{ "background": "Systemic risk assessment in manufacturing plants within developing economies is often hampered by sparse, heterogeneous data and complex interdependencies between mechanical, electrical, and human factors. Traditional reliability models struggle to integrate these diverse data sources and quantify epistemic uncertainty effectively.", "purpose and objectives": "This study presents and evaluates a novel Bayesian hierarchical modelling framework designed to quantify risk reduction in plant systems. The primary objective is to provide a robust methodological tool for engineers to integrate multi-level operational data and infer the efficacy of implemented safety interventions.", "methodology": "The proposed model structures plant risk hierarchically, with system-level failure rates modelled as a function of subsystem performance and intervention status. The core specification is $\\lambda{ij} \\sim \\text{Gamma}(\\alphaj, \\betaj)$, where $\\alphaj, \\beta_j$ are subsystem-specific parameters informed by plant-level covariates through log-linear regressions. Inference was performed using Hamiltonian Monte Carlo sampling on data from multiple Ugandan manufacturing sites.", "findings": "The model successfully synthesised disparate data, revealing that electrical subsystem interventions yielded the greatest marginal risk reduction, with a posterior probability of 0.92 that the true reduction exceeded 30%. Parameter estimates showed robust convergence, with 95% credible intervals for key hyperparameters excluding zero.", "conclusion": "The Bayesian hierarchical framework offers a statistically rigorous and practically viable methodology for risk assessment in data-scarce environments. It formally incorporates uncertainty, providing a more nuanced understanding of intervention impact than point-estimate approaches.", "recommendations": "Plant engineers should adopt probabilistic risk models that explicitly account for data hierarchy and uncertainty. Further research should focus on developing prior distributions for similar industrial contexts to enhance model portability and facilitate faster implementation.", "key words": "Bayesian inference, hierarchical modelling, risk assessment, manufacturing systems, reliability engineering, uncertainty quantification", "contribution statement": "This paper introduces a novel hierarchical Bayesian model specifically tailored for integrated risk assessment in manufacturing systems with sparse, multi-source data, demonstrating