Analyzing Self-Presentation in Twitter Bios and its Associations with Misinformation Sharing

Conceitos Básicos

Self-presentation in Twitter bios, as captured through novel embedding models, is associated with increased sharing of low-quality news, particularly for older, right-leaning, and more religious users.

Resumo

This study proposes and evaluates three models to embed Twitter bios in socially meaningful latent spaces, capturing dimensions like age, gender, partisanship, and religiosity. The models leverage patterns in the multiple social identities that users include in their bios.
The evaluation shows that the fine-tuned SBERT model outperforms several baselines in capturing perceptions of self-presentation, as validated through human surveys. The authors then use this model to explore two key research questions:

Interaction between age and partisanship: The study finds that Twitter users who present as older and right-leaning share a much higher proportion of low-quality news compared to those who are only older or only right-leaning.

Role of religiosity: The analysis reveals that individuals who self-present as more religious are also more likely to share a higher proportion of low-quality news, potentially in interaction with their political orientation.

These findings provide new insights into how different dimensions of self-presentation on social media are associated with the spread of misinformation, with implications for targeted interventions. The publicly available models can also be used by other researchers to study self-presentation and its links to behavior across different contexts.

Estatísticas

"Older and right-leaning Twitter users share a much higher proportion of low-quality news compared to those who are only older or only right-leaning."
"Individuals who self-present as more religious are also more likely to share a higher proportion of low-quality news."

Citações

"Self-presentation in Twitter bios, as captured through novel embedding models, is associated with increased sharing of low-quality news, particularly for older, right-leaning, and more religious users."
"The fine-tuned SBERT model outperforms several baselines in capturing perceptions of self-presentation, as validated through human surveys."

Principais Insights Extraídos De

Measuring Dimensions of Self-Presentation in Twitter Bios and their Links to Misinformation Sharing

by Navid Madani... às arxiv.org 09-19-2024

https://arxiv.org/pdf/2305.09548.pdf

Perguntas Mais Profundas

How do the associations between self-presentation and misinformation sharing vary across different social media platforms?

The associations between self-presentation and misinformation sharing can vary significantly across different social media platforms due to the unique user demographics, platform functionalities, and cultural contexts inherent to each site. For instance, Twitter, as highlighted in the study, allows users to present themselves through concise bios, which can encapsulate multiple social identities. This self-presentation is linked to behaviors such as sharing URLs from low-quality news sites, particularly among older, right-leaning individuals. In contrast, platforms like Facebook or Instagram may encourage more visual self-presentation, where users curate their identities through images and posts rather than textual bios. This could lead to different associations with misinformation sharing, as the motivations for sharing content may be influenced by the visual context and social interactions prevalent on these platforms.
Moreover, the nature of misinformation itself can differ across platforms. For example, TikTok's algorithm-driven content discovery may lead to rapid dissemination of misinformation through engaging video formats, which could attract a younger audience that self-presents differently than older users on Twitter. The interactive features of platforms like Reddit, where users can engage in discussions and debates, may also shape how self-presentation influences the sharing of misinformation, as users may feel more accountable for their shared content in a community-driven environment. Thus, understanding the nuances of self-presentation across various social media platforms is crucial for comprehensively addressing the spread of misinformation.

What other dimensions of self-presentation, beyond age, partisanship, and religiosity, might be linked to the spread of misinformation online?

Beyond age, partisanship, and religiosity, several other dimensions of self-presentation could be linked to the spread of misinformation online. One significant dimension is education level. Users with varying educational backgrounds may present themselves differently and have differing capacities to critically evaluate information. For instance, individuals who self-identify as highly educated may be less likely to share misinformation, while those who emphasize a lack of formal education might be more susceptible to sharing low-quality news.
Another dimension is geographic location. Users from different regions may have distinct cultural contexts and exposure to misinformation, influencing their self-presentation and sharing behaviors. For example, individuals in politically polarized regions may self-present in ways that align with local narratives, potentially increasing the likelihood of sharing misinformation that resonates with their community.
Occupational identity is also a relevant dimension. Users who identify as professionals in fields such as journalism, education, or healthcare may present themselves with a sense of authority, which could impact their sharing behaviors. Conversely, individuals in less formal occupations may engage with misinformation differently, potentially sharing sensational content that aligns with their social circles.
Lastly, social identity factors such as race, gender, and sexual orientation can also play a role in how users self-present and engage with misinformation. These identities can shape the narratives users are exposed to and the types of misinformation they may be inclined to share, reflecting broader societal biases and stereotypes.

Could targeted interventions that account for the interactive effects of age and partisanship, as well as religiosity, be more effective in combating the spread of misinformation?

Yes, targeted interventions that account for the interactive effects of age, partisanship, and religiosity could be significantly more effective in combating the spread of misinformation. The findings from the study indicate that individuals who self-present as older and right-leaning are more likely to share low-quality news, suggesting that these demographic factors interact in ways that amplify misinformation sharing.
By designing interventions that specifically address these interactions, stakeholders can tailor their messaging and strategies to resonate with the unique motivations and concerns of these groups. For instance, educational campaigns that focus on critical media literacy could be developed for older, right-leaning individuals, emphasizing the importance of verifying sources and understanding the implications of sharing misinformation.
Additionally, interventions could leverage the role of religiosity, which was found to correlate with misinformation sharing. Campaigns that engage religious communities through trusted leaders or platforms could help disseminate accurate information and counteract misinformation narratives that may be prevalent within those groups.
Furthermore, utilizing data-driven approaches to identify high-risk demographics based on self-presentation patterns can enhance the precision of interventions. By understanding the specific contexts and identities of users, interventions can be more effectively designed to mitigate misinformation sharing, ultimately fostering a more informed public discourse.