With generative artificial intelligence’s (GenAI) growing popularity, individuals are increasingly using it when searching for election-related information. This scenario raises concerns that GenAI usage may result in widespread dissemination of misinformation, given its ability to generate seemingly authentic information. Nevertheless, despite the importance of Gen AI, few researchers have examined how individuals use this tool to search for election-related information. This study aims to assess how GenAI’s perceived system (i.e., accessibility and integration) and information quality (i.e., completeness, accuracy, and neutrality) impact its usage. Focusing on the 2024 U.S. presidential election, we conducted a two-wave survey and data was collected from 364 Americans. Participants were found to have a favorable attitude overall toward GenAI. Further, accuracy and neutrality were positively associated with GenAI usage. A fuzzy set qualitative comparative analysis was also conducted to identify different configurations of perceived system and information quality that led to high GenAI usage. Analyzing the qualitative responses further confirmed the results. This study contributes to the literature on the role of GenAI during elections, providing a nuanced understanding of how dimensions of GenAI’s perceived system and information quality impact individuals’ GenAI usage. The findings have significant practical implications for dealing with the (mis)information generated by GenAI.

The potential of large language models (LLMs) to mitigate the time- and cost-related challenges associated with inductive thematic analysis (ITA) is being increasingly explored in the literature. However, the use of LLMs to support ITA has often been opportunistic, relying on ad hoc prompt engineering (PE) approaches, thereby undermining the reliability, transparency, and replicability of the analysis. The goal of this study is to develop a structured approach to PE in LLM-assisted ITA. To this end, a comprehensive review of the existing literature is conducted to examine how researchers applying ITA integrate LLMs into their workflows and, in particular, how PE is utilized to support the analytical process. Built on the insights generated from this review, four key steps for effective PE in LLM-assisted ITA are identified and proposed. Furthermore, the study explores advanced PE techniques that can enhance the execution of these steps, providing researchers with practical strategies to improve their analyses. In conclusion, the main contributions of this paper include: (i) mapping the existing research on LLM-assisted ITA to enable a better understanding of the rapidly developing field, (ii) proposing a structured four-step PE process to enhance methodological rigor, (iii) discussing the application of advanced PE techniques to support the execution of these steps, and (iv) highlighting key directions for future research.

Online political campaigning takes place on several platforms, suggesting the need for those seeking voter support—such as political parties—to adapt to the characteristics of each platform. Getting voters to take action—be it online (such as asking them to engage with posts) or offline (such as asking them to attend rallies or to vote)—is a key element of campaigning efforts. Here, we focus on the use of what is referred to as calls to action as employed by Norwegian political parties on three different social media platforms (Facebook, Instagram, and Twitter) between 2013 and 2024. Using a combination of automated and manual content analysis, the results indicate that while Facebook is the preferred platform for providing calls to action during roughly the first half of our time period, Instagram takes the lead in this regard for the latter half. Overall, though, we see a clear decrease of calls to action across all three platforms, indicating the changing priorities of parties. Using likes as a common measurement of engagement across all three studied platforms, posts containing calls to action emerged as less popular towards the end of the time period for Facebook and Twitter, while users of Instagram appear to be more interested in engaging with such posts also during these latter stages. The study ends with a discussion of the main findings, also suggesting some ways forward for future research efforts.

This study employs an agent-based model (ABM) simulation grounded in polarization theory to explore how social and media influences shape affective polarization, social diversity, and media diversity within a hostile partisan information environment. We examine four scenarios: minimal, strong, social-dominated effect, and media-dominated effect, by iterating 16 combinations of parameters, including influence strength, homogeneous discussion, and selective exposure rates. Calibrated with a real-world U.S. population dataset, results show that while social and media influences accelerate sorting, selectivity structure is the primary driver of affective polarization. Notably, low homogeneous discussion and low selective exposure produce the highest polarization levels across four scenarios due to a probabilistic backfire effect that reflects identity-protective cognition. Strong, locally concentrated social influence sharply reduces social diversity, whereas media influence alone cannot produce convergence without social reinforcement. Media diversity proves more resilient due to global exposure, though it declines under high selective exposure and strong media influence. Despite initial partisan gaps, final-stage outcomes reveal minimal differences between Democrats and Republicans across all conditions.

