Social Science Computer Review

The use of algorithmic hiring is on the rise, becoming more and more common due to its capacity to process a large number of applications with efficiency. Despite the advantages of this approach, there have been numerous cases of discriminatory hiring outcomes. A significant contributing factor to such outcomes is the lack of representativeness in the data used to develop these systems. This results in a significant decline in performance for underrepresented groups, disproportionately impacting marginalized communities. Addressing bias in algorithmic hiring requires access to comprehensive datasets that include curriculum vitae (CV) and demographic information reflecting diverse backgrounds. Unfortunately, there is a lack of datasets that serve such a purpose. This paper introduces a data donation campaign designed to collect real-world CVs, including demographic and sensitive information, from a representative sample of individuals in the EU workforce. The paper discusses the design decisions underpinning the campaign, along with the challenges encountered during its deployment and execution. Finally, it offers lessons learned and practical solutions to overcome these challenges, thereby contributing valuable insights for future efforts in this domain.

The Supreme Court’s Dobbs v. Jackson Women’s Health Organization decision in June 2022 reversed 50 years of precedent by allowing states to formulate their own abortion policies. This resetting of abortion policy created a new raft of opportunities and threats across the states for both pro-life and pro-choice supporters. In this study, we aim to analyze how public discourse around abortion responded to this changed political context. Using a dataset of 288,325 abortion-related Tweets posted in 2022, we examine public reaction to Dobbs using both quantitative and qualitative approaches. We categorize Tweets by abortion stance (pro-choice and pro-life) and geo-political context by state groups (protected, restricted, and unsettled based on abortion access policy). Our temporal analysis shows that while both pro-choice and pro-life Twitter activity spiked after both the leaked draft in May 2022 and the final decision, only pro-choice discussions maintained a heightened level of engagement over time. Analyzing the discussion frames among the Tweets reveals that pro-choice users emphasized a wider range of arguments that varied by state context, while pro-life Tweets were generally unresponsive to state context. Our findings indicate that the new threats and opportunities had uneven effects within pro-life and pro-choice public discourse.

This essay conceives artificial intelligence as a chapter in the history of writing through reconsidering the eighteenth-century automaton writer created by Jaquet Droz, a Swiss clock making workshop. As much an innovation in writing technology as it was an early example of artificial intelligence, the Jaquet Droz automaton writer reveals how artificial intelligence is a historical idea and material artifact deeply entangled with the history of writing, an embodied as well as deeply emotional form of cognitive activity and one of the oldest human technologies.

Digital platforms now serve as crucial archives for analysing societal trends, yet their research APIs pose methodological challenges. This study critically evaluates TikTok Research API through comparative analysis of 6,373 videos from 14 creators in the United States and United Kingdom (2020–2022), contrasting API-derived outputs with manual collection and third-party analytics. The API demonstrated scalability, retrieving more videos than alternative methods and providing 22 variables, including eight unavailable elsewhere. However, limitations were substantial: transcriptions covered about 10% of the content, with more transcripts returned from American male creators. Engagement metrics exhibited inconsistent accuracy across data sources, with the API showing systematically lower view counts but higher comment and share counts compared to manual collection. The number of videos varied depending on sample composition, indicating that small changes in inclusion criteria could shift outcomes disproportionately. These results highlight systematic inconsistencies, showing why multi-method approaches remain necessary despite automation. While TikTok Research API offers valuable scale and ethical compliance, its demographic biases and metadata inconsistencies compromise validity. The study advocates integrated auditing protocols and targeted API refinements to improve representativeness and accuracy in platform-based research.

This study examines how elite figures shape polarisation on Twitter/X through the interplay of content, structure, and engagement strategy. Drawing on data from nine globally influential users (2010–2021), the research integrates natural language processing, network analysis, and causal modelling to test five hypotheses grounded in social identity, agenda-setting, and two-step flow theories. Entity co-occurrence networks reveal that polarised discourse forms denser, more clustered networks than non-polarised content, indicating tighter semantic cohesion around socially and politically charged entities. Thematic and sentiment analyses show that posts addressing non-core topics – particularly those concerning social justice, environmental sustainability, philanthropy, and global welfare – are nearly five times more likely to be polarised than core professional themes. Negative emotional tone further amplifies this effect, while higher tweet-to-retweet ratios reduce polarisation, underscoring the moderating role of original content production. A user-level mediation analysis tested whether topical diversity transmits the effect of follower scale on polarisation but found no significant indirect pathway, suggesting that larger audiences do not necessarily foster communicative moderation. The findings advance understanding of elite discourse by linking structural and thematic polarisation to behavioural mechanisms of engagement. Theoretically, the study bridges social identity, agenda-setting, and two-step flow frameworks to explain how elites balance audience alignment and expressive risk. Practically, it highlights how emphasising original content, inclusive framing, and professional identity consistency can mitigate divisive online dynamics and foster more cohesive digital publics. To support transparency and reproducibility, the dataset and analytical code are made publicly available.

