Improving reliability and trust with pre-launch experimentation in Citizen Science context using water quality sensors

Details

Supervisors

Raskin, Jean-Pierre

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en mathématiques appliquées, à finalité spécialisée

Abstract

According to the United Nations, by 2030, 4.8 billion people’s health and livelihoods could be at risk if water quality monitoring is not improved. Collecting and analysing water data is an essential prerequisite for informed decision-making, yet many regions most at risk face a severe data gap. Citizen science (CS) projects, where citizens collect data, could help fill this gap. The success of these projects heavily relies on their adaptation to the socio-economic, environmental and cultural context. Moreover, given the social interactions involved in such projects, trust is of central importance. This thesis tries to fill the literature gap about trust and proposes a pre-launch experiment that project carriers could conduct to adapt the project to the context and citizens’ beliefs. More specifically, the goal of this experiment is to understand the influence of predefined factors (such as the training type and analyte knowledge) and different types of sensors on the confidence that a citizen has in their ability to handle the sensor or in the sensor’s ability to produce reliable data, and on the accuracy of the data. Instead of a traditional one-factor-at-a-time (OFAT), the study compares groups differing by single factors, a more resource-efficient approach. We conducted the experiment in Belgian secondary school classes and in workshops with adults, using pH as the water quality parameter. The study revealed that in this specific case, training was the most influential factor. The results also showed that the choice of participants matters: pre-launch experiment should involve citizens resembling the future target group, as findings differ depending on who participates. The contribution of this thesis does not lie in the specific Belgian results but in the methodology, which provides a practical, resource-efficient way to anticipate challenges, enhance trust and improve the design of citizen science water monitoring projects. As this work constitutes an initial version of the experiment, it necessarily leaves room for improvement. Several recommendations are given to refine the design and strengthen its capacity to produce reliable insights across different contexts.