Person–organization 'fit' has two incompatible faces: one measurable, one constructed. Automated matching and hiring tools that treat fit as mere attribute alignment risk overlooking identity work that shapes retention, fairness and productivity, so AI-driven systems and evaluations should incorporate methods that capture lived, interpretive fit.

Main Finding

PO (person–organization) fit research rests on two distinct, often unacknowledged ontologies: a correspondence ontology (aligned fit) that treats fit as measurable attribute alignment between person and organization, and a constitutive ontology (constructed fit) that treats fit as an enacted, interpretive, affective capacity to sustain a workable sense of self in organizational life. Each ontology carries different epistemological assumptions, process explanations, and methodological implications; recognizing both enables clearer theorizing and a pluralist research agenda.

Key Points

• Two ontologies:
  • Correspondence (aligned fit): Fit = degree of match/correspondence between stable traits/preferences/values of individuals and objective features of organizations. Fits naturally with quantitative measurement, alignment metrics, and predictive models linking match → outcomes.
  • Constitutive (constructed fit): Fit = dynamic, interpretive accomplishment (identity work, affective experience, sense-making) that allows individuals to live a coherent self within the organization. Focuses on processes, narratives, and lived experience rather than static alignment.
• Distinct epistemologies:
  • Correspondence: realist/positivist assumptions, observable attributes, measurement validity concerns focused on alignment metrics.
  • Constitutive: constructivist/interpretivist assumptions, meaning-making, situational contingencies, and subjective experience.
• Process models: The article develops separate processual explanations for how organizational participation leads to outcomes under each ontology, showing divergent causal pathways (e.g., trait alignment → satisfaction/retention vs. interpretive work → meaningfulness/adjustment).
• Methodological implications:
  • Researchers should explicitly state which ontology they adopt and design methods accordingly.
  • Aligned-fit work favors surveys, dyadic/market-matching metrics, and causal estimation; constructed-fit work favors qualitative methods, longitudinal/process tracing, and attention to interpretation and affect.
• Normative recommendation: adopt a pluralist agenda that preserves theoretical coherence within each ontology while expanding explanatory scope by integrating multiple approaches where appropriate.

Data & Methods

• Nature of the paper: conceptual/theoretical synthesis rather than an empirical study.
• Intellectual sources: draws on constructivist psychology and interpretivist organizational theory to critique predominant correspondence approaches and to build the constitutive account.
• Methodological output: development of explicit process models for each ontology and argumentation about epistemological and methodological consequences.
• No new quantitative datasets or experiments are reported; empirical validation is suggested as future work using methods matched to each ontology (e.g., survey-based alignment measures vs. qualitative longitudinal studies of identity work).

Implications for AI Economics

• Measurement and modeling of “fit” in AI-driven labor markets:
  • AI systems (hiring algorithms, recommender systems, matching platforms) often operationalize fit under a correspondence ontology (feature matching). This paper cautions that such operationalizations capture only one notion of fit and may miss constitutive, meaning-based dimensions that affect retention, performance, and welfare.
• Algorithmic hiring and fairness:
  • Reliance on observed trait-fit signals can entrench biases if those signals proxy cultural homogeneity rather than workers’ capacity to construct fit. AI economists should consider how automated matchers affect diversity of identities and whether they suppress identity work important for inclusion.
• Predicting turnover and productivity:
  • Predictive models trained on alignment measures may fail to predict outcomes driven by interpretive processes (e.g., identity adjustments, changing affect). Mixed-methods data (qualitative, longitudinal signals, temporal behavior) can improve prediction and causal understanding.
• Design of AI for workplace integration:
  • Tools that support constructed fit (onboarding assistants, adaptive task allocation, identity-affirming nudges, explainable matching recommendations) may yield different organizational outcomes than pure matching algorithms. Evaluation should include subjective wellbeing, meaning, and identity metrics, not only short-run performance.
• Market design and matching theory:
  • Standard matching models assume stable preferences and observable attributes; incorporating constitutive dynamics suggests preferences and suitability are endogenous and shaped by organizational experience. This affects equilibrium analysis, dynamics of labor reallocation, and welfare assessments.
• Policy and regulation:
  • Regulators should be aware that measurement-based fit assessments used by AI can have normative consequences for worker autonomy and inclusion. Policies might require disclosure of the ontology underpinning automated fit assessments and mandate mixed-evidence evaluation (including qualitative impact).
• Empirical agenda for AI economics:
  • Use pluralist methods: combine large-scale administrative/matching data with survey measures of subjective fit, repeated qualitative interviews to capture identity work, and causal inference techniques that respect processual timing (panel methods, event studies).
  • Investigate how AI-mediated matching changes the balance between alignment and construction (e.g., do automated matches reduce opportunities for identity work?), and how that mediates macro outcomes like mobility, inequality, and productivity.

In short, AI economics research and applications that treat person–organization fit as purely a measurable alignment risk missing important, constructed dimensions of fit; adapting models, algorithms, and evaluations to incorporate both ontologies will yield more accurate prediction, fairer systems, and richer policy insights.