This study examines how the #MeToo movement reshaped long-term public discourse on sexual violence in South Korea. While research has documented the surge of attention following prosecutor Seo Ji-hyun’s 2018 disclosure, little is known about how the movement reorganized the broader discursive field beyond the initial moment of crisis. Drawing on discursive institutionalism, we argue that #MeToo acted as a discursive shock that consolidated previously fragmented conversations into more coherent and durable interpretive frameworks. Using an 8-year corpus of 351,582 Korean-language tweets (2015–2023), we apply Dirichlet Multinomial Regression topic modeling and time-series analysis to identify shifts in both topical content and discursive structure. Our findings show a transition from episodic, emotion-driven narratives to stable, thematic frames emphasizing human rights, systemic discrimination, and institutional accountability. We further demonstrate increasing convergence between social media and mainstream news, indicating diffusion and stabilization of these new frames. The results suggest that digital activism can produce enduring transformations in public meaning, illuminating how social movements institutionalize discourse over time.

Bias in word embeddings is often measured using bipolar dimensions, constructed as the difference between two anchor centroids. This technique assumes both poles are symmetrical and equally informative. However, normativity literature shows that one category may function as the unmarked norm, with others framed as marked deviations. In race, whiteness typically holds the normative position, and embedding-based race dimensions may inherit the skew. We test this possibility using dimensions constructed from validated African–European name anchors, probed with neutral and valence words. In three embedding models (Wiki-News, South African news, Google News), we assess whether race dimensions favour whiteness as a normative anchor, whether this skew is stronger in culturally specific models (SA, Google), and whether bipolar offsets amplify one pole, given unipolar evidence. Results show that neutral and valence terms cluster nearer to the white pole (most strongly in the Wiki-News model), indicating whiteness as the semantic default. Overshoot favoured Black in Google and Wiki-News, while White overshoot only occurred in the South African model. We argue that this captures racialised variance where the pole with more spread tends to exert greater leverage on the bipolar axis. The study provides quantitative evidence of white-normative anchoring and diagnostics for asymmetric amplification in embedding-based bias measures.

Drawing on information ecology theory, this study investigates how online toxicity spreads and drives emotional group polarization in Reddit discussions surrounding the Israel–Palestine conflict. Based on a dataset comprising 8,725 posts and 1,628,366 comments, we employ Google’s Perspective API to detect toxic content, BERTopic to extract discussion topics, and a dictionary-based method to measure affective features. The results show that, from the information content perspective, content related to conflict and political topics is more prone to generating toxicity. From the information user perspective, post toxicity reduces the scale of user interaction while simultaneously increasing the level of comment toxicity. Furthermore, the analysis shows that post toxicity provokes comment toxicity by triggering users’ negative or high-arousal emotions, with discrete negative emotions such as anger, sadness, and fear enhancing the positive impact of post toxicity on comment toxicity. From the information environment perspective, the results indicate that post toxicity is not significantly associated with emotional group polarization. Accordingly, post toxicity alone plays a very limited role in explaining emotional group polarization in online discussions. These findings advance our understanding of toxicity dynamics in online environments and offer evidence-based strategies for moderators, platform designers, and policymakers to mitigate harmful discourse and foster healthier online communities.

This essay argues that the black box—both as cryptic device and as critique of illegibility—is not unique to modern technology and has deep roots in the medieval Mediterranean world. Technical opacity was frequently addressed in Latin and Arabic sources, often with a critical undertone. Then as now, techno-skeptical writers saw the self-acting device as treacherous, due to its reliance on hidden labor and mechanisms. This critique arose especially in relation to unfamiliar or foreign devices, like animated idols; as such, it was often racializing, attributing opacity as well as deceit to the object and its makers. Modern critiques of technology that focus on invisible labor may reproduce similar biases by enforcing a privileged, first-world perspective. A transhistorical approach thus not only shows the enduring history of the black box; it also illuminates the religious genealogy of techno-skepticism, as well as the biases that inhere in the black box, especially when deployed as a critical discourse.

