Caution is needed when using crowdsourced data to study environmental disparities

What is it about?

Crowdsourced data platforms (also known as citizen science platforms) have been used to collect an impressive amount of environmental data from users around the world. These data offer researchers new opportunities to study environmental problems. However, data collection efforts for these platforms are often unstructured, making them ill-suited for some types of research questions. In this study, we examine how crowdsourced data might be used to investigate unequal access to environmental and natural amenities, like biodiversity. We show that data availability in a well-known crowdsourced data platform (eBird) is strongly correlated with the demographics (like income and racial composition) of census tracts in two large American cities. Thus, research efforts to understand the relationship between demographics and bird sightings on eBird will suffer from problems of statistical bias.

Featured Image

Photo by Nerea Martí Sesarino on Unsplash

Photo by Nerea Martí Sesarino on Unsplash

Why is it important?

Environmental disparities represent a growing and important field of study. However, quantitative research in this realm requires thoughtful consideration of how the data on environmental outcomes are produced, and in particular, whether those data are generated in ways that are also influenced by the disparities of interest. In our study, we show how the availability of eBird data is strongly correlated with income and racial composition of the tract from which data is recorded (or not). Because the underlying availability of these crowdsourced data is related to demographic disparities, such data will be ill-suited for studying environmental disparities due to statistical bias that arises in such settings.