Contributions
- Comprehensive taxonomy of nine evidence-based intervention types against online misinformation
- Synthesis of 81 scientific papers across behavioral, cognitive, political, and educational research
- Conceptual framework organizing interventions into three categories: nudges, boosts, and refutation strategies
- Identification of four main experimental paradigms: misinformation-correction, headline-discernment, technique-adoption, and field studies on social media
- Interactive online toolbox (https://interventionstoolbox.mpib-berlin.mpg.de) with searchable databases and global evidence map
- Analysis of geographic representation, longevity of effects, and heterogeneity in measurement
Method
The authors conducted a literature review based on five inclusion criteria: definition and scope, problem addressed, non-redundancy, empirical evidence, and expert opinion from 30 misinformation researchers across 11 countries. They synthesized evidence from 81 peer-reviewed papers published or about-to-be-published, covering interventions that explicitly tackle misinformation (including disinformation, false and misleading information, and fake news). The conceptual overview organizes interventions along ten dimensions (target, mode of implementation, entry point, outcome type, etc.), while the evidence overview summarizes each paper's sample, experimental paradigm, study design, outcome measures, and effect sizes. The authors mapped studies geographically and identified four primary experimental paradigms used across interventions.
Results
Intervention taxonomy (nine types):
Nudges (targeting behavior):
- Accuracy prompts: Direct attention to accuracy (e.g., asking users to evaluate headline accuracy); outcome: sharing discernment
- Friction: Add effortful steps before sharing (e.g., pause-and-think prompts); outcome: sharing intentions
- Social norms: Leverage peer influence to discourage misinformation sharing; outcome: beliefs and sharing behavior
Boosts and educational interventions (targeting competence):
- Inoculation: Pre-emptive exposure to weakened misinformation or manipulation tactics; outcome: manipulation technique recognition
- Lateral reading and verification strategies: Teach credibility assessment via lateral reading (searching for information about a source elsewhere online), image searching, or context tracing; outcome: credibility assessment of websites
- Media-literacy tips: Provide checklist strategies for spotting false news; outcome: accuracy and sharing discernment
Refutation strategies (targeting beliefs):
- Debunking and rebuttals: Provide corrective information with explanation of why misinformation is false; outcome: misconception reduction
- Warning and fact-checking labels: Inform users of trustworthiness via labels (e.g., "False—Independent fact-checkers say this has no factual basis"); outcome: accuracy judgments
- Source-credibility labels: Show how news sources are rated by fact-checking organizations; outcome: sharing intentions and information diet quality
Evidence summary:
- Evidence is heavily concentrated in the Global North, especially the USA, though some interventions (debunking, accuracy prompts, media-literacy tips) have been tested across multiple countries
- Interventions are sensitive to cultural differences and often less effective for less-educated rural populations
- Few studies test long-term effects; many show decay in effectiveness over time
- Heterogeneity in experimental paradigms, test stimuli, and outcome measures makes direct comparison difficult
- Field studies on social media represent an emerging and growing paradigm
Four experimental paradigms:
- Misinformation-correction: Measure belief reduction after presenting corrections
- Headline-discernment: Participants rate true/false headlines; outcome: sharing discernment (difference in sharing intentions for true vs. false)
- Technique-adoption: Assess whether participants learn and apply skills (e.g., lateral reading, manipulation detection)
- Field studies on social media: Deploy interventions in real social networks; measure information-diet quality and sharing behavior
Notes
Strengths:
- Authoritative international collaboration; represents consensus across 30 experts from 11 countries
- Accessible, interactive online toolbox makes the evidence searchable and updatable
- Clear taxonomy facilitates future meta-analyses and systematic reviews
- Honest about limitations: heterogeneity in methods, geographic bias, knowledge gaps on long-term effects
- Pragmatic framing—interventions as tools to be combined rather than ranked
Weaknesses:
- Does not evaluate interventions' comparative effectiveness or implementation feasibility
- Limited evidence from Global South; interventions may not generalize equally across cultures
- Mechanism of long-term effects (or decay) poorly understood; few longitudinal studies
- Heterogeneity in paradigms and outcome measures slows cumulative progress in the field
- Platform-internal research (e.g., friction treatments, community notes) not included due to lack of published evidence
Open questions:
- How can medium- and long-term effects be better tested and understood?
- Which interventions work best for which populations and cultural contexts?
- How do interventions scale across school curricula, apps, and platform integrations?
- Can integrative interventions be designed to handle multiple content types (headlines, websites, video)?
- What system-level platform design changes would complement individual-level interventions?
Implication for policy and practice: The toolbox is explicitly designed for use by researchers (meta-analyses, method standardization), policymakers (policy discussions, regulation), educators (curriculum design), and individuals (self-nudging). The authors emphasize that individual-level interventions alone are unlikely to fully solve misinformation at the platform level—system-level approaches (platform design, content moderation, regulatory frameworks) are necessary for long-term effect.