Summary:
- In two survey experiments, we showed American participants (N=2,217) a series of actual headlines from social media, presented in Facebook format.
- For some headlines, we made publisher information more visible (by adding a logo banner), while for others we made publisher information less visible (by removing all publisher information).
- We found that publisher information had no significant effects on whether participants perceived the headline as accurate, or expressed an intent to share it – regardless of whether the headline was true or false. In other words, seeing that a headline came from a misinformation website did not make it less believable, and seeing that a headline came from a mainstream website did not make it more believable.
- To investigate this lack of an effect, we conducted three follow-up surveys (total N=2,770). This time, we asked participants to rate either (i) the trustworthiness of a range of publishers or (ii) the accuracy of headlines from those publishers, given no source information (i.e., the headline’s “plausibility”).
- We found a strong, positive correlation between trust in a given outlet and the plausibility of headlines it published. Thus, in many cases, learning the source of a headline does not appear to add information about its accuracy beyond what was immediately apparent from the headline alone.
- Finally, in a survey experiment with 2,007 Americans, we found that providing publisher information only influenced headline accuracy ratings when headline plausibility and publisher trust were “mismatched” – for example, when a headline was plausible but came from a distrusted publisher (e.g., fake-news or hyperpartisan websites).
- In these cases of mismatch, identifying the publisher reduced accuracy ratings of plausible headlines from distrusted publishers, and increased accuracy ratings of implausible headlines from trusted publishers.
However, when we fact-checked the 30% of headlines from distrusted sources in our set that were that were rated as plausible by participants, we found they were mostly true. In other words, providing publisher information would have increased the chance that these true headlines would be mistakenly seen as false – raising the possibility of unintended negative consequences from emphasizing sources.
Our results suggest that approaches to countering misinformation based on emphasizing source credibility may not be very effective and could even be counterproductive in some circumstances.