Hate Speech in the Political Discourse on Social Media

Summary

This presentation explores the impact of Twitter on political communication, focusing on hate speech directed at U.S. Congress members. It examines how gender, ethnicity, and party affiliation affect the amount of hate speech politicians receive in their replies. The study finds that women and people of color are disproportionately targeted, with Democrats facing more hate speech than Republicans. It highlights how sentiment in original tweets correlates with the level of harassment. The research suggests that social media's dual role in fostering dialogue and amplifying hate speech poses risks for political participation, particularly for underrepresented groups.

Takeaways

• 😀 Social media, especially Twitter, has become a major tool for politicians to engage with constituents, allowing for unidirectional communication, direct dialogue, and political mobilization.
• 😀 The shift to social media is not entirely positive; it fosters echo chambers, us-vs-them rhetoric, and can lead to harassment and hate speech, especially in emotionally charged discussions like politics.
• 😀 Hate speech on social media is typically defined as abusive or threatening speech directed at specific groups based on factors like ethnicity or sexual orientation.
• 😀 The study focuses on understanding how hate speech on Twitter affects U.S. Congress members based on their gender, ethnicity, and political affiliation.
• 😀 Research questions aim to explore whether U.S. Congress members are more likely to receive hate speech depending on their personal characteristics and the sentiment of their tweets.
• 😀 The dataset for this research includes data from all 541 members of the 117th U.S. Congress, looking at political affiliation, gender, ethnicity, and tweet sentiment.
• 😀 The study used sentiment analysis to assess the emotional tone of original tweets, ranging from highly negative (-4) to highly positive (+4).
• 😀 Hate speech detection was achieved through a machine learning model trained on an annotated dataset of tweets, categorizing them as regular, offensive, or hateful.
• 😀 Key findings include that women and people of color in Congress receive significantly more hate speech than their male and white counterparts.
• 😀 The research shows that tweet sentiment is directly linked to the level of hate speech; more negative tweets tend to receive more hateful replies, with a stronger effect for Democrats.
• 😀 The study highlights a major issue of inequality in how different groups of politicians are treated online, with implications for their participation in politics and democracy as a whole.
• 😀 Future research could expand by comparing results across countries and social media platforms and exploring the social networks of users posting hateful replies to better understand the sources of hate speech.

Q & A

What is the primary focus of the research presented in the script?

The research focuses on examining the prevalence of hate speech in replies to U.S. Congress members on Twitter, particularly how hate speech varies based on gender, ethnicity, and political affiliation.

How has social media usage by politicians evolved over time?

Social media usage by politicians has significantly increased. In 2009, around 70 members of Congress used Twitter, while today, almost all politicians use it for both professional and personal communication.

What are the three main ways social media is used by politicians according to the script?

Politicians use social media for unidirectional communication with constituents, dialogue and direct feedback from constituents, and political mobilization, such as rallying support for projects or crowdfunding.

What negative effects does social media have, as mentioned in the study?

Social media can foster echo chambers, us-vs-them rhetoric, harassment, and hate speech. It also contributes to further polarization, erosion of intergroup relationships, and the spread of misinformation.

What are the main research questions explored in the study?

The study explores whether U.S. Congress members are more likely to receive hate speech based on their gender and ethnicity, and whether hate speech in replies depends on the sentiment of the original tweet.

What dataset was used in this study and how was it constructed?

The dataset consists of data from 541 members of the 117th U.S. Congress, collected from their official websites and Twitter feeds. This includes demographic details, political party affiliation, gender, ethnicity, and their original tweets, not replies or retweets.

How was hate speech identified in the study?

Hate speech was identified using an annotated Twitter dataset from Davidson Attack, which categorized tweets as regular speech, offensive speech, or hate speech. A neural network was trained using this dataset to classify replies as hate speech or not.

What are the main findings regarding hate speech based on gender and ethnicity?

The study found that women and people of color received more hate speech than men and white individuals, with persons of color facing 40% more hate speech. Additionally, Democratic Congress members were more likely to receive hate speech compared to Republicans.

How does sentiment in the original tweet affect the likelihood of receiving hate speech?

Negative sentiment in the original tweet increases the likelihood of receiving hate speech. The study found that negative tweets were correlated with higher levels of hate speech, but this effect was stronger for Democrats compared to Republicans.

What are the implications of these findings for political participation?

The findings suggest that marginalized groups, such as women and people of color, may be discouraged from participating in politics due to the hostile environment on social media, which could negatively impact democracy and the inclusivity of political discourse.

What are the two main areas of future research suggested by the study?

Future research could compare the prevalence of hate speech on social media across different countries and platforms, and investigate the social networks of individuals posting hate speech to understand their motivations and context.