Abstract
{ "background": "System reliability is a critical determinant of productivity and competitiveness in manufacturing, yet robust methodological frameworks for its evaluation in resource-constrained industrial settings are scarce. Existing approaches often lack the rigour to isolate causal effects from operational noise.", "purpose and objectives": "This study aimed to develop and apply a novel quasi-experimental methodological framework to evaluate the reliability of production systems within Tanzanian manufacturing plants, focusing on the causal impact of specific maintenance interventions.", "methodology": "A difference-in-differences quasi-experimental design was implemented across a matched sample of treatment and control production lines in multiple plants. System reliability was measured via mean time between failures (MTBF). The core statistical model was $\\Delta Y{it} = \\beta0 + \\beta1 \\text{Treatment}i + \\beta2 \\text{Post}t + \\beta3 (\\text{Treatment}i \\times \\text{Post}t) + \\epsilon{it}$, with inference based on cluster-robust standard errors.", "findings": "The intervention produced a statistically significant positive effect. The estimated coefficient $\\beta_3$ was 14.7 (95% CI: 8.2, 21.2), indicating that the treatment increased MTBF by approximately 15 hours relative to the control group. This represents a 22% improvement in the reliability metric for the treated systems.", "conclusion": "The proposed framework successfully isolates the effect of reliability-centred maintenance interventions, demonstrating its utility for causal analysis in real-world industrial environments. It provides a more rigorous alternative to observational studies.", "recommendations": "Manufacturing plant managers should adopt quasi-experimental designs for evaluating operational improvements. Policymakers should support the development of local expertise in industrial experimental methodology to enhance evidence-based decision-making.", "key words": "system reliability, quasi-experimental design, manufacturing, maintenance engineering, difference-in-differences, industrial research methods", "contribution statement": "This paper