Explainable AI for Predicting Blockchain Adoption Readiness in Government Agencies: A Hybrid SEM-Machine Learning Study
Main article
Abstract
This study develops and evaluates an explainable artificial intelligence framework for predicting blockchain adoption readiness in government agencies. The paper integrates a technology-organization-environment perspective with institutional trust, regulatory legitimacy, and public value concerns to explain why public organizations may be technically interested in blockchain yet unevenly prepared for implementation. A scenario-calibrated dataset of 418 agency-level observations is used to demonstrate a hybrid analytical design in which structural equation modeling validates latent readiness mechanisms, while machine learning models classify high-readiness agencies from construct scores and operational indicators. The SEM results show that regulatory legitimacy, technological capability, top management support, data governance, and institutional trust are the strongest direct predictors of adoption readiness. The machine learning comparison indicates that the hybrid SEM-ML stack achieves the best held-out performance, with higher AUC and F1 values than logistic regression, support vector machines, random forests, extra trees, and gradient boosting alone. Explainability analysis further reveals that regulatory clarity, data governance, top management support, interoperability readiness, and institutional trust have the greatest marginal contribution to readiness classification. The study contributes a transparent readiness analytics architecture for public-sector blockchain governance and argues that adoption readiness should be assessed as a sociotechnical and institutional condition rather than as a purely technological capability. Keywords: Explainable AI; Blockchain adoption; Government agencies; Public-sector innovation; Structural equation modeling; Machine learning; Technology-organization-environment framework; Regulatory legitimacy
