Governing Epistemic Risk in Data-Driven Decision-Making: A Conceptual Framework Integrating Data Governance, Platform Power, Algorithmic Governance and Information Integrity, with Application to Kenya

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Data-driven decision-making increasingly depends on long socio-technical chains in which data is collected, integrated, modelled, ranked and circulated before it reaches a decision-maker. Existing governance instruments focus mainly on safety, rights and compliance, and existing scholarship treats data governance, platform power, algorithmic governance and information integrity as separate problems. As a result, the reliability and legitimacy of the knowledge on which decisions rest — their epistemic quality — remains largely ungoverned. This conceptual paper addresses that gap. Using an integrative literature review and a model-type conceptual design, it synthesises datafication theory, platform capitalism, algorithmic governance and social epistemology into a single data-to-decision lens. It contributes (i) a taxonomy of thirteen epistemic risks located along the data-to-decision lifecycle; (ii) the Epistemic Risk Governance Framework (ERGF), which maps four governance pillars onto six actionable governance dimensions; (iii) seven testable propositions, including a weakest-link principle of epistemic governance; and (iv) a comparative analysis showing where the ERGF extends the NIST AI Risk Management Framework, ISO/IEC 42001, the EU Artificial Intelligence Act, the UNESCO Recommendation and Kenya’s emerging data and AI governance instruments. The framework is illustrated through Kenyan decision contexts and accompanied by a design science agenda for its empirical evaluation.