidenceRetention (preserving the revised account for future use)Posterior Updating (revising the prior in light of data)Accumulating revised knowledgeAnomaly Detection (identifying where the retained account fails)Posterior Predictive Checking (identifying where the model misfit occurs)Triggering revision through disconfirmationIteration (re-enacting with the revised account)Next-Cycle Prior (using the posterior as the prior for subsequent analysis)Cumulative, self-correcting process
Both loops are iterative and provisional. Neither terminates in a final product. Sensemaking returns to enactment after retention fails; the Bayesian cycle returns to prior specification after predictive checks reveal misfit. The critical difference is that the Bayesian loop formalizes each stage, making the transitions between them explicit, comparable across studies, and subject to mathematical scrutiny.
The prior corresponds to the initial plausible account. The likelihood function corresponds to the evidential encounter. The posterior predictive check corresponds to the anomaly that triggers revision. What sensemaking describes as a social and cognitive process, Bayesian inference formalizes as a mathematical one, not to replace the qualitative judgment but to make its logic explicit, cumulative, and comparable across scholars and studies. Bayesian epistemology is, in this reading, the accumulated methodology of sensemaking.
Institutional Barriers and the Sociology of Theorizing
No epistemological framework can solve an institutional problem. The barriers to ambitious theorizing in management science are sociological, not statistical, and they require sociological remedies. The dominant mode of constructing research questions is “gap-spotting,” identifying understudied areas within existing paradigms. A review of 52 published articles found that gap-spotting dominated across paradigmatic camps, with three variants: confusion spotting, neglect spotting, and application spotting. “Problematization” (identifying and challenging assumptions underlying existing theories) is a methodology for generating more influential work. Still, gap-spotting remains more politically acceptable than undermining the assumptions on which prominent scholars have built careers (Alvesson & Sandberg, 2011).
Gap-spotting serves a legitimate function in normal science (Kuhn, 1962). Not every paper needs to overthrow a paradigm. Incremental refinement, application to new contexts, and clarification of boundary conditions constitute productive scholarly activity. The problem is not that gap-spotting exists but that it dominates to the near-exclusion of paradigm-challenging work. When 100% of articles in the top journals must contain “theory” (Hambrick, 2007), and the safest way to meet that requirement is to spot a gap in an existing framework, the system selects for incremental contributions. It selects against the ambitious speculation that built the field.
Business schools adopted an “inappropriate, and ultimately self-defeating, model of academic excellence” after the 1959 Ford and Carnegie Foundation reports called for more rigorous research. The result is a system where “faculty know more about academic publishing than about actual business problems” (Bennis & O’Toole, 2005, p. 98). Business schools exhibit exactly the isomorphic processes that institutional theory predicts: coercive pressures from accreditation bodies (AACSB, EQUIS), mimetic pressures from rankings, and normative pressures from disciplinary PhD training (Wilson & McKiernan, 2011).
The assumed dichotomy between basic and applied research is false. A matrix distinguishing pure basic research, pure applied research, and use-inspired fundamental research reveals a third option. Pasteur studied fermentation among French winemakers and discovered microbiology; he also studied livestock diseases and developed vaccines. Business schools are uniquely positioned for this kind of use-inspired basic research, producing work that is simultaneously theoretically fundamental and practically relevant. Most management research, however, pursues theoretical elegance without consideration of use (Stokes, 1997; Tushman & O’Reilly, 2007).
The field also suffers from remoteness from practical problems, failure to replicate findings, poor writing, “endless obscure theorizing,” and research fraud. The publish-or-perish culture encourages questionable research practices, including HARKing (Hypothesizing After Results are Known) and p-hacking (Tourish, 2019). Elite management journals ignore the significant problems facing humanity (Harley & Fleming, 2021).
Despite a decade of advocacy, Bayesian methods remain “essentially absent” across 10,000 or more articles in 15 organizational science journals (Kruschke et al., 2012). Doctoral training remains overwhelmingly frequentist, journal reviewers are unfamiliar with Bayesian methods, and no established norms for reporting results exist (McKee & Miller, 2015). Recent guides for strategy scholars are beginning to change this landscape and demonstrating substantively different conclusions from Bayesian versus frequentist analyses (McCann & Schwab, 2023; Certo et al., 2024).
A reviewer trained exclusively in frequentist null-hypothesis significance testing will likely view explicit priors as “biasing the results.” This objection confuses two distinct claims. The first (legitimate) claim is that a poorly chosen prior can dominate a posterior when sample sizes are small. The second (illegitimate) claim is that having any prior expectation at all constitutes bias. Every frequentist analysis embeds theoretical assumptions in model specification, variable selection, and functional form. These assumptions are priors in everything but name. The Bayesian framework does not introduce subjectivity into an otherwise objective process; it makes existing subjectivity visible and debatable. Whether the current reviewer pool will accept that argument is an institutional question, not an epistemological one.
