How Source Attribution Affects Responses to Misinformation on Personal Messaging

• In Event: The Online and Offline Effect of Misinformation

Abstract

Based on a pre-registered experiment, this study shows that attributing news shared on WhatsApp to a reputable source can boost the perceived accuracy of a message, as well as affect the likelihood of other important behavioral outcomes such as staying silent and asking for additional information. Source attribution swayed participants’ responses more strongly than the factual accuracy of the news being shared, and responses to factually inaccurate news attributed to a source were statistically indistinguishable from responses to factually accurate news. These findings foreground the enduring importance of trusted news brands in complex media ecosystems but also highlight the challenges of combating misinformation on personal messaging, where malicious users can easily share falsehoods masqueraded as coming from reputable sources.

We conducted a pre-registered between-subjects experiment embedded in an online survey of 2,500 British WhatsApp users that resembled the adult population based on gender, age, education, and region of residence. Participants were randomly assigned to seeing a vignette that invited them to imagine that another user had posted a news story, embedded in a realistic mock-up of the WhatsApp interface. This paper focuses on two experimental manipulations: whether the news was attributed to the BBC (the most trusted news organization in the UK) or not attributed to a specific source, and whether the news was factually accurate or not.

Our hypotheses focus on six different outcomes, all measured after exposure to the treatments: a cognitive outcome (perceived message accuracy) and five intended behavioral outcomes (likelihood to share or forward the message, likelihood to stay silent, likelihood to ask for additional information, likelihood to provide additional information, and likelihood to verify the accuracy of the message). These behavioral outcomes are crucial on personal messaging, where end-to-end encryption precludes automated interventions and users’ actions are essential to combating misinformation.

We theorized that participants’ responses would be affected by the factual accuracy of the news reported in the message and by the presence of a reputable source. This would mean that participants would consider factually accurate news and news attributed to an authoritative source to be more accurate, and that they would be more likely to share it and less likely to verify it. We also theorized that, in the social context of personal messaging, participants exposed to accurate news and to news attributed to an authoritative source would be more likely to stay silent after seeing it and less likely to ask for and provide additional information on it—all signs of conflict avoidance that our previous research shows are common reactions to information shared in informal conversations on personal messaging. Finally, we hypothesized that the effects of source attribution and factual accuracy of the message would interact with each other.

We analyzed the data based on ANOVAs with Bonferroni corrections for multiple comparisons. As predicted, when participants saw news attributed to the BBC, they perceived the message as significantly more accurate than when the news was not attributed to a source. Participants exposed to news attributed to the BBC also saw themselves as significantly more likely to stay silent and less likely to ask for additional information, as we hypothesized. Source attribution also increased participants’ intention to share or forward the message and decreased their intention to provide additional information and to verify the accuracy of the message, but these relationships were not statistically significant.

The importance of source attribution emerges even more clearly in light of our second key set of findings. As we predicted, participants considered factually true messages as significantly more accurate than false ones. However, contrary to our expectations, participants’ behavioral intentions did not differ based on whether they had seen true or false messages. Thus, participants discerned between accurate and inaccurate information in their cognitive assessments but not in their intended reactions, which suggests that enlisting personal messaging users in anti-misinformation work may be challenging.

Finally, the effects of attribution to a reputable source did not vary between factually accurate and inaccurate news. When participants saw news attributed to the BBC, their reactions were similar irrespective of whether the news was true or false. This null finding suggests personal messaging users may be vulnerable to misinformation packaged as coming from a trusted source. Overall, our study highlights how source attribution can work as a double-edged sword on personal messaging, as authoritative news brands can enhance the credibility of accurate messages but also facilitate deception when they are attached to inaccurate content.

Authors

• Cristian Vaccari, University of Edinburgh

• Andrew Chadwick, Loughborough University

• Natalie-Anne Hall, Loughborough University

• Brendan Lawson, Loughborough University