Bridging Epistemologies: Integrating Faith into STEM Education and Its Impact on Student Learning Outcomes

Abstract

Background: Global STEM reforms increasingly privilege technical proficiency while sidelining moral teleology, leaving graduates adept with code but adrift on questions of purpose. This study advances a Scripture‑anchored, interdisciplinary framework that reunites scientific inquiry with a Christian worldview and empirically tests its influence on student learning. Methodology: Adopting a qualitative‑dominant mixed‑methods design, we executed an AI‑assisted meta‑synthesis of 180 peer‑reviewed sources (2019‑2025) and triangulated those insights with multi‑site case studies of Wheaton College (USA), Daystar University (Kenya), and Redeemer University (Canada). Data—curricula, alumni surveys, and classroom artefacts—underwent thematic coding and cross‑case pattern matching, generating testable propositions linking faith‑based instruction, epistemic coherence, and learning outcomes. Results: Across contexts, faith‑integrated STEM programmes exhibited markedly greater cognitive engagement, deeper ethical reasoning, and heightened spiritual formation than secular comparators. Meta‑synthesised effect sizes (Hedges g = 0.35–0.55) show that explicit biblical framing elevates moral‑reasoning and purpose metrics, with the most pronounced gains among students previously unfamiliar with faith‑learning integration. Key mediators include faculty integration skill, mission‑aligned accreditation standards, and culturally attuned pedagogy. Conclusions: Positioning sola Scriptura as the ultimate epistemic authority does not dilute scientific rigour; it re‑humanises STEM by welding empirical facts to transcendent purpose. The proposed framework supplies educators, policymakers, and accreditation bodies with a scalable roadmap for producing technically competent, ethically resilient graduates equipped to steward AI, biotechnology, and green‑tech frontiers. Future research should deploy longitudinal, multi‑institutional designs and neurocognitive measures to further validate the model’s causal pathways and global applicability.