{ "background": "Previous studies on water treatment risk reduction in Uganda have relied on conventional frequentist models, which can be limited when analysing sparse or heterogeneous facility-level data. The replication of these studies is crucial for validating prior findings and assessing methodological robustness in a practical engineering context.", "purpose and objectives": "This study aimed to replicate a prior risk assessment using a Bayesian hierarchical framework to methodologically evaluate its advantages in quantifying uncertainty and integrating multi-level data from decentralised treatment systems.", "methodology": "We conducted a replication study using operational data from a network of treatment facilities. A Bayesian hierarchical model was specified: $yi \\sim \\text{Binomial}(ni, pi)$, $\\text{logit}(pi) = \\alpha{j[i]} + \\beta Xi$, $\\alphaj \\sim \\text{Normal}(\\mu\\alpha, \\sigma_\\alpha^2)$, with weakly informative priors. Posterior distributions were estimated using Markov chain Monte Carlo sampling.", "findings": "The Bayesian model yielded more conservative and precise estimates of risk reduction compared to the original study. The posterior median for overall log-odds reduction was 1.85, with a 95% credible interval of [1.62, 2.11], indicating high certainty. Facility-level random effects revealed substantial heterogeneity, with an estimated 30% of facilities showing negligible risk reduction.", "conclusion": "The replication confirms the core finding of significant risk reduction but demonstrates that a hierarchical Bayesian approach provides a superior methodological framework for engineering decision-making by formally quantifying uncertainty and variability.", "recommendations": "Future engineering risk assessments for water treatment should adopt hierarchical Bayesian models, especially for heterogeneous infrastructure systems. Regulatory monitoring programmes should prioritise identifying and investigating facilities with posterior estimates near zero effect.", "key words": "Bayesian hierarchical model, replication study, water treatment, risk assessment, uncertainty quantification, Uganda", "contribution statement": "This paper provides a novel methodological evaluation by demonstrating how Bayesian hierarchical modelling