Data, decisions, and drinking water safety: an interdisciplinary analysis of the complex adaptive response to monitoring in rural Kenya

For decades, international and national development efforts have endeavoured to improve rural drinking-water safety in Sub-Saharan Africa by focusing on infrastructure provision, devolving responsibility for operations and maintenance to the community level. Rural areas continue, however, to have severely limited access to both safe water and to water quality information for operational decision-making. This thesis argues that sustained provision of safe water is an emergent function of a complex adaptive system, and that information-response feedback mechanisms must be positioned to support decision-making at the local level if this function is to be realised. Through empirical research situated in rural Kenya, it explores how a systems-based understanding of the links between data and decision-making can elucidate leverage points for establishing an information feedback that improves water safety and thereby mitigates health risk. Guided by a structurationist systems framing and questions that align with normative, descriptive, and prescriptive modes of decision analysis, the research described in this thesis intersects with both natural and social science disciplines. Pragmatically, it draws on a diverse range of literature and methods. The empirical work is structured in three parts: Data uncertainty has implications for normative conceptions of risk management. Research demonstrating growth of E. coli in the environment challenges interpretations of microbial water quality based on E. coli test results. Characterising strains using whole genome sequencing offered insight into the utility of E. coli as an indicator of microbial contamination in rural water supplies. The difficulty of interpreting health risk from grab samples, especially at the household level, is highlighted. Monitoring can be an effective feedback mechanism at supply level, informing more reliable understandings of hazard dynamics, by accounting for sanitary conditions and temporal variability in indicator concentrations. Data-informed understandings of risk interact with other drivers of decision-making. An integrated fear appeal framing, which responds to key weaknesses in the behaviour change literature, is applied a) to assess user perceptions of drinking-water safety and b) to evaluate an information intervention through which monitoring data were shared with local lay water managers. The findings emphasise that data should be reported with sensitivity to self-efficacy limitations and the threatscapes that decision-makers navigate. Specific, contextualised, and repeated messaging can reinforce engagement with water safety precautions at supply level – especially if lay water manager self-efficacy is supported through infrastructure design, training, and ongoing resourcing. Institutional structure enables and constrains monitoring activity and data flows. Dilemma analysis is used to synthesise views from stakeholders in bureaucratic, market-based, and community domains. The findings challenge the common practice of conceptualising water quality versus quantity as dichotomous objectives. They advance the literature on pluralistic water governance by explicitly considering quality risks, which have received less coverage than other service dimensions. Progress on securing safe water is inhibited by unclear divisions of responsibility and risks associated with not being able to respond to data, particularly at the local level. This thesis concludes that, although the health of water users is the ultimate concern, monitoring activity should focus at supply level – with research and policy developments needed to establish technical and institutional structures that support the efficacy of lay management. A structurationist systems framing was instrumental in guiding the research to evidence a range of stock and flow, feedback, and structural leverage points for implementing effective monitoring. This framing, which links international policy and local practice by recognising the interaction of agency and structure in stakeholder responses to hazards, could be applied broadly to problems of health risk management. It is useful for research that aims to contribute prescriptive decision analysis to bridge the gap between normative models and decision-making in practice.