Mitigating Methodological Challenges in Citizen Science using Data Science

Abstract Citizen science initiatives offer an unprecedented scale of volunteer-driven data collection but often face scrutiny regarding their methodology, research design, and data collection as well as analysis. Addressing these concerns, this paper adopts a data science approach to process and enhance the integrity of data generated from citizen science projects. We present a methodological framework that employs data science techniques to effectively mitigate data noisiness and coverage biases, issues commonly associated with citizen science datasets. The paper features a case study involving a collaboration with JGM, a citizen science research group specializing in serious gaming and training. This partnership provides a unique lens to examine the application of data science techniques in citizen science, focusing on analysing team dynamics in escape room scenarios. This article outlines rigorous data preprocessing and processing workflows implemented from a data science standpoint to ensure data quality. The processed dataset, comprising 291 observations and 55 variables, is a blueprint for enhancing data reliability in citizen science endeavours. In summary, this paper demonstrates how data science methods can make citizen science projects more reliable and replicable. We encourage more work that combines these two fields to improve the quality of research.