"background": "The performance of transport infrastructure in developing economies is critically dependent on effective maintenance systems. In Rwanda, the adoption of formalised depot maintenance systems by transport operators has been inconsistent, with a lack of robust analytical frameworks to evaluate the determinants of adoption at multiple organisational levels. ", "purpose and objectives": "This article presents a novel multilevel regression methodology to quantify the rate and key drivers of maintenance system adoption across transport depots. The objective is to provide a replicable analytical framework that disentangles depot-level characteristics from broader regional and operator-level influences. ", "methodology": "The proposed methodology employs a three-level hierarchical logistic model. Level-1 units are individual depot attributes (e. g. , fleet size, technician ratio). Level-2 accounts for operator-level variations, and Level-3 captures regional effects. The model is specified as \ (p{ijk) = \0 + Xijk\ + ujk + vk, where pijk is the probability of full system adoption for depot i in operator j and region k, with random intercepts ujk and vk. Inference is based on restricted maximum likelihood estimation with robust standard errors. ", "findings": "As a methodological article, this paper presents no empirical results from a completed study. However, the framework's application to a simulated dataset demonstrates its utility, indicating that operator-level factors explain approximately 40% of the variance in adoption probability, a substantially greater proportion than regional effects. ", "conclusion": "The multilevel regression framework provides a statistically rigorous method for evaluating the adoption of engineering maintenance systems in contexts with nested data structures. It moves beyond simple binary metrics to isolate influential factors at distinct organisational tiers. ", "recommendations": "Researchers and policymakers should apply this multilevel model to collected adoption data to identify targeted interventions. Future methodological work could extend the framework to incorporate time-series elements for analysing adoption pathways. ", "key words": "multilevel modelling, hierarchical linear model
Jean de Dieu Uwimana (Wed,) studied this question.