Measuring the Stigmatizing Effects of a Highly Publicized Event on Online Mental Health Discourse

Title of the Paper

Measuring the Stigmatizing Effects of a Highly Publicized Event on Online Mental Health Discourse

Paper Information

• Subject Areas: Social Psychology, Online Communities, Human-Computer Interaction (HCI)
• Keywords: Online Communities, Mental Health, Mental Health Stigmatization, Media Influence, Social Media, Depp-Heard Trial, LIWC, Linguistic Analysis, Causal Impact Analysis, Thread Psychology

Research Background and Issues

• Problems and Challenges:

  • Mental health stigmatization has long been a significant issue affecting the lives of individuals with psychological disorders, including societal prejudice and self-stigmatization.
  • Media coverage plays a crucial role in shaping public understanding of mental health but often employs stigmatizing language or misrepresentation, potentially exacerbating societal biases.
  • Social media has become a primary venue for discussions on mental health topics, yet little is known about how external events influence the use of stigmatizing language in online mental health discourse.

• Research Significance:

  • Understanding the spread and impact of stigmatizing language in media, particularly on social media, can inform the design of effective anti-stigmatization interventions.
  • The language used in online communities profoundly affects users' mental health and their behaviors in seeking support.

• Research Motivation and Related Work:

  • Existing studies indicate that media events can have profound effects on public attitudes toward mental health issues, but the specific mechanisms and impacts remain unclear.
  • Related work has primarily focused on traditional media; this study emphasizes the online environment, particularly the dynamic changes in mental health-related language on social media platforms like Twitter.

Proposed Solution

• Proposed Method:

  • Using the 2022 Johnny Depp and Amber Heard defamation trial as a case study, this research investigates the impact of a highly publicized media event on the language used in mental health discussions on Twitter.
  • By combining linguistic analysis methods (e.g., Linguistic Inquiry and Word Count, LIWC tool) with causal impact analysis, the study quantifies changes in the levels of stigmatizing and de-stigmatizing language.

• Innovations:

  • The first study to use detailed psychological linguistic models to analyze the dynamic impact of high-profile media events on stigmatizing language in social media.
  • By setting control time periods and using data unrelated to the event, the study employs Bayesian structural time-series models (Causal Impact Analysis) to assess the direct causal effects of the event.

• Implementation Steps and Techniques:

  1. Data Collection:
     • Tweets were collected from 51 days before the event to 51 days after its conclusion, focusing on keywords such as "mental health" and "borderline personality disorder."
  2. Linguistic Analysis:
     • The LIWC tool was used to score tweets based on stigmatizing and de-stigmatizing language using multiple psychological dictionaries.
  3. Causal Analysis:
     • Causal impact models were employed to evaluate the probability and effect size of changes in stigmatizing and de-stigmatizing language scores due to the event.
  4. Comparative Analysis:
     • Language changes in tweets unrelated to specific mental health disorders were compared to verify the event-specific impact.

Research Findings

• Specific Findings:

  • During and after the trial, stigmatizing language in tweets related to personality disorders increased significantly, while de-stigmatizing language decreased notably.
  • Stigmatizing language saw a marked rise in terms of animalistic dehumanization (e.g., "creature") and words related to power and violence (e.g., "assault").
  • Self-disclosure (use of first-person pronouns like "I" and "We"), an important form of de-stigmatizing language, significantly decreased (by 48%).
  • The impact of stigmatizing language persisted for nearly two months after the trial, indicating the event's lasting effects.

• Advantages Compared to Existing Solutions:

  • Provides fine-grained, quantified data and model support for changes in stigmatizing language.
  • Expands the understanding of the dynamics of stigmatization in online mental health discussions, particularly on social media.

• Experimental Evaluation and Model Results:

  • Causal analysis confirmed that the Depp-Heard trial increased the probability of stigmatizing language by 97.3%, with an approximate growth of 17%.
  • General "mental health" tweets were also affected by the event, with stigmatizing language increasing by 11% and de-stigmatizing language decreasing by 3%.

• Limitations and Future Directions:

  • This study focuses solely on Twitter; results may differ on other social media platforms (e.g., Facebook, Reddit).
  • Only the LIWC tool was used for quantitative analysis; future studies could incorporate topic modeling and other methods for richer analysis.
  • Further refinement of stigmatization categories (e.g., the influence of gender, occupation, etc.) and extending the time window are potential directions for future research.