Abstract
{ "background": "The optimisation of capital-intensive industrial machinery fleets is a critical engineering management challenge in developing economies. A previously proposed Bayesian hierarchical model offered a novel framework for evaluating the cost-effectiveness of such fleets, but its methodological robustness and applicability in a real-world context required independent verification.", "purpose and objectives": "This study aimed to replicate and methodologically evaluate the published Bayesian hierarchical model for machinery fleet cost-effectiveness. The primary objective was to assess the model's computational reproducibility, sensitivity to prior specifications, and the stability of its inferences when applied to operational data from Kenya.", "methodology": "We executed a full computational replication using the original algorithms and a new dataset comprising maintenance logs, fuel consumption records, and productivity metrics from a fleet of earth-moving equipment. The core model, $y{ij} \\sim \\text{Normal}(\\mu + \\alphai + \\betaj, \\sigma^2)$, with $\\alphai \\sim \\text{Normal}(0, \\tau\\alpha)$ and $\\betaj \\sim \\text{Normal}(0, \\tau_\\beta)$, was re-implemented. Sensitivity analyses were conducted using alternative prior distributions (e.g., half-Cauchy for variance components).", "findings": "The replication confirmed the model's structural logic but revealed a substantive sensitivity to the choice of priors on variance parameters. Specifically, the 95% credible interval for the cost-effectiveness ratio of loader fleets widened by approximately 40% under more conservative, weakly informative priors compared to the original specification. This indicates that original conclusions were moderately dependent on prior assumptions.", "conclusion": "The Bayesian hierarchical model provides a valuable analytical structure, but its outputs require careful interpretation with explicit reporting of prior sensitivity. The study underscores the importance of methodological transparency and robustness checks in engineering economic models.", "recommendations": "Future applications of the model should mandate and report comprehensive sensitivity analyses. Practitioners should adopt robust or weakly informative priors as a default to guard against over