This article presents a sociological dialogue between six researchers who specialise in different sociological subfields. Each researcher explores the possible consequences of generative AI within their specific area of expertise. More concretely, the article develops insights around directions in social theory, the political economy of intellectual property, matters of identities and intimacies, evidence and evidentiary power, racial and reproductive inequalities, as well as work and social class. This is followed by a collective discussion on six interconnected themes across these areas: agency, authorship, identity, visibility, inequality, and hype. We also consider our role as cultural producers, understanding our reactions to generative AI as part of the empirical, theoretical, and methodological shifts this knowledge controversy engenders, as well as highlighting our duty as critical sociologists to keep the knowledge controversy about generative AI open.

How does repression reshape the way online activists engage with target audiences? While prior research has primarily examined changes in overall online participation, it has paid less attention to how activists adjust their strategies in response to repression. Addressing this gap, this article argues that repression incentivizes online activists to broaden their support base by promoting inter-group engagement and signaling inclusivity. Focusing on the 2011 Occupy Wall Street movement, the study analyzes Twitter interactions using network measures of assortativity and cross-group tie proportions. It applies permutation tests and ARIMA-based Interrupted Time Series (ITS) analysis to compare network patterns across key phases, delineated by the Brooklyn Bridge mass arrests on October 1 and the eviction threat of Zuccotti Park on October 13. The results show that repression triggers a significant decrease in assortativity, indicating increased inter-group engagement, while cross-group tie proportions remain stable, suggesting structural rather than isolated behavioral changes.

This study introduces a machine learning-based framework aimed at identifying and interpreting the most influential factors contributing to divorce. Utilizing data from the 2021 Turkey Family Structure Survey, we apply Random Forest and Logistic Regression models to rank predictors based on their relative impact on marital dissolution. The goal is to uncover which socio-legal, temporal, and behavioral variables most significantly contribute to the divorce outcome within a culturally grounded dataset. Both models converge on a set of dominant features—psychological conflict responses, cultural marital rituals, and political disagreements—demonstrating their robust influence across different algorithmic paradigms. Feature importance scores derived from model outputs and explainability tools (e.g., permutation and coefficient-based rankings) reveal consistent patterns and offer interpretable insights aligned with sociological theory. This approach contributes to computational sociology by showcasing how machine learning can be used not only for prediction, but more importantly, for identifying statistical patterns that reflect social structures and behavioral dynamics associated with divorce outcomes.

This study investigates individuals’ lay definitions—naïve mental representations—of artificial intelligence (AI). Two national surveys in the United States explored lay definitions of AI in the workplace (Study 1) and in everyday life (Study 2) using both open- and closed-ended questions. Open-ended responses were analyzed with natural language processing, and quantitative survey data identified factors associated with these definitions. Results show that conceptions of AI differed by context: workers emphasized efficiency and automation in the workplace, while the general public linked AI to diverse everyday technologies. Across both groups, conceptions remained nuanced yet limited. Sociodemographic factors and personality traits were related to sentiments expressed in definitions, and greater trust in AI predicted more positive sentiments. These findings underscore the need for targeted training and education to foster a more comprehensive public understanding of what AI is and what it can do across different contexts.

The spread of modern digital technologies, such as social media online platforms, digital marketplaces, smartphones, and wearables, is increasingly shifting social, political, economic, cultural, and physiological processes into the digital space. Social actors using these technologies (directly and indirectly) leave a multitude of digital traces in many areas of life that sum up an enormous amount of data about human behavior and attitudes. This new data type, which we refer to as “digital behavioral data” (DBD), encompasses digital observations of human and algorithmic behavior, which are, amongst others, recorded by online platforms (e.g., Google, Facebook, or the World Wide Web) or sensors (e.g., smartphones, RFID sensors, satellites, or street view cameras). However, studying these social phenomena requires data that meets specific quality standards. While data quality frameworks—such as the Total Survey Error framework—have a long-standing tradition survey research, the scientific use of DBD introduces several entirely new challenges related to data quality. For example, most DBD are not generated for research purposes but are a side product of our daily activities. Hence, the data generation process is not based on elaborate research designs, which in turn may have profound implications for the validity of the conclusions drawn from the analysis of DBD. Furthermore, many forms of DBD lack well-established data models, measurement (error) theories, quality standards, and evaluation criteria. Therefore, this special issue addresses (i) the conceptualization of DBD quality, methodological innovations for its (ii) assessment, and (iii) improvement as well as their sophisticated empirical application.