Chen Chen

Purpose: With the advances in AI-driven audiovisual synthesis, multimodal disinformation has become prevalent in social media, where user engagement may complicate its evaluation. It remains unclear why individuals are susceptible to such content, especially in contexts featuring social engagement cues. Therefore, this study aims to investigate the roles of multimodality, online comments and virality metrics in disinformation evaluation. Design/methodology/approach: Guided by the heuristic–systematic model, multimodal framing theory, and social proof theory, we carried out a mixed-methods study with two phases. In Phase I, a between-subjects experiment (N = 433) examined how various cues independently and interactively influenced credibility judgments of disinformation. In Phase II, follow-up interviews (N = 24) uncovered the causes driving the observed effects. Findings: Text-with-video disinformation was perceived more credibly than text-with-image and text-only types. Participants generally considered videos more difficult to forge. This view served as a heuristic that triggered less critical evaluation. Additionally, online comments could correct or mislead credibility judgments, whereas virality metrics showed no such effect. Interestingly, the impact of online comments could be attenuated by multimodality, making participants more susceptible to disinformation. Because audiovisual cues offer sufficient intuitive evidence, participants relied less on social proof for judgment-making. Originality/value: Synthesizing the abovementioned theories, this study advances the understanding of individual susceptibility to multimodal disinformation. This susceptibility stems not only from multimodality per se, but also from its tendency to overshadow the efficacy of social proof in correcting falsehoods. From the information processing perspective, this phenomenon is driven by intuitive sufficiency of audiovisual realism, which satisfies judgmental confidence and renders social proof cues less significant.

As deepfake videos become increasingly difficult for people to recognise, understanding the strategies humans use is key to designing effective media literacy interventions. We conducted a study with 195 participants between the ages of 21 and 40, who judged real and deepfake videos, rated their confidence, and reported the cues they relied on across visual, audio, and knowledge strategies. Participants were more accurate with real videos than with deepfakes and showed lower expected calibration error for real content. Through association rule mining, we identified cue combinations that shaped performance. Visual appearance, vocal, and intuition often co-occurred for successful identifications, which highlights the importance of multimodal approaches in human detection. Our findings show which cues help or hinder detection and suggest directions for designing media literacy tools that guide effective cue use. Building on these insights can help people improve their identification skills and become more resilient to deceptive digital media.