Looks Real, or Really Fake? Warnings, Visual Attention and Detection of False News Articles
1. Introduction
1.1. In recent years, online misinformation designed to resemble news by adopting news design conventions has proven to be a powerful vehicle for deception and persuasion.
1.2. Misinformation spread through digital media is a complex problem that became more salient in consumers’ minds as a result of coverage of the role it may have played in the 2016 United States presidential election.
1.3. Less studied, but crucial to the understanding of misinformation psychology, have been the process through which consumers view, process and evaluate the veracity of online articles from unfamiliar sources that use news design elements to make misinformation seem factual or credible.
1.4. These false news articles rely on social media for their spread, to an even greater degree than other online news stories do.
1.5. In a 2 x 2 eye-tracking experiment, news consumers viewed four science news articles from unfamiliar sources, then rated each article for credibility before being asked to classify each as true news or as false information presented as news.
1.6. The present research utilized a 49-participant mixed factorial eye-tracking experiment in the aim of contributing to the nascent literature on false news and information processing in three key areas.
1.6.1. First, it sought to provide one of the first examinations of how online news readers visually attend to areas of the article page on fake news stories in comparison to real news stories.
1.6.2. Secondly, by examining the effects of forewarning participants about the existence of misinformation on science topics, the study sought to examine whether the salience of fake news played a role in which elements of news stories participants viewed.
1.6.3. Thirdly, the study sought to gauge the relationship between users’ visual attention to several design elements on the article pages, such as source information, story recency and authorship information, internal story links, and external page links, on consumers’ credibility evaluations and ability to detect “fake” news stories.
2. Results
2.1. 1
2.2. 2
2.3. 3
3. Method
3.1. Forty-nine university students (undergraduate and graduate) at a large public U.S. university were recruited from a student research participation pool.
3.2. In a two (warning: present-absent) x two (article veracity: real/fake) mixed-factorial design, participants were asked to view and evaluate four science articles online.
3.3. Participants were randomly assigned to either receive or not receive a textual forewarning reminding them that fake science news is prevalent on the internet before browsing the 4 articles.
3.4. Each participant was asked to read the four articles as they would on their own and allowed to progress to the next article at their own pace, while the location and duration of their gaze was recorded unobtrusively by eye-tracking hardware.
3.5. Each participant viewed the same four articles, which were presented in a randomly generated order in each session.
3.5.1. Two articles were chosen to represent false content from disreputable publishers, and two articles were chosen to represent real content from real publishers that would likely not be familiar to participants.
3.5.1.1. The first false science news article was entitled “Amateur Divers Find Long-Lost Nuclear Warhead”.
3.5.1.2. The other false science news article was entitled, “Devastating Snowfall Predicted for NW Montana in Early 2018”.
3.5.1.3. The first true science news article was titled “The FDA Says MDMA is a ‘breakthrough’ drug for PTSD patients”.
3.5.1.4. The second true science news article was titled “Skin pigment could power new implantable battery”.
4. Literature Review
4.1. 1
4.1.1. Misinformation styled to resemble professionally produced news, also known as false news or “fake” news, reached the public consciousness in part due to concerns about its potential role in influencing the 2016 Brexit vote in the U.K. and the outcome of the 2016 U.S. presidential election.
4.1.2. The spread of health and science misinformation can have negative consequences for both those who believe it and the general public.
4.1.3. Fringe communities such as “anti-vaxxers” rely upon social media platforms to spread misinformation amongst each other and to convert the uninformed into believers.
4.1.4. These communities usually single out a topic that is ideologically driven, such as climate change denial, the autism-vaccine link or the Ebola outbreak of 2014, and cultivate a narrative that fits their belief system in spite of substantial evidence to the contrary.
4.2. 2
4.2.1. Producers of fake news tend to share two common characteristics that differentiate them from other, even heavily slanted, news producers
4.2.1.1. First, they have no investment or interest in accurate reporting or the inherent truth of their content
4.2.1.2. Second they are not interested in their long run reputation or credibility, but instead focus on short-term exposure and number of clicks in an initial period of exposure
4.2.2. It is common for fake news pieces to take true content and present it in a completely false context.
4.2.3. Fake news websites often are intentionally constructed and named so that they resemble websites of more credible news sources.
4.3. 3
4.3.1. Researchers have sought to identify effective interventions to decrease the likelihood of consumers falling for misinformation. Two broad categories of such research can be defined by the timing of the intervention: either “inoculation” approaches that include a warning delivered prior to a user’s exposure to an article, or exposure to “correction” after reading the article
4.3.1.1. Studies of correction approaches have shown a persistent effect of misinformation consumption; despite exposure to correction, some portion of message consumers will continue to believe and rely upon misinformation in decision-making.
4.3.1.1.1. Encoding information from a fake news story and then being informed that the information is not true requires users to wholly discard that mental model.
4.3.1.1.2. If users cannot replace the information with a complete model, they show persistent reliance on misinformation to complete it.
4.3.1.2. Inoculation approaches have instead gauged the effects of reminding online news readers of the existence or threat of misinformation, either prior to general browsing, or prior to clicking a specific disputed article.
4.3.1.2.1. Because a warning should heighten users’ defensive processing of content they are exposed to, it should increase the likelihood of users applying greater scrutiny to the content and/or design of fake news pages, which should lead to greater recognition of fake news articles.
4.3.2. Despite the recent rise in research on way to combat the spread and influence of misinformation, little work has been done to identify factors in the format and appearance of fake news articles on which consumers base their initial determination that an article is “real” or “fake".
4.3.2.1. Schema theory, first outlined by Bartlett (1932), focuses on the existence, and persistence, of mental models of information in the mind of individuals, which shape the way they process new information and interact with the world.
4.3.2.2. Individuals often process information using conceptual processing, which involves retrieving a schema, or mental model, from past experience if it is has the possibility of making it easier to interpret a new situation or stimulus.
4.4. Research questions
4.4.1. RQ1: For readers of online articles from unfamiliar sources, what is the relationship between a prior warning about the existence of science misinformation and participants’ ability to correctly identify science misinformation as true or false?
4.4.2. RQ2: For readers of online articles from unfamiliar sources, to what extent do participants pay attention to article identifier elements (byline, datestamp, headline) when viewing news articles from an unfamiliar source?
4.4.3. RQ3: For readers of online articles from unfamiliar sources, what is the relationship between visual attention to identifier elements, visual attention to site internal links, visual attention to site external links, and perceived credibility of the article?