By Jenna Somers
Because half of Americans “often” or “sometimes” get their news from social media, according to the Pew Research Center, it’s important to do two things: reduce the sharing of false information on social media, and increase users’ ability to distinguish truth from fiction.
Vanderbilt professor and researcher Lisa Fazio will lead an international collaboration of 80 misinformation scientists in a project called “Building a better toolkit (for fighting misinformation)” to compare the effectiveness of eight common strategies to help combat false information on social media. A two-year, $477,916 grant from the Social Science Research Council will support the large collaborative project.
“I’m really excited about this project and the large group of collaborators who will be helping to design, conduct and analyze the research,” Fazio said. “Our goal is to provide practical tips for practitioners fighting misinformation while also learning more about the psychological processes involved in identifying false information.”
Fazio, who is an associate professor of psychology and human development, will work with her team to finalize the design of the eight interventions; test their effectiveness in an online experiment with 30,000 participants using true, false and misleading political and health-related headlines; examine the duration and diversity of the effects across participants and headlines; and conclude by assessing the effectiveness of the most promising interventions in a real-world context on YouTube.
The team will test these interventions:
- Inoculation: warning people about manipulative techniques often used in false posts
- Debunking: providing corrective information after viewing false information
- Warnings: warnings as currently implemented by Facebook (e.g., “False Information: Checked by independent fact-checkers”)
- Accuracy prompt: reminding people to think about accuracy
- Digital media literacy tips: short tips on how to identify true information
- Source credibility: providing information about the credibility of the source
- Thinking mode: encouraging people to think about what they already know on the topic
- Descriptive norms: reminding people that most users value accurate information and don’t like people who share false information
Accuracy rating and sharing intention are the two key outcomes of interest. An example accuracy rating question is, “To the best of your knowledge, is the above headline accurate?” and an example sharing intention question is, “If you were to see the above headline online, how likely would you be to share it?”
The researchers do not intend to prove that one intervention is more or less effective than another. Instead, they want to document the effectiveness of the interventions under controlled conditions so that researchers and practitioners can be better aware of the pros and cons of using them.
Critical research questions will guide the team’s assessment of these interventions:
- How effective are the interventions at helping people discern which posts are true or false? How effective are the interventions at reducing false information sharing? Are misinformation experts and/or the public accurate at predicting which interventions will be effective?
- How long do the effects of the intervention last? How do the benefits of the intervention vary based on the type of information (e.g., true/false/misleading; health-related vs. politics, partisanship, plausibility) and characteristics of the participants (e.g., political affiliation, education, media literacy)?
- Are the most promising interventions effective when implemented as advertisements on YouTube?
Fazio’s team will preregister information about the study, including analyses, on the Open Science Network and debrief all participants on which social media posts were false or misleading so that no one leaves the study misinformed. They also will design a “cookbook” resource to guide practitioners on which interventions to use and how to mix and match them to achieve desired goals. While no single intervention is superior in all scenarios, Fazio and her colleagues intend to provide useful information on the relative benefits of various interventions.
