Evaluating crowdsourcing behaviors in identifying online misinformation

Illustrated by Lynette Gow

SECTOR

Governance / Political Behavior

PROJECT TYPE

Lab experiments

DOI

Location

Kenya

BEHAVIORAL THEME

Perception | Accuracy | Bias
OVERVIEW

We conducted an experimental demonstration of the Healthy Internet Project (HIP) plugin in collaboration with UNDP Accelerator Lab Kenya and the HIP initiative incubated at TED. The plugin, an open-source browser extension, allows users to flag online content, aiming to counter falsehoods and negativity while promoting valuable ideas. The experiment sought to understand user motivations and behaviors in flagging misinformation through a volunteer-driven platform.

In the experiment involving 205 onboarded users, only 128 engaged with the platform, with most falling into the Low Flagging Group, submitting fewer than five flags. User perceptions revealed concerns about subjectivity, potential harm, and personal risk associated with flagging negative content, particularly political material. User accuracy was compromised as “misinformation” was often linked to negative sentiment rather than actual falsehoods.

THEMATIC AREAS

External fact-checkers struggled to identify misinformation due to underutilized or inadequately detailed flag comments. The study concludes that volunteer-driven misinformation identification’s effectiveness is limited without addressing user safety perceptions and accuracy concerns.

To enhance platforms like HIP, recommendations include ensuring user anonymity to address risk concerns, providing a primer on misinformation to improve accuracy, reconsidering the Worthwhile flag, and introducing a required “Misinformation Identification” field for clearer reporting to assist fact-checkers.