Methodological Evaluation and Multilevel Regression Analysis of Power-Distribution System Yield in Tanzania, 2000–2026

"background": "The reliability and yield of power-distribution systems in sub-Saharan Africa are critical for economic development, yet systematic, methodologically robust evaluations of equipment performance are scarce. Existing assessments often lack the statistical rigour to account for hierarchical data structures inherent in regional infrastructure networks. ", "purpose and objectives": "This study aims to methodologically evaluate power-distribution equipment systems and quantify the determinants of system yield improvement. The primary objective is to develop and apply a multilevel regression model to isolate the effects of equipment upgrades, maintenance regimes, and contextual factors on technical yield. ", "methodology": "A longitudinal dataset of technical and operational parameters from multiple distribution networks was constructed. A three-level hierarchical linear model was specified: Y{ijk = \0 + \1X1ijk + uj + vk +, where uj and v₊ are random intercepts for district and zone levels, respectively. Estimation used restricted maximum likelihood with robust standard errors. ", "findings": "The multilevel model explained 74% of the variance in system yield. A one-unit increase in the standardised condition index of primary substation transformers was associated with a 0. 23 increase in yield (95% CI: 0. 18 to 0. 28), holding other factors constant. The variance partition coefficient indicated that 31% of the unexplained variance resided at the zonal level. ", "conclusion": "The methodological approach confirms that technical yield is significantly influenced by equipment condition within a nested geographical structure. The analysis provides a statistically sound framework for prioritising infrastructure investments. ", "recommendations": "Investment planning should prioritise transformer health monitoring and replacement programmes. Utilities should adopt multilevel modelling for performance benchmarking to account for regional heterogeneity in future system evaluations. ", "key words": "power distribution, multilevel modelling, infrastructure yield, hierarchical linear model, asset management", "contribution statement": "This paper provides a novel