The Truth in Sharing: Dissecting the Dynamics of Misinformation Spread

• In Event: The Diffusion of Misinformation

Abstract

Misinformation continues to be a critical threat to public health and democratic governance worldwide. While most existing research focuses on individual-level determinants of misinformation sharing, message-level attributes that make a particular piece of (mis-)information more viral and shareable are understudied. We also know little about the contextual drivers of misinformation spread, such as perceived social norms, the identity of the transmitter, or the forum in which sharing takes place, and how their relative importance compares to that of content, language, and source. Furthermore, the determinants of information transmission likely differ across online versus offline settings, but these contexts are rarely compared explicitly. Most research on misinformation, indeed, focuses on online settings and does so primarily in Western, mostly urban contexts and with adult populations.

Recognizing these gaps, this paper seeks to answer the following two related research questions: (1) Which factors determine whether a piece of information is more likely to be shared, and how do these factors compare in terms of their importance in affecting sharing behavior? (2) How do the effects of these factors differ for online versus offline sharing intentions? We address these research questions through a two-arm conjoint experiment, fielded in-person with about 14,000 adolescents in non-urban settings in the Indian region of Bihar. Focusing on adolescents in a hard-to-reach, mostly offline with low digital literacy provides an important yet understudied perspective, complementing prior work that has mostly been conducted with Western adult samples in online settings. Furthermore, while the target population is still relatively disconnected from internet access (previous data shows that only 1 in 3 households in the study area have internet access), misinformation is still a problem, highlighting the importance of offline and word-of-mouth sharing, as well as reliance on informal sources, village networks, and ethnic ties.

In our conjoint experiment, respondents are presented with pairs of informational statements and asked to choose which of the two statements they are more likely to share. The informational statements are presented as paragraphs and randomly vary in terms of topic (science, medicine, health, politics), veracity (true, false), original source (quality daily newspaper, government agency, YouTube video, no source cue), transmitter identity (Hindu stranger, Muslim stranger, local doctor, community leader, own relative), and perceived level of peer endorsement (widely accepted, controversial, mostly rejected).

In addition, we randomly assign participants to one of two conditions before they complete the conjoint task. In the offline sharing condition, participants are first prompted to recall an in-person situation in which friends or family shared information and what they believe matters most when deciding whether to share a particular piece of information in such a setting. They then complete the conjoint choice task, in which they are asked to state which piece of information they would rather share in an offline setting, such as a casual meeting with friends or family. In contrast, in the online sharing condition, participants are prompted to think about sharing in a WhatsApp group chat before being asked to choose which information they would rather share in such a group chat.

This study makes several important contributions to the literature on (mis-)information sharing. First, we study information-level, contextual, and social drivers of sharing intention, and we do so with an in-person sample of a highly important but hard-to-reach population of rural, mostly low-literacy Indian adolescents. Furthermore, most existing work relies on self-reported measures, whereas our forced-choice conjoint design better recreates naturalistic sharing decisions that involve tradeoffs. Lastly, instead of exclusively focusing on online misinformation spread, we also study offline interpersonal sharing which remains the predominant transmission mode for rumor and misinformation in developing countries despite increasing technology penetration. By studying whether drivers of information sharing differ between online and offline contexts, we add significant nuance to the study of misinformation spread.

Data collection for this study is currently ongoing, and we expect to have a full set of findings by April 2024.

Authors

• Florian Sichart, Princeton University

• Sumitra Badrinathan, American University

• Simon Chauchard, University Carlos 3 of Madrid

• Priyadarshi Amar, University of Wisconsin Madison