Community‐Driven Fact‐Checking on <scp>WhatsApp</scp> : Who Fact‐Checks Whom, Why, and With What Effect?

ABSTRACT This paper studies community‐driven fact‐checking—the members of a community fact‐checking their own content—on WhatsApp, with the aim of determining its prevalence, who does it, and whether it is effective. The study leverages two large datasets of WhatsApp group chats, encompassing both public and private group conversations with varying levels of intimacy among members. Adopting a mixed‐methods approach, the research combines quantitative analysis of observational data with qualitative measures to shed light on these research questions. The findings reveal that community‐driven corrections are infrequent, and when they do occur, they are typically conveyed through polite requests aimed at alerting individuals about the presence of misinformation. However, users often exhibit apathy towards self‐correction, disregard the corrections, or even feel offended by public corrections within the group. Notably, the responsibility of correcting misinformation primarily falls on active community members, with group administrators accounting for a relatively small portion (up to 20\%) of the corrections. Additionally, the study uncovers significant variations in the types of corrections and responses to corrections, influenced by group norms and the degree of familiarity among group members. These observations suggest the existence of underlying dynamics of power and trust within these groups. The insights from this research hold implications for fact‐checking and policies in encrypted chat platforms, as well as the role of the community in facilitating accurate information dissemination even in non‐encrypted discussions.