Who’s in the crowd? Addressing bias in crowdsourced data

Abstract

National forests and grasslands are popular recreation destinations. Providing opportunities for visitors while protecting these natural attractions falls under the scope of recreation planning, but for recreation planners to do their jobs effectively, they need data—information that includes visitor demographics, reasons for visiting, and where visitors are coming from.

Increasingly, recreation planners are using real-time visitor data from posts and photos of trip activities posted on crowdsourced platforms, such as social media or citizen or community science projects, to inform planning efforts. Yet these crowdsourced data are not representative of all visitors. If applied to models without correcting for how, and by whom, these platforms are used, estimates will skew toward more frequent visitors that provide the data, instead of the general population of visitors that planners aim to survey.

Sonja Kolstoe, a research economist with the U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, has spent nearly 10 years exploring and refining techniques to use crowdsourced data and developing data correction methods to address selection bias. Her work was the basis for The Nature Conservancy’s study of visitors’ willingness to pay to visit the Waikamoi Preserve on the island of Maui, Hawai‘i. The Washington Department of Fish and Wildlife is also developing a pilot study to analyze the economic value of salmon viewing and the project’s survey methods will be based on Kolstoe’s research.

Citation

Watts, Andrea; Kolstoe, Sonja. 2024. Who’s in the crowd? Addressing bias in crowdsourced data. Science Findings 270. Portland, OR: U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station. 6 p.