(PS9-11) Life-Threatening Thoughts and Behaviors on a Social Media Platform: Exploring Bystander Recognition

Abstract Text: Over the past decade, the United States has experienced a surge in social media use. While designed to connect users through a virtual platform for social connection, these platforms have become a commonplace for individuals sharing life-threatening thoughts and behaviors. Once shared, an audience of followers subsequently become a bystander to the individual threatening to lethally injure oneself or others. As a bystander, one may have to decide whether to intervene with the author of the post. To reach the point of intervention, however, the bystander must first recognize they have encountered a life-threatening post. To date, there is little known literature exploring such bystander ability in life-threatening social media. As such, the current study sought to evaluate bystanders’ recognition of life-threatening posts on a social media platform, and factors which may implicate this ability. The participants of the study (N = 709) were from an undergraduate university and identified as mostly White (54.9%) and female (83.5%). Participants were shown a series of 12 vignettes which mocked social media feeds found on X/Twitter, followed by a series of questions to assess recognition of life-threatening behavior within the previously viewed feed. A total of 6 life-threatening posts which directed harm at the poster, and 6 life-threatening posts which directed harm toward others on school grounds, were randomly placed within the feeds. Additionally, 3 posts from each life-threatening category were further manipulated to either contain a photo of the lethal means (e.g., firearm, pills) or no photo in the post. Sensitivity and specificity were measured for the overall accuracy of the model, followed by a series of chi-squared tests of independence which examined recognition differences between groups of posts. Results indicate that the overall model demonstrated a sensitivity of 80.1% and specificity of 92.9%. Participants were significantly better at recognizing posts with life-threatening remarks directed towards others rather than directed at the poster (X2 (1, N = 709) = 656.60, p < .001, V = .39). Additionally, participants were significantly better at recognizing life-threatening posts which included a photo of lethal means compared to posts without a photo (X2 (1, N = 709) = 596.79, p < .001, V = .37). This held true for both life-threatening posts directed at the self (X2 (1, N = 709) = 274.38, p < .001, V = .36) and others (X2 (1, N = 709) = 322.81, p < .001, V = .39). Bonferroni’s post-hoc correction was applied to the models and yielded no changes to p-values. Findings provide preliminary support that factors within life-threatening social media posts (e.g., directionality of threat, inclusion of photo) may affect one’s ability to recognize threats. Intervention and prevention efforts targeting such life-threatening behaviors on social media may consider incorporating psychoeducation for bystanders into their programs which target factors commonly missed or not recognized as life-threatening. Future research should continue investigating the role of the bystander in life-threatening social media, including threat assessment and intervention motivation.