Abstract
{ "background": "The adoption of advanced manufacturing systems in developing economies is critical for industrial growth, yet robust methodological frameworks for evaluating adoption rates are lacking. Existing approaches often fail to account for the hierarchical structure of plant-level data and inherent uncertainties in self-reported adoption metrics.", "purpose and objectives": "This paper proposes a novel Bayesian hierarchical model to provide a rigorous methodological framework for evaluating the adoption of manufacturing systems. The objective is to estimate true adoption rates while quantifying uncertainty and accounting for plant-level heterogeneity.", "methodology": "A three-level hierarchical model is developed, where the observed adoption status $y{ij} \\sim \\text{Bernoulli}(p{ij})$ is modelled with a plant-specific probability linked via a logit function to latent true adoption $\\thetai$, such that $\\text{logit}(p{ij}) = \\alphaj + \\betaj \\thetai$. Priors are placed on $\\thetai$, $\\alphaj$, and $\\betaj$ to complete the Bayesian specification. Inference is performed using Hamiltonian Monte Carlo.", "findings": "The framework, applied to a simulated dataset reflecting the Nigerian industrial context, demonstrates that naive estimates of adoption can be significantly biased. The model indicates a high posterior probability (exceeding 0.95) that true adoption rates for computer-aided design systems are overestimated by more than 15 percentage points when measurement error is ignored.", "conclusion": "The Bayesian hierarchical model offers a superior methodological approach for evaluating technology adoption in manufacturing, providing a mechanism to separate measurement noise from genuine adoption signals and yielding more reliable, uncertainty-aware estimates.", "recommendations": "Researchers and policymakers should adopt probabilistic modelling frameworks that explicitly account for measurement error and hierarchical data structures when assessing technological uptake. Future data collection efforts should be designed to inform such models.", "key words": "Bayesian inference, hierarchical modelling, technology adoption, manufacturing systems, measurement error, industrial policy", "contribution statement": "This paper introduces a novel Bayesian hierarchical model