Psychological interventions countering misinformation in social media: A scoping review

Paweł Gwiaździński, Aleksander B. Gundersen, Michał Piksa, Izabela Krysińska, Jonas R. Kunst, Karolina Noworyta, Agata Olejniuk, Mikołaj Morzy, Rafał Ryguła, Tomi Wójtowicz, Jan PiaseckiFrontiers in Psychiatry 2023 · DOI

TL;DR

This scoping review synthesizes 75 empirical studies on psychological interventions designed to combat social media misinformation. It categorizes interventions into "Boosting," "Nudging," and "Technocognition," ultimately finding that while many strategies work in the lab, automated solutions like Source Ratings and Messages from Trusted Organizations offer the highest viability for real-world implementation.

What Problem Does This Paper Address?

The "infodemic" of misinformation on social media is a documented threat to public health and democratic stability. While psychologists have developed various theories to counter "fake news," there is a significant disconnect between theoretical effectiveness and practical scalability.

The authors address three primary gaps:

  1. Mapping the Landscape: Identifying which psychological interventions actually exist in the literature.
  2. The "Real-World" Gap: Determining how many of these interventions have been tested in actual social media environments versus controlled labs.
  3. Viability Assessment: Creating a framework to evaluate which interventions are technically and economically feasible for platforms like Facebook or X (formerly Twitter) to actually deploy.

Methods

The researchers conducted a systematic scoping review following the PRISMA-ScR guidelines. They searched PubMed, Embase, and Scopus for peer-reviewed empirical studies published between 2004 (the launch of Facebook) and 2022.

Key methodological features include:

  • Taxonomy: They adopted a framework by Kozyreva et al. (2020) to classify 15 types of interventions into three broad categories:
  • Boosting: Fostering cognitive competencies (e.g., media literacy, "inoculation" or pre-bunking).
  • Nudging: Subtle changes in choice architecture (e.g., warnings, social correction).
  • Technocognition: Designing UI/UX to account for human cognitive biases (e.g., adding "friction" before sharing, source ratings).
  • Intervention Assessment Score (IAS): A novel 5-point Likert scale used by an interdisciplinary team to rate each intervention on technical ease, resource requirements, and the level of user motivation needed for success.

Datasets

As a scoping review, the "dataset" consists of the literature itself:

  • Initial Pool: 4,279 records.
  • Final Sample: 75 peer-reviewed empirical papers.
  • Criteria: Studies had to present primary data (qualitative or quantitative) on psychological interventions targeting misinformation in a social media context.

Key Insights and Ideas

  • The Ecological Deficit: Most research is "non-ecological." Out of 75 studies, only 7 were conducted in real-world social media environments. 30 were purely lab-based, and 29 used "mimical" designs (fake interfaces that look like Facebook).
  • The 2016 Inflection Point: Research into this field was nearly non-existent before 2013 but exploded after 2016, coinciding with major global political events and the rise of the "post-truth" narrative.
  • Viability vs. Effort: Interventions like Inoculation (teaching users to spot manipulation techniques) are theoretically powerful but scored low on viability (2.93/5) because they require highly motivated users and significant time.
  • Automation is the Path Forward: Interventions that can be automated via algorithms or bots (Nudging and Technocognition) are deemed far more viable for platforms than those requiring human-led education (Boosting).

Key Results

  • Effectiveness: Approximately 65% (49/75) of the studies reported successful outcomes for their interventions. Only two studies reported "counterproductive" results (where the intervention actually increased susceptibility).
  • Top Performers: The interventions with the highest Intervention Assessment Scores (IAS)—meaning they are the most "ready" for the real world—were:
  • Message from a trusted organization (4.20/5.0)
    • Source rating (4.20/5.0)
    • UX Manipulation/Friction (3.88/5.0)
  • Platform Bias: Research is heavily skewed toward text-based platforms. Facebook (39 papers) and Twitter (25 papers) dominate the data, while video-heavy platforms like TikTok and YouTube are severely under-researched in this context.

Limitations and Future Work

  • Demographic Blind Spots: The review notes a lack of research on participants over age 65, a group often cited as being more vulnerable to sharing misinformation.
  • The "Visual" Gap: Most interventions are text-based (e.g., a written warning). There is little data on how to counter misinformation in images, memes, or short-form videos.
  • Cultural Concentration: The vast majority of studies come from the US, UK, and Germany. The authors call for more research in different cultural and political climates where misinformation dynamics may differ.
  • Hybrid Approaches: The authors suggest that the most effective future systems will likely combine Technocognition (UI friction) with Nudging (social pressure indicators) to create a seamless, less "preachy" user experience.

Summary updated

Abstract

[https://figshare.com/], identifier [https://doi.org/10.6084/m9.figshare.14649432.v2].
BibTeX
@inproceedings{gwiazdzinski2023psychological,
  author = {Paweł Gwiaździński and Aleksander B. Gundersen and Michał Piksa and Izabela Krysińska and Jonas R. Kunst and Karolina Noworyta and Agata Olejniuk and Mikołaj Morzy and Rafał Ryguła and Tomi Wójtowicz and Jan Piasecki},
  title = {{Psychological interventions countering misinformation in social media: A scoping review}},
  booktitle = {Frontiers in Psychiatry},
  year = {2023},
  doi = {10.3389/fpsyt.2022.974782},
}
Figure