Inclusive education is increasingly positioned as a lever for both equity and economic participation. Yet in contexts shaped by superdiversity—where disability, migration background, socio-economic vulnerability, gender, and linguistic difference intersect—education systems often rely on linear policy assumptions that underestimate the non-linear, emergent dynamics of exclusion. This paper applies complexity systems theory to explain why inclusive education reforms may fail to reduce persistent NEET (Not in Employment, Education, or Training) trajectories, even when policy commitments appear strong on paper.

We argue that exclusion and NEET outcomes are not the result of isolated deficits but emerge through feedback loops across multiple system levels: school-level practices (e.g., tracking, resource allocation, teacher beliefs), family constraints (e.g., economic precarity, caregiving burdens), labor-market signals (e.g., credential inflation, employer discrimination), and public narratives (e.g., deficit discourses about disability and diversity). These interactions produce tipping points and path dependencies that lock vulnerable youth—particularly those with disabilities, neurodivergence, or migration backgrounds—into trajectories of educational disengagement and labor-market exclusion.

The study positions Turkey as a "middle-income laboratory" to generate insights relevant for low-income contexts (LICs), where institutional capacity and resources are constrained but demographic and social complexity is high. Turkey's education system exhibits characteristics common to many LICs: rapid urbanization, large-scale internal and external migration, uneven implementation of inclusive education policies, and high youth unemployment. By analyzing Turkey's experience through a complexity lens, we aim to identify transferable lessons for LICs facing similar challenges.

We synthesize research across four domains: (i) complexity theory and education governance (e.g., Byrne & Callaghan, 2014; Mason, 2008), (ii) superdiversity and disability (Vertovec, 2007; Artiles et al., 2011), 

(iii) self-determination and agency as transformative mechanisms (Shogren et al., 2015), and (iv) NEET-related educational risk pathways (Eurofound, 2016; Elder, 2015). This synthesis reveals a critical gap: while inclusive education research has documented barriers at the individual and school levels, it has rarely examined how these barriers interact across system levels to produce aggregate outcomes like NEET rates.

Our conceptual contribution is a systems model that maps the mechanisms linking inclusive education practices to NEET risk under conditions of superdiversity. The model identifies key leverage points where interventions may have disproportionate impact: self-determination supports, transition planning, school-family collaboration, and institutional responsiveness to diversity. Importantly, the model also highlights unintended consequences of well-intentioned policies (e.g., how mainstreaming without adequate support can increase dropout risk) and threshold effects (e.g., how small increases in teacher capacity can trigger large improvements in student outcomes once a critical mass is reached).

We conclude with implications for policy design: inclusive education strategies should explicitly account for non-linearities, anticipate unintended consequences, and invest in adaptive mechanisms that improve educational-to-work transitions. This perspective strengthens the link between inclusion and structural transformation by making visible how "ignored complexity" translates into resource loss and stalled system change. By reframing inclusion as an adaptive system capacity rather than a checklist of interventions, we offer a new lens for understanding why some education systems successfully convert diversity into economic dynamism while others remain trapped in cycles of exclusion.

