Developing an Original Theoretical Framework for Responsible Data Science Practice in Nigerian Public Sector Contexts

The deployment of data science in Nigerian public sector decision-making raises complex ethical, governance, and accountability questions that existing responsible AI frameworks developed for Western contexts inadequately address. This study developed an original theoretical framework for responsible data science practice in Nigerian public sector contexts. A theory-building methodology combining systematic review, empirical grounding, and expert validation was employed. Systematic review of 76 responsible AI and data governance frameworks from 2019 to 2024 confirmed that frameworks inadequately theorise the specific accountability deficit of weak institutional environments, the democratic legitimacy gaps in algorithmic government decisions, and the data colonialism dimensions of dependency on foreign AI infrastructure. Empirical grounding through structured interviews with 40 Nigerian government data officials, academics, civil society representatives, and international organisation specialists generated findings on the practical ethics of data use in Nigerian public administration. The original Responsible Data Science Framework for Nigerian Government proposes five principles: institutional accountability (clear authority and audit trails for algorithmic decisions), epistemic humility (mandatory uncertainty communication in model outputs), data sovereignty (preference for Nigerian-hosted and managed data systems), community co-design (inclusive consultation in system design for citizen-facing applications), and harm anticipation (prospective impact assessment for vulnerable populations). Each principle includes implementation criteria, governance mechanisms, and assessment indicators. Expert validation by 22 data ethics and governance specialists confirmed the framework's original theoretical contribution and practical applicability to Nigerian public sector contexts.

Keywords: responsible data science, public sector AI ethics, Nigeria, data governance framework, original contribution