Objective: This paper analyzes gender equality across countries in the year 2024 by using the GGGI, with the intention of disentangling the unseen structural and non-deterministic patterns. Instead of repeating the process of calculating the index, it is openly recognizing the compositional feature of the GGGI and the unseen similarities between the indices. Methods: This research employs a global cross-sectional study of 146 countries over the four primary GGGI sectors: economic participation, education, health and survival, and empowerment. Where OLS is only employed as a diagnostic test, as its almost perfect fit (R²∼1) is squarely mechanical and lacks relevance for inference. Apart from ensemble models employed for predictions, K-means clustering, SHAP analysis, and GridSearchCV optimization are also used. Findings: The out-of-sample predictions demonstrate high levels of predictive accuracy, with Gradient Boosting models yielding an R² of approximately 0.90 and an RMSE of approximately 0.045, indicating that there is significant nonlinear information beyond index aggregation. Unsupervised clustering techniques show that there are seven distinct country clusters that go beyond traditional geographic and income divisions, which can be identified with more than 93% accuracy. The SHAP results show that empowerment and economic participation are drivers, while there is insignificant variation in healthcare. Contribution: This study identifies the boundaries of regression analysis in index research, as well as the advantages of machine learning analysis in determining structural patterns related to gender equity.

As Generative AI (GenAI) becomes increasingly integrated into everyday information environments, many individuals remain uncertain or reluctant to engage with these technologies despite their growing potentials. By extending the Cognitive Mediation Model (CMM), the study emphasizes the cognitive and emotional processes that underpin public engagement with emerging digital information technologies. A survey of 527 Chinese participants was conducted using an online questionnaire. Structural equation modeling was employed to test the extended CMM, examining the relationships between exposure to traditional and social media, elaboration, interpersonal communication, knowledge acquisition, anxiety, and GenAI usage intention. Results indicate that exposure to both traditional and online social media were positively associated with elaboration and interpersonal communication, which were in turn negatively associated with anxiety and positively associated with perceived knowledge. However, elaboration does not significantly relate to factual knowledge, while interpersonal communication was negatively associated with factual knowledge. Anxiety was negatively associated with GenAI usage intention, whereas factual and perceived knowledge were positively associated with adoption. Theoretical and practical contributions are discussed as well.

Digital trace “data donation” studies offer researchers a unique opportunity to collect high-quality behavioral data, but decisions about the scope of requested data can impact both dataset richness and participant compliance. This paper examines the tradeoffs between requesting larger data packages, which include more extensive historical records, and participants’ willingness to donate. In a randomized experiment with Facebook and Instagram data donations, we compare a control condition where participants are asked to request the default 1-year data period to a treatment condition in which they are asked to request data for their entire account history. We analyze how different request sizes affect (1) participants’ compliance rates and (2) the characteristics of the data resulting from these different requests. We find that participants asked to request more data are less likely to complete the task. However, we propose that this is not primarily due to heightened privacy concerns, but rather because these data packages are significantly larger and therefore take longer for the platforms to deliver. This additional time to deliver data packages results in increased attrition. In terms of the effects on the data itself, we show that decisions about the time-span of the data impacts not only the volume of data requested, but also has implications for measurement validity, as the temporal window fundamentally redefines what key constructs represent, potentially transforming intended static indicators into narrow snapshots of recent behavior. We provide guidance for researchers navigating these decisions, considering both the benefits of richer longitudinal data and the risks of reduced participation.

Social evaluation is fundamental to everyday interactions, yet our understanding has been constrained by fragmented theories and the lack of a scalable method for tracking group attitudes in real time. This paper resolves this methodological gap by introducing and validating a computational framework that empirically synthesizes three major theoretical models (Stereotype Content Model, Dual Perspective Model, and Semantic Differential) within a unified word embedding space. We demonstrate that social evaluation is structured by two core latent dimensions: Warmth-Communion-Evaluation (WCE), capturing affective and moral judgments, and Competence-Agency (CA), reflecting perceptions of ability and effectiveness. To validate its real-world utility, we apply this framework to U.S.-based Twitter posts about Chinese and Japanese individuals before and during the COVID-19 pandemic. Our analysis reveals that while perceptions of competence (CA) remained stable, affective evaluations (WCE) of Chinese individuals declined sharply, a dynamic not observed for Japanese individuals. This work offers a robust, scalable instrument for tracking intergroup attitudes during crises and provides a crucial bridge between social psychological theory and computational social science, enabling the real-time analysis of intergroup dynamics.

Topic discovery and integration are vital for maintaining vocabularies that categorize textual corpora. Automated approaches are often computationally expensive and lack domain-specific conceptual nuance; manual approaches are costly in terms of time and potential bias. To address this dilemma, we introduce the segments-as-topic (SAT) methodology, a four-stage process that combines automation and human expertise to assess candidate topics for vocabulary inclusion. In the SAT generation stage, a topic is formulated and refined through collaboration with domain experts, and then a sentence-level semantic similarity model retrieves corpus segments semantically aligned with the topic. The SAT expansion stage uses this seed set to find additional semantically similar segments, which are iteratively accepted or rejected to build a final segment set. During the review stage, a panel of scholars evaluates the topic for inclusion. In the integration stage, all segments in the final segment set are automatically tagged with the new topic. We apply this methodology to the Comparative Constitutions Project vocabulary that tracks over 330 topics in national constitutions, and demonstrate the addition of three new topics to the vocabulary. The SAT approach balances computational efficiency with expert judgment, offering a systematic, user-friendly, and replicable framework for social scientists to expand domain-specific vocabularies.