This paper is designed as a conceptual and integrative review leading to a testable systems model. The methodology consists of three phases: Phase 1: Structured Literature Mapping We conducted a systematic search of peer-reviewed literature (2010–2025) across four databases: ERIC, Scopus, Web of Science, and Google Scholar. Search terms combined three concept clusters: (i) inclusive education OR special education OR disability, (ii) complexity theory OR systems thinking OR non-linear dynamics, and (iii) NEET OR youth unemployment OR school-to-work transition. We prioritized studies from middle-income and low-income contexts, with particular attention to Turkey, Jordan, and comparable settings. Inclusion criteria required empirical or theoretical engagement with system-level dynamics (not just individual-level outcomes). We identified 87 relevant studies, which were coded thematically using NVivo software. Phase 2: Mechanism Identification and Causal Loop Diagramming Drawing on complexity systems methodology (Sterman, 2000; Meadows, 2008), we extracted causal mechanisms from the literature and organized them into feedback loops. For example: inadequate teacher training → low-quality inclusive practices → student disengagement → family withdrawal from school collaboration → reduced institutional accountability → continued inadequate training (reinforcing loop). We constructed causal loop diagrams to visualize these dynamics and identify leverage points. This phase was iterative, involving cross-validation among the three authors to ensure theoretical coherence and empirical grounding. Phase 3: Model Development and Validation Logic We synthesized the causal loops into an integrated systems model with four subsystems: (i) school practices, (ii) family resources and agency, (iii) labor-market structures, and (iv) policy and public discourse. The model specifies hypothesized relationships (e.g., "increased self-determination supports reduce NEET risk via improved transition planning") and identifies testable predictions (e.g., "systems with higher teacher capacity variability will show non-linear NEET outcomes"). While this paper does not test the model empirically, we outline a validation roadmap for subsequent work using accessible secondary data: national education statistics (e.g., dropout rates, special education enrollment), labor force surveys (e.g., NEET rates by demographic group), and policy documents (e.g., inclusive education implementation plans). We also propose qualitative validation through stakeholder interviews (teachers, families, youth) to assess face validity and refine causal assumptions. This approach prioritizes feasibility and transparency: we do not claim to have collected new data, but rather to have synthesized existing evidence into a novel framework that can guide future empirical research and policy experimentation.

This paper makes three contributions to inclusive education research and policy. First, we demonstrate that complexity systems theory offers a powerful lens for understanding why inclusive education reforms often fail to deliver expected outcomes. By mapping feedback loops and non-linear dynamics, we reveal how well-intentioned interventions can produce unintended consequences (e.g., mainstreaming without support increasing dropout risk) and how small changes at leverage points (e.g., teacher capacity, self-determination supports) can trigger disproportionate system-wide improvements. Second, we reframe superdiversity not as a static demographic characteristic but as a dynamic system property that amplifies non-linearity. In superdiverse contexts, the interactions among disability, migration, socio-economic status, gender, and language create exponential rather than additive complexity. This insight challenges linear policy models (e.g., "add more resources → improve outcomes") and calls for adaptive, context-responsive strategies. Third, we position Turkey as a middle-income laboratory for generating insights relevant to low-income countries. Turkey's combination of rapid demographic change, uneven policy implementation, and high youth unemployment mirrors challenges faced by many LICs. By analyzing Turkey's experience through a complexity lens, we identify transferable lessons: the importance of investing in teacher capacity as a leverage point, the risks of policy-practice gaps, and the need for transition supports that bridge education and labor markets. For policy, our model suggests that inclusive education strategies should: (i) explicitly account for feedback loops and threshold effects, (ii) invest in adaptive mechanisms (e.g., school-family collaboration, transition planning) rather than static inputs, and (iii) monitor unintended consequences through system-level indicators (e.g., NEET rates, dropout patterns). For research, we offer a testable framework and a roadmap for empirical validation using accessible secondary data. Future work should examine how the identified mechanisms operate in different contexts and test interventions at the leverage points we have identified.

Artiles, A. J., Kozleski, E. B., Dorn, S., & Christensen, C. (2011). Learning in inclusive education research: Re-mediating theory and methods with a transformative agenda. Teachers College Record, 113(10), 2122–2155. Byrne, D., & Callaghan, G. (2014). Complexity theory and the social sciences: The state of the art. Routledge. Elder, S. (2015). What does NEETs mean and why is the concept so easily misinterpreted? ILO Technical Brief No. 1. International Labour Organization. Eurofound. (2016). Exploring the diversity of NEETs. Publications Office of the European Union. Mason, M. (Ed.). (2008). Complexity theory and the philosophy of education. Wiley-Blackwell. Meadows, D. H. (2008). Thinking in systems: A primer. Chelsea Green Publishing. Shogren, K. A., Wehmeyer, M. L., Palmer, S. B., Forber-Pratt, A. J., Little, T. J., & Lopez, S. (2015). Causal agency theory: Reconceptualizing a functional model of self-determination. Education and Training in Autism and Developmental Disabilities, 50(3), 251–263. Sterman, J. D. (2000). Business dynamics: Systems thinking and modeling for a complex world. McGraw-Hill. Vertovec, S. (2007). Super-diversity and its implications. Ethnic and Racial Studies, 30(6), 1024–1054.