Social
Social media-derived LGBTQ+ sentiment and teen mental health Junaid S Merchant* Thu T. Nguyen Heran Mane Xiaohe Yue
The proliferation of state-level policies targeting LGBTQ+ youth reflects growing divisions in public attitudes and raises concerns about the psychological impacts of these trends amidst historical lows in teen mental health. This study utilized social media data to measure public sentiment about LGBTQ+ issues and examined associations with mental health outcomes of US teens in the 2019 release of the Youth Risk Behavior Surveillance System (YRBSS). X (formerly Twitter) posts (N=18,000,000+) from 2016-2022 were collected for an ongoing study. Posts were filtered for LGBTQ+ keywords, classified for positivity or negativity using a machine learning sentiment model, and averaged at geographic and temporal scales to quantify LGBTQ+ sentiment as the proportion of posts classified in each category. Time-series of nationally averaged LGBTQ+ exhibit sensitivity to LGBTQ+-relevant historical events, like the rise in positivity during Pride events and spikes in negative sentiment following anti-LGBTQ+ policy discussions in 2019 (Figure 1a). We evaluated the relationship between state-level LGBTQ+ sentiment and self-reported suicidal ideation in the past 12 months (yes/no binary responses) among high school students (N=160,112) across 44 states. Results show that state-level negative LGBTQ+ sentiment is associated with slightly elevated risk for suicidal ideation (odds ratio = 1.014 [1.00, 1.027]) even when adjusted for individual- and state-level factors. We uncovered novel patterns of disparities when disaggregating identity dimensions across 5 races, 3 sexual orientations, and 2 genders (=30 intersecting strata; Figure 1b) indicating stronger associations between LGTBQ+ sentiment and health among homo- and bisexuals compared to heterosexual teens. Our findings highlight the utility of social media data for examining the ongoing challenges of LGBTQ+ health and the importance of considering intersecting identities in health disparities research.
