Digital Hostility Toward LGBTQIA+ Research Recruitment on Social Media Using Topic Modeling and Sentiment Analysis of Facebook Comments: Quantitative Content Analysis Study.

BACKGROUND

Lesbian, gay, bisexual, transgender, queer/questioning, intersex, asexual (LGBTQIA+) researchers and participants frequently encounter hostility in virtual environments, particularly on social media platforms where public commentary on research advertisements can foster stigmatization. Despite a growing body of work on researcher virtual hostility, little empirical research has examined the actual content and emotional tone of public responses to LGBTQIA+-focused research recruitment.

OBJECTIVE

This study aimed to analyze the thematic patterns and sentiment of social media comments directed at LGBTQIA+ research recruitment advertisements, in order to better understand how virtual stigma is communicated and how it may impact both researchers and potential participants.

METHODS

A total of 994 publicly visible Facebook comments posted in response to LGBTQIA+ recruitment advertisements (January to April 2024) were collected and analyzed. Text preprocessing included tokenization, stop-word removal, and lemmatization. Latent Dirichlet allocation was used to identify latent themes across the dataset. Sentiment analysis was conducted using the Bing Liu and National Research Council lexicons, with scores ranging from -1 (most negative) to 1 (most positive). Linguistic Inquiry and Word Count was used to quantify psychological and moral language features. Comments were also manually coded into four audience target groups (researchers, LGBTQIA+ community, general public, and other commenters), and language category differences were analyzed using 1-way ANOVAs with Bonferroni corrections.

RESULTS

Topic modeling identified three key themes: (1) "Transitions, Health, and Gender Dysphoria," (2) "Polarized Debate and Response," and (3) "Religious and Ideological Debates." Topic 2 had the highest average prevalence (average γ=0.486, SD 0.21). Sentiment analysis revealed negative mean sentiment scores for all three topics: Topic 1 (-0.41, SD 0.48), Topic 2 (-0.21, SD 0.44), and Topic 3 (-0.35, SD 0.46). No topic exhibited a statistically significant predominance of positive sentiment. A 1-way ANOVA showed significant differences in linguistic tone across target groups: negative tone (F=12.84; P<.001), swearing (F=16.07; P<.001), and anger-related language (F=9.45; P<.001), with the highest levels found in comments directed at researchers. Comments targeting LGBTQIA+ individuals showed higher references to mental illness, morality, and threats to children. While affirming responses were less frequent and typically appeared within confrontational contexts, their presence highlights significant moments of solidarity and resistance.

CONCLUSIONS

This study documents a persistently hostile virtual environment for LGBTQIA+ research, where researchers are frequently dehumanized and LGBTQIA+ identities are pathologized. These findings reinforce stigma communication models and suggest a need for institutional responses that include mental health support, enhanced moderation tools, and policy advocacy. Future research should investigate how hostile discourse affects researchers' well-being and recruitment outcomes, and evaluate interventions to foster more respectful engagement with LGBTQIA+ studies.