Comparative Field Diagnostics of Rural Water Supply Reliability in Rwanda: A Randomised Control Trial

Reliable rural water supply is critical for agricultural productivity and community health, yet standard monitoring often fails to capture frequent, localised service failures. Existing diagnostic methods lack rigorous field validation, particularly in sub-Saharan African contexts. This study aimed to methodologically evaluate and compare the diagnostic accuracy of two field-based protocols for measuring the functional reliability of municipal piped water systems serving agricultural communities. A randomised control trial was conducted across 120 rural villages. Villages were randomly assigned to one of two diagnostic protocols: a 'snapshot' compliance check or a 'longitudinal' sensor-based monitoring protocol. System reliability was modelled using a multilevel logistic regression: (P(Yᵢⱼ=1)) = β₀ + β₁ X₁ᵢⱼ + β₂ Zⱼ + uⱼ, where Yᵢⱼ is functional status, with robust standard errors clustered at the district level. The longitudinal protocol detected a significantly higher prevalence of intermittent supply than the snapshot method, identifying system failure in 47% of sampled communities compared to 18% (95% CI for difference: 22% to 36%). The primary failure mode was pressure loss, not complete outage. Longitudinal, sensor-based field diagnostics provide a substantively different and more accurate assessment of rural water supply reliability than standard compliance checks, revealing hidden intermittency. National monitoring programmes should integrate continuous sensor-based diagnostics into routine assessments. Donor projects should fund longitudinal data collection to better target infrastructure investments. water security, monitoring and evaluation, randomised trial, infrastructure, sub-Saharan Africa This paper provides the first experimental evidence comparing field methods for water point diagnostics and introduces a validated, scalable protocol for detecting intermittent supply.