The Field at a Crossroads, 2018 to 2025
AMR’s editorial team has produced a sustained stream of guidance on theory contributions and positioning (Barney, 2018, 2020) and the practical mechanics of writing theory papers, acknowledging that “most PhD students are trained extensively on how to write an empirical paper, but few are trained on how to write a conceptual paper” (Thatcher et al., 2021, p. 4). The AMR Origins Series (2023) captures the stories behind published articles to demystify theorizing for junior scholars.
The journal Organization Theory, launched circa 2020, is explicitly dedicated to theory development. Its inaugural issue traced how positivist scientism creates “mechanistic self-dehumanization” and called for approaches that embrace science while making it more humane (Petriglieri, 2020). A 2024 Journal of Management Studies debate argued that the field over-relies on describing “how things are” at the expense of theorizing “how things should be” and “how they can be achieved,” creating blind spots for grand challenges such as climate change and inequality (Hanisch, 2024).
Recent work has continued to clarify the meaning of theory through typification (Sandberg & Alvesson, 2021), defended construct clarity against calls for more action and less abstraction (Suddaby, 2024), and critiqued the masculinized discourse of theory-building in management studies (Cunliffe, 2022). A crucial distinction between unit theory (specific empirical frameworks) and programmatic theory (the settled science that unit theories collectively support) reveals that the field obsessively focuses on improving unit theory, without the architectural processes needed to integrate findings into programmatic knowledge (Cronin et al., 2021). Bayesian updating provides a formal mechanism for precisely this kind of cross-study integration.
Management lags behind psychology in adopting open science practices (pre-registration, open data, registered reports), although awareness is growing through the Responsible Research in Business and Management (RRBM) network. A fundamental tension exists between the open science movement’s emphasis on reproducibility and pre-registration and the ambitious theorizing community’s embrace of speculation, creative leaps, and initially fuzzy constructs. Pre-registered Bayesian analyses with explicit priors may resolve this tension, representing both transparency and theoretical commitment simultaneously.
Metaphor functions as a generative device in early-stage theorizing, enabling scholars to see unfamiliar phenomena through familiar conceptual lenses (Cornelissen, 2005; Cornelissen et al., 2008). Bayesian formalization operates downstream of metaphor, at the point where a metaphorical insight generates specific empirical predictions. The two are sequential rather than competing. Bayesianism does not demand immediate quantification of metaphorical constructs; it provides a framework for formalizing the empirical expectations that emerge once metaphorical reasoning has done its generative work.
Conclusion: Toward Transparent Ambition
The theory-building debate in management science is not a binary between more and fewer theories. It is a multidimensional contest over what kind of knowledge the field should produce.
The field’s most influential theories were products of ambitious conceptual speculation, not incremental empiricism. Transaction cost economics, institutional theory, the resource-based view, population ecology, and resource dependence theory all emerged theory-first, from interdisciplinary synthesis and subjective priors. A publishing system that would reject such work would be operating against its own interests.
Performativity means management theories are not passive descriptions but active forces shaping organizational reality (Ferraro et al., 2005; MacKenzie, 2006). Critical realism redirects attention from surface correlations to generative mechanisms (Bhaskar, 1975). Phronesis reframes the goal from prediction to practical wisdom (Flyvbjerg, 2001). Abduction legitimizes the creative leap from puzzling observation to novel explanation (Peirce, 1903/1998). These considerations determine which research questions the field treats as legitimate and which methods it accepts as rigorous.
Bayesian epistemology offers an underexploited bridge between the warring camps. Formalizing theoretical commitments as explicit prior distributions makes speculation transparent, testable, and cumulative. Bold theories (sharp priors) are rewarded when confirmed, creating incentives for theoretical ambition rather than incremental hedging. Prior elicitation methods provide formal procedures for translating expert knowledge and theoretical intuition into probability distributions that the scholarly community can scrutinize, debate, and update (O’Hagan et al., 2006; Vanpaemel & Lee, 2012). Bayesian updating is the formalized methodology of the iterative, self-correcting, provisional theorizing that sensemaking has always championed (Weick, 1995; Gelman & Shalizi, 2013).
Institutional reform remains necessary. Gap-spotting dominates because it is politically safe (Alvesson & Sandberg, 2011). Tenure systems reward quantity over paradigm-challenging quality. Business schools exhibit the very isomorphic pressures they theorize about (Wilson & McKiernan, 2011). Epistemological frameworks do not change institutions. What they can do is remove the intellectual cover for rejecting ambitious work. When a reviewer’s only objection to a speculative theoretical paper is “there are no data,” the Bayesian response is precise: the priors are the theory, and the data the field gathers over the next decade will update them. The question is whether the priors are well-formed, transparent, and sufficiently bold to warrant updating.
