Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Social Media and Academia: How Gender Influences Online Scholarly Discourse

Rrubaa Panchendrarajan, Harsh Saxena, Akrati Saxena

TL;DR

This paper investigates how gender influences online scholarly discourse among computer science academics at the top US universities on X/Twitter. It employs a multi-method pipeline, including topic modeling on 11 clusters derived from BERTweet embeddings with KMeans, sentiment analysis via Pysentimiento, emotion analysis with TweetNLP, and writing-style assessment using a Mixtral LLM, complemented by harassment analysis through Perspective API and a BertTweet-based gender classifier. Analyzing 501,708 posts from 151 female and 476 male academics (collected in Jan 2023, with 2022 replies), the study finds that men engage more on AI/ML and current US-society topics, while women show relatively more activity around AI events/workshops; women also exhibit stronger positive and negative sentiments and greater empathy in writing. Notably, replies to female academics contain more threats and severe toxicity, highlighting gender-based disparities in online harassment. The findings underscore the need for more inclusive online scholarly spaces and inform platform moderation and policy design, while acknowledging limitations such as sample bias and the binary gender assumption.

Abstract

This study investigates gender-based differences in online communication patterns of academics, focusing on how male and female academics represent themselves and how users interact with them on the social media platform X (formerly Twitter). We collect historical Twitter data of academics in computer science at the top 20 USA universities and analyze their tweets, retweets, and replies to uncover systematic patterns such as discussed topics, engagement disparities, and the prevalence of negative language or harassment. The findings indicate that while both genders discuss similar topics, men tend to post more tweets about AI innovation, current USA society, machine learning, and personal perspectives, whereas women post slightly more on engaging AI events and workshops. Women express stronger positive and negative sentiments about various events compared to men. However, the average emotional expression remains consistent across genders, with certain emotions being more strongly associated with specific topics. Writing-style analysis reveals that female academics show more empathy and are more likely to discuss personal problems and experiences, with no notable differences in other factors, such as self-praise, politeness, and stereotypical comments. Analyzing audience responses indicates that female academics are more frequently subjected to severe toxic and threatening replies. Our findings highlight the impact of gender in shaping the online communication of academics and emphasize the need for a more inclusive environment for scholarly engagement.

Social Media and Academia: How Gender Influences Online Scholarly Discourse

TL;DR

This paper investigates how gender influences online scholarly discourse among computer science academics at the top US universities on X/Twitter. It employs a multi-method pipeline, including topic modeling on 11 clusters derived from BERTweet embeddings with KMeans, sentiment analysis via Pysentimiento, emotion analysis with TweetNLP, and writing-style assessment using a Mixtral LLM, complemented by harassment analysis through Perspective API and a BertTweet-based gender classifier. Analyzing 501,708 posts from 151 female and 476 male academics (collected in Jan 2023, with 2022 replies), the study finds that men engage more on AI/ML and current US-society topics, while women show relatively more activity around AI events/workshops; women also exhibit stronger positive and negative sentiments and greater empathy in writing. Notably, replies to female academics contain more threats and severe toxicity, highlighting gender-based disparities in online harassment. The findings underscore the need for more inclusive online scholarly spaces and inform platform moderation and policy design, while acknowledging limitations such as sample bias and the binary gender assumption.

Abstract

This study investigates gender-based differences in online communication patterns of academics, focusing on how male and female academics represent themselves and how users interact with them on the social media platform X (formerly Twitter). We collect historical Twitter data of academics in computer science at the top 20 USA universities and analyze their tweets, retweets, and replies to uncover systematic patterns such as discussed topics, engagement disparities, and the prevalence of negative language or harassment. The findings indicate that while both genders discuss similar topics, men tend to post more tweets about AI innovation, current USA society, machine learning, and personal perspectives, whereas women post slightly more on engaging AI events and workshops. Women express stronger positive and negative sentiments about various events compared to men. However, the average emotional expression remains consistent across genders, with certain emotions being more strongly associated with specific topics. Writing-style analysis reveals that female academics show more empathy and are more likely to discuss personal problems and experiences, with no notable differences in other factors, such as self-praise, politeness, and stereotypical comments. Analyzing audience responses indicates that female academics are more frequently subjected to severe toxic and threatening replies. Our findings highlight the impact of gender in shaping the online communication of academics and emphasize the need for a more inclusive environment for scholarly engagement.
Paper Structure (13 sections, 7 figures, 6 tables)

This paper contains 13 sections, 7 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Academic Discourse on Social Media
  4. Gender Biases in Social Media
  5. Data Collection
  6. Analysis and Findings
  7. Analysis of Tweets and Retweets
  8. Analyzing Topics.
  9. Analyzing Sentiments.
  10. Analyzing Emotions.
  11. Analyzing Writing Style.
  12. Analysis of Replies
  13. Discussion and Conclusion

Figures (7)

  • Figure 1: The Silhouette scores across varying cluster size.
  • Figure 2: Statistics of the identified topics and tweets, retweets and response corresponding to each topic.
  • Figure 3: Weekly avg. of positive and negative sentiment scores by both gender with key events identified at sentiment peaks.
  • Figure 4: Average emotion expressed in tweets and retweets posted by male and female academics
  • Figure 5: Writing Analysis
  • ...and 2 more figures