Abstract
{ "background": "The original quasi-experimental study proposed a novel method for measuring the adoption rates of predictive diagnostics in industrial machinery fleets. Its findings, suggesting high potential uptake, have influenced maintenance policy discussions, yet its methodological robustness in a real-world operational setting required verification.", "purpose and objectives": "This study aimed to replicate the original quasi-experimental design to evaluate its methodological rigour and empirical validity for measuring technology adoption in an industrial engineering context. The objective was to test the stability of the original effect estimates and the feasibility of the field implementation protocol.", "methodology": "We executed a direct replication of the stepped-wedge, quasi-experimental design across a comparable sample of Nigerian industrial sites. Adoption was measured via logged system usage data. The primary analysis estimated the intervention effect using a generalised linear mixed model: $\\logit(P(Y{ij}=1)) = \\beta0 + \\beta1 T{ij} + ui + e{ij}$, where $u_i \\sim N(0, \\sigma^2)$. Robust standard errors were clustered at the site level.", "findings": "The replication yielded a statistically non-significant intervention effect (\(OR = 1\).12, 95% CI: 0.87 to 1.44), contrasting with the original study's positive finding. A key theme from implementation logs was the critical influence of intermittent electrical power supply on diagnostic system engagement, a contextual factor not fully accounted for in the original design.", "conclusion": "The replication did not corroborate the original study's positive effect size, indicating that the proposed methodology may be highly sensitive to unmeasured contextual and operational variables prevalent in industrial settings.", "recommendations": "Future applications of this design must incorporate more robust power infrastructure metrics and longer lead-in periods to establish baseline usage. Adoption studies for industrial technologies should prioritise hybrid methods that integrate sensor data with structured operational audits.", "key words": "replication study, quasi-experimental design, technology adoption, predictive maintenance