The field does not require different theoretical quantities. It needs different theorizing: more transparent in its assumptions, more ambitious in its scope, more honest about its performative effects, and more formally connected to cumulative knowledge-building. Bayesian epistemology, abductive reasoning, and phronetic social science collectively point toward a model of management scholarship that treats theorizing as what it has always been: informed, disciplined speculation about how organizations work and how they might work better.
References
Abbott, A. (1988). The system of professions: An essay on the division of expert labor. University of Chicago Press.
Alvesson, M., & Sandberg, J. (2011). Generating research questions through problematization. Academy of Management Review, 36(2), 247–271.
Barney, J. (1991). Firm resources and sustained competitive advantage. Journal of Management, 17(1), 99–120.
Barney, J. (2018). Editor’s comments: Positioning a theory paper for publication. Academy of Management Review, 43(3), 345–348.
Barney, J. (2020). Editor’s comments: From “That’s Interesting” to “That’s Important.” Academy of Management Review, 45(1), 1–4.
Bennis, W. G., & O’Toole, J. (2005). How business schools lost their way. Harvard Business Review, 83(5), 96–104.
Bhaskar, R. (1975). A realist theory of science. Leeds Books.
Callon, M. (1998). Introduction: The embeddedness of economic markets in economics. In M. Callon (Ed.), The laws of the markets (pp. 1–57). Blackwell.
Certo, S. T., Busenbark, J. R., Kalm, M., & LePine, J. A. (2024). Bayesian analysis in strategic management research. Strategic Management Journal, 45(2), 243–268.
Cornelissen, J. P. (2005). Beyond compare: Metaphor in organization theory. Academy of Management Review, 30(4), 751–764.
Cornelissen, J. P., Oswick, C., Christensen, L. T., & Phillips, N. (2008). Metaphor in organizational research: Context, modalities, and implications for research. Organization Studies, 29(1), 7–22.
Corley, K. G., & Gioia, D. A. (2011). Building theory about theory building: What constitutes a theoretical contribution? Academy of Management Review, 36(1), 12–32.
Cronin, M. A., Stouten, J., & van Knippenberg, D. (2021). The theory crisis in management research: Solving the right problem. Academy of Management Review, 46(4), 667–683.
Cunliffe, A. L. (2022). Must I grow a pair? On being a female scholar in management studies. Academy of Management Learning & Education, 21(1), 1–17.
Davis, G. F. (2015). Editorial essay: What is organizational research for? Administrative Science Quarterly, 60(2), 179–188.
DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48(2), 147–160.
Edwards, P. K., O’Mahoney, J., & Vincent, S. (Eds.). (2014). Studying organizations using critical realism: A practical guide. Oxford University Press.
Ferraro, F., Pfeffer, J., & Sutton, R. I. (2005). Economics language and assumptions: How theories can become self-fulfilling. Academy of Management Review, 30(1), 8–24.
Fleetwood, S. (2005). Ontology in organization and management studies: A critical realist perspective. Organization, 12(2), 197–222.
Flyvbjerg, B. (2001). Making social science matter: Why social inquiry fails and how it can succeed again. Cambridge University Press.
Gelman, A. (2008). Objections to Bayesian statistics. Bayesian Analysis, 3(3), 445–449.
Gelman, A., & Hennig, C. (2017). Beyond subjective and objective in statistics. Journal of the Royal Statistical Society: Series A, 180(4), 967–1033.
Gelman, A., & Shalizi, C. R. (2013). Philosophy and the practice of Bayesian statistics. British Journal of Mathematical and Statistical Psychology, 66(1), 8–38.
Ghoshal, S. (2005). Bad management theories are destroying good management practices. Academy of Management Learning & Education, 4(1), 75–91.
Giddens, A. (1984). The constitution of society: Outline of the theory of structuration. Polity Press.
Hambrick, D. C. (2007). The field of management’s devotion to theory: Too much of a good thing? Academy of Management Journal, 50(6), 1346–1352.
Hanisch, M. (2024). Prescriptive theorizing in management research: A new impetus for addressing grand challenges. Journal of Management Studies, 61(3), 1012–1038.
Hannan, M. T., & Freeman, J. (1977). The population ecology of organizations. American Journal of Sociology, 82(5), 929–964.
Harley, B., & Fleming, P. (2021). Not even trying to change the world: Why do elite management journals ignore the major problems facing humanity? Journal of Management Studies, 58(6), 1697–1704.
Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305–360.
Khurana, R. (2007). From higher aims to hired hands: The social transformation of American business schools and the unfulfilled promise of management as a profession. Princeton University Press.
Kruschke, J. K., Aguinis, H., & Joo, H. (2012). The time has come: Bayesian methods for data analysis in the organizational sciences. Organizational Research Methods, 15(4), 722–752.
