Abstract

"background": "Understanding the drivers of adoption for engineered water treatment systems in sub-Saharan Africa is critical for public health and infrastructure planning. Existing studies often fail to account for the hierarchical structure of data, where household decisions are nested within communities influenced by regional policies and environmental factors. ", "purpose and objectives": "This case study presents a methodological framework for applying multilevel regression analysis to evaluate the adoption of point-of-use and communal water treatment technologies. It aims to demonstrate the model's utility in isolating community-level effects from household-level predictors. ", "methodology": "A longitudinal dataset from a national monitoring programme was analysed using a three-level mixed-effects logistic regression model. The model, specified as \ (p{ijk) = \0 + \ Xijk + ujk + vk, where pijk is the probability of adoption for household i in community j in district k, with random intercepts ujk and vk, accounted for clustering. Inference was based on 95% confidence intervals using robust standard errors. ", "findings": "The analysis quantified significant variation attributable to district-level governance structures, accounting for approximately 15% of the residual variance. A key concrete result is that household proximity to a maintained borehole showed a negative association with adoption of point-of-use systems (OR = 0. 65, 95% CI 0. 52, 0. 81). ", "conclusion": "Multilevel modelling is a robust methodological approach for engineering adoption studies, providing nuanced insights by partitioning variance across contextual tiers. It confirms that adoption is not merely a function of household attributes but is substantially shaped by higher-level infrastructural and administrative contexts. ", "recommendations": "Engineering policy assessments should routinely employ hierarchical models. Programme evaluations must integrate community-level variables, particularly existing infrastructure access, into their design. Future data collection should be structured to support nested analysis. ", "key words": "multilevel modelling, water treatment adoption, infrastructure