Journalists are most likely to receive abuse: Analysing online abuse of UK public figures across sport, politics, and journalism on Twitter

Overview

• Engaging with social media is important for public figures to connect with audiences and share ideas
• However, public figures often face disproportionate levels of abuse on these platforms, degrading public discourse
• While research on abuse of groups like politicians and journalists exists, little has been done to understand differences in abuse dynamics across different types of public figures

Plain English Explanation

Prominent people in society, like politicians, sports stars, and journalists, often use online social media platforms to interact with their audiences and spread their ideas. However, these public figures tend to receive a higher amount of hateful and abusive messages on these platforms compared to regular users. While there has been research on the abuse faced by certain groups like politicians and journalists, not much work has been done to systematically understand how the patterns of abuse may differ across various types of public figures.

To address this, the researchers analyzed a large dataset of 45.5 million tweets targeted at 4,602 public figures from three different domains: members of parliament (MPs), professional football (soccer) players, and journalists. They used advanced language models to identify abusive content in these tweets.

The key findings are:

• MPs receive the most abuse in total, but journalists are the most likely to receive abuse when accounting for other factors
• Abuse is unevenly distributed, with a small number of individuals receiving the majority of it
• Abuse is more event-driven for some groups, like footballers, compared to others
• Having a larger online presence and being male are associated with higher levels of abuse across all groups

Technical Explanation

The researchers analyzed a dataset of 45.5 million tweets directed at 4,602 UK public figures across three domains: members of parliament (MPs), professional footballers, and journalists. They used fine-tuned transformer-based language models to label the tweets as abusive or not.

Their analysis revealed several key insights:

• Abuse Prevalence: While MPs received the most abuse in absolute terms, journalists were the most likely to receive abuse after controlling for other factors like online presence and gender.
• Uneven Distribution: Abuse was unevenly distributed, with a small number of individuals in each group receiving the majority of the abusive tweets.
• Temporal Patterns: Abuse was more event-driven for some groups, like footballers, compared to others, where it was more consistent over time.
• Predictive Factors: Having a larger online presence and being male were associated with higher levels of abuse across all three domains.

The researchers argue that understanding these differences in abuse dynamics can inform efforts to protect and support public figures, particularly those who are more vulnerable to online harms.

Critical Analysis

The study provides valuable insights into the dynamics of online abuse targeting different types of public figures. However, it is important to note some potential limitations and areas for further research:

1. Context and Nuance: The study relies solely on automated text analysis, which may miss important contextual nuances and fail to capture the full complexity of abusive interactions.

2. Generalizability: The study focuses on UK-based public figures, and the findings may not be directly applicable to other cultural or political contexts.

3. Intersectionality: The analysis examines abuse along the dimensions of occupation and gender, but does not consider how other intersecting factors, such as race or socioeconomic status, may shape the experience of online abuse.

4. Platform Differences: The study only examines abuse on Twitter, and the dynamics may differ on other social media platforms with distinct user demographics and norms.

5. Potential Biases: The use of machine learning models, even when carefully fine-tuned, may introduce biases that could affect the accuracy of the abuse detection.

Further research is needed to address these limitations and provide a more comprehensive understanding of the multifaceted nature of online abuse targeting public figures.

Conclusion

This study offers valuable insights into the differential patterns of online abuse faced by various types of public figures, including politicians, football players, and journalists. The findings highlight the uneven distribution of abuse, with a small number of individuals bearing the brunt of it, as well as the role of factors like online presence and gender in shaping abuse dynamics.

These insights can inform efforts to protect and support public figures who are disproportionately targeted by online abuse, helping to maintain the health and integrity of public discourse. However, further research is needed to address the limitations of the study and provide a more comprehensive understanding of this complex issue.