Kuhn, T. S. (1962). The structure of scientific revolutions. University of Chicago Press.
MacKenzie, D. (2006). An engine, not a camera: How financial models shape markets. MIT Press.
March, J. G., & Simon, H. A. (1958). Organizations. Wiley.
McCann, B. T., & Schwab, A. (2023). Bayesian analysis in strategic management research: Time to update your priors. Strategic Management Review, 4(1), 75–109.
McKee, M. C., & Miller, D. (2015). Institutionalizing Bayesianism within the organizational sciences: A practical guide. Journal of Management, 41(2), 471–490.
Meehl, P. E. (1978). Theoretical risks and tabular asterisks: Sir Karl, Sir Ronald, and the slow progress of soft psychology. Journal of Consulting and Clinical Psychology, 46(4), 806–834.
Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83(2), 340–363.
O’Hagan, A., Buck, C. E., Daneshkhah, A., Eiser, J. R., Garthwaite, P. H., Jenkinson, D. J., Oakley, J. E., & Rakow, T. (2006). Uncertain judgements: Eliciting experts’ probabilities. Wiley.
Peirce, C. S. (1998). Pragmatism as the logic of abduction. In the Peirce Edition Project (Eds.), The essential Peirce: Selected philosophical writings, Vol. 2 (pp. 226–241). Indiana University Press. (Original lecture delivered 1903)
Penrose, E. T. (1959). The theory of the growth of the firm. Wiley.
Petriglieri, G. (2020). F**k science!? An invitation to humanize organization theory. Organization Theory, 1(1), 1–18.
Pfeffer, J. (2005). Why do bad management theories persist? A comment on Ghoshal. Academy of Management Learning & Education, 4(1), 96–100.
Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations: A resource dependence perspective. Harper & Row.
Powell, W. W., & DiMaggio, P. J. (2023). The iron cage redux: Looking back and forward. Organization Theory, 4(4), 1–19.
Priem, R. L., & Butler, J. E. (2001). Is the resource-based “view” a useful perspective for strategic management research? Academy of Management Review, 26(1), 22–40.
Sætre, A. S., & Van de Ven, A. H. (2021). Generating theory by abduction. Academy of Management Review, 46(4), 684–701.
Sandberg, J., & Alvesson, M. (2021). Meanings of theory: Clarifying theory through typification. Journal of Management Studies, 58(2), 487–516.
Stokes, D. E. (1997). Pasteur’s quadrant: Basic science and technological innovation. Brookings Institution Press.
Suddaby, R. (2014). Editor’s comments: Why theory? Academy of Management Review, 39(4), 407–411.
Suddaby, R. (2024). What theory is, what theory was: Whence and whither theory. Organization Theory, 5(4), 1–22.
Sutton, R. I., & Staw, B. M. (1995). What theory is not. Administrative Science Quarterly, 40(3), 371–384.
Thatcher, S. M. B., Fisher, G., & Criado, A. R. (2021). The nuts and bolts of writing a theory paper: A practical guide. Academy of Management Review, 46(4), 647–666.
Tourish, D. (2019). Management studies in crisis: Fraud, deception and meaningless research. Cambridge University Press.
Tushman, M. L., & O’Reilly, C. A. (2007). Research and relevance: Implications of Pasteur’s Quadrant for doctoral programs and faculty development. Academy of Management Journal, 50(4), 769–774.
Vanpaemel, W., & Lee, M. D. (2012). Using priors to formalize theory: Optimal attention and the generalized context model. Psychonomic Bulletin & Review, 19(6), 1047–1056.
Weick, K. E. (1995). What theory is not, theorizing is. Administrative Science Quarterly, 40(3), 385–390.
Wernerfelt, B. (1984). A resource-based view of the firm. Strategic Management Journal, 5(2), 171–180.
Whetten, D. A. (1989). What constitutes a theoretical contribution? Academy of Management Review, 14(4), 490–495.
Wickert, C., Cornelissen, J. P., Schultz, M., Gehman, J., & Haack, P. (2025). Abduction as a mode of theorizing: From puzzling observations to novel explanations. Journal of Management Studies, 62(1), 1–28.
Williamson, O. E. (1975). Markets and hierarchies: Analysis and antitrust implications. Free Press.
Williamson, O. E. (1985). The economic institutions of capitalism: Firms, markets, relational contracting. Free Press.
Williamson, O. E. (2010). Transaction cost economics: The natural progression. American Economic Review, 100(3), 673–690.
Wilson, D. C., & McKiernan, P. (2011). Global mimicry: Putting strategic choice back on the business school agenda. British Journal of Management, 22(3), 457–469.
Zucker, L. G. (1977). The role of institutionalization in cultural persistence. American Sociological Review, 42(5), 726–743.