Relevance & Research Question

4chan is an anonymous online forum known for ephemeral content and internet subcultures. This research examines how masculinities are constructed and contested in such spaces, where gender norms are negotiated through irony, confrontation, and subcultural slang. Combining theories of masculinity with computational text analysis, it bridges qualitative research and data-driven modeling of online discourse. It asks how different types of masculinities (hegemonic, hybrid, negotiating, caring) are constructed and circulated online and how theory-driven annotation and lexicon development can support computational identification.
Methods & Data

This study employs a mixed-method design integrating computational text analysis and qualitative annotation. It uses a comprehensive 2.5-year dataset of over 328M 4chan posts, encompassing textual data and metadata across all boards collected via a 4chan text collection tool. The platform’s unfiltered nature allows gender and politics to frequently intersect across diverse contexts. The empirical pipeline includes preprocessing (cleansing, fixing misspelled, broken, or repeated words), identifying masculinity-related discourse through SentenceBert and extracting candidate terms for a theory-driven annotation schema grounded in gender studies and discourse theory. The schema operationalizes masculinity types through parameters such as dominance, emotionality, caregiving, and norm stance. Human annotators use a web-based interface (Doccano) supported by automated term suggestions and inter-annotator agreement metrics. The resulting annotations form the basis for a masculinity lexicon and subsequent computational modeling and clustering to explore discursive variations across online communities.

Results

This study generates a dataset and lexicon capturing linguistic construction of different masculinity types. These resources support qualitative interpretation and large-scale computational modeling, revealing patterns in identity performance and negotiation of gender norms. Methodologically, it demonstrates how theory-driven annotation combined with embedding-based term extraction and automated quality control, providing a scalable framework for future studies of digital gender expression.
Added Value

Conceptually, this study illuminates how different types of masculinities are performed and negotiated in online communities, linking these patterns to broader social and political discourses. Methodologically, it demonstrates how theory-driven annotation and embedding-based lexicon development can create scalable tools for analyzing gendered language. The resulting dataset, annotation framework, and lexicon provide reusable resources for future research across platforms and contexts.

Relevance & Research Question
As survey self-reports on digital behavior are often inaccurate, social and behavioral researchers increasingly augment web surveys with digital traces. Existing research has mostly focused on collecting these traces, leaving open the question of how to further analyze them. The present study utilize traces in the form of metered data to measure one digital behavior: online information-seeking in the context of (expectant) parenthood. We investigate the following research question: How can traces of website visits, online searches, and app usage be transformed into measurements of online information-seeking behavior?
Methods & Data
We obtained metered data and self-reports on (expectant) parenthood through the Netquest opt-in online panel in Spain. The sample (n = 800) includes both participants who self-reported being and not being (expectant) parents. In 2024, Netquest extracted participants’ metered data to previously specified websites, search terms, apps, and URL keywords over a period of 12 to 24 weeks. Before data extraction, an external ethics advisor reviewed and approved the study. As the large volume of extracted traces (approx. 184,000) renders manual, human-based coding infeasible, we classify the traces using locally deployed Large Language Models (LLMs).
Results
Preliminary results indicate that about 85% of parents engaged in website visits, 70% in online searches, and 30% in app usage related to online information-seeking behavior. Next, we will investigate the impact of key analytical decisions on the measurements generated. Specifically, we will systematically vary the type of trace considered (e.g., website visits, search terms, or app usage), the LLM used (e.g., Llama or Gemma), and the aggregation rules applied (e.g., total duration or frequency). To evaluate the impact of these decisions, we will determine criterion validity by using participants’ self-reports on (expectant) parenthood.
Added Value
Our study stands out from previous methodological research, which mostly focused on the collection of traces, and builds an important knowledge basis for future studies analyzing traces (or metered data). By shedding light on key analytical decisions and their impact on measurement outcomes, it contributes to informed research workflows and ultimately more accurate measurements of digital behavior, including online information-seeking.

Relevance & Research Question
This study aimed to evaluate the effectiveness of social media platforms, specifically Meta and TikTok, as data collection tools and to explore their influence on body image perception, with a focus on gender differences. The goal was to assess these platforms' efficacy in gathering data on sensitive topics like body image.
Methods & Data

Advertising campaigns were conducted on Meta and TikTok to recruit participants for an online survey about body image perception. Throughout the campaign, various images were used for the advertisements, comparing and testing the performance. Additionally, an experiment was conducted to assess the impact of an incentive as a motivation for engaging with the questionnaire. Data analysis focused on campaign performance metrics and survey responses, employing logistic regression and cluster analysis to examine the relationship between social media use and body satisfaction.
Results
The campaigns demonstrated social media's potential to reach a large audience quickly, with TikTok generating high visibility and Meta platforms, especially Facebook, ensuring acceptable participation rates. However, self-selection bias and sample representativeness emerged as concerns. The survey results revealed a clear link between social media use and body perception. Logistic regression showed that perceiving social media content as realistic and influential significantly increased the likelihood of body dissatisfaction, up to four times higher. Age emerged as a protective factor, while gender differences were less pronounced than expected. Cluster analysis identified three distinct user profiles based on perceived social media influence and content realism.

Added Value
This study provides insights into the effectiveness of social media as a data collection tool for sensitive topics and contributes to understanding the complex relationship between social media use and body image perception. It highlights the need for cautious interpretation of data collected through these platforms due to potential biases. The findings suggest that perceived realism of social media content plays a crucial role in body dissatisfaction, offering valuable implications for promoting more conscious social media use and personal well-being. Future research directions are proposed to address limitations and expand on these findings.