Survey research is entering a new era which centres on its linkage with other forms of digitally generated data such as social media. Many suggest that this can help to address existing weaknesses in self-report surveys such as non-response and measurement bias. However, to link a participant’s survey responses to their social media data, consent from the participant is required. Previous studies have shown that consent to linkage is typically low and selective. This paper expands on the existing literature by comparing Twitter (now X) usage and consent to survey linkage across five national contexts. Testing the effects of several sociodemographic and attitudinal predictors in the US, the UK, France, Germany and Poland, our study finds that overall consent rates vary significantly by age, political attention, privacy concern, trust in social media companies and frequency of political posting on Twitter/X. However, our results also confirm that variable effects differ significantly between nations, suggesting a moderating cultural influence. Within-country variation in the US between 2020 and 2024 is also present, indicating that effects are not necessarily fixed over time. These findings dictate the need for caution when conducting substantive comparisons across countries and time when using social media data.

The automatic classification of digital hate is a pressing challenge, yet many existing computational models remain opaque and insufficiently evaluated as measurement tools in social science contexts. This study examines the utility of Google’s Perspective API as a measurement instrument by modeling higher-order constructs of harmful discourse (i.e., incivility and intolerance) as outcomes of multiple lower-level behavioral indicators captured by distinct API scores rather than using a single aggregate score as a proxy for complex social behaviors while enabling the evaluation of state-of-the-art black-box classifiers beyond classification metrics. Drawing on 4,000 manually annotated English-language YouTube comments in the context of the Israel-Hamas war, we test whether multiple API scores predict incivility and intolerance using generalized linear and additive models, assess classification performance across non-hateful, uncivil, and intolerant content, and benchmark a recent deep learning model. Results show that Identity Attack is a strong predictor of intolerance, whereas Insult and Profanity are indicative of incivility. While classification performance is somewhat below state-of-the-art deep learning models, our approach offers important advantages: transparency, interpretability, accessibility for non-technical researchers, and potential cross-linguistic applicability. We argue that typology-driven, multi-indicator-based classification provides a practical and theoretically grounded complement to more aggregated black-box models, particularly in human-in-the-loop workflows that can help reduce annotator exposure through pre-filtering of content.

Drawing on rational choice and social learning theories, this study examines how social networks, subcultural immersion, and specific activities shape dark web actors’ perceptions of risks and rewards. Using original 2024 survey data from 123 active dark web users recruited across English- and Russian-language forums and Telegram channels, we analyze how peer involvement, sustained participation in deviant subcultures, and interest in particular illicit activities influence perceived risks of detection and perceived benefits. Findings suggest that larger peer networks and deeper subcultural immersion may lower perceived risks and heighten perceived rewards. We also find important activity differences: interest in malware correlates with lower perceived risks, ransomware with higher perceived risks, and selling illicit goods with particularly high perceived rewards. These results underscore the fluid, socially constructed nature of risk and reward perceptions in clandestine spaces, offering rare empirical insight into the subjective mechanisms underlying cybercriminal choices and informing targeted prevention strategies.

Emerging research in computational social science has applied LLMs to automate content analysis, often by prompting a single model to act as a human coder. While a single LLM may suffice for a few manifest variables, it still falls short on diverse latent constructs. And the impact of LLM agent attributes on measurement outcomes remains unclear, limiting their validity for communication research. Drawing upon the literature on interacting agents and communication, this study examines the impact of agent diversity, agent open-mindedness, and human–AI collaboration (HAIC) in a multi-LLM-agent system for automated content analysis. The results demonstrate reliable and accurate measurement of four communication variables across three datasets, with improved performance following agent discussion. Additionally, agent open-mindedness, but not agent diversity, significantly affects measurement outcomes. These results highlight the potential of multi-LLM-agent systems for automated content analysis and suggest the importance of considering agent attributes and values in system design.

In June 2024, youth-led protests in Kenya against a controversial Finance Bill demonstrated the connection between digital technologies and political activism in the Global South. This study examines how generative artificial intelligence (GAI) shapes political participation by focusing on Kenyan Gen Z activists who used ChatGPT to create custom models: Finance_Bill_GPT, Corrupt_Politicians_GPT, and MPs_Contribution_GPT (collectively called Protest_GPT_KE). These tools simplified complex laws, exposed corruption, and mobilized young people online, allowing them to bypass traditional sources such as media and elites. However, using GAI for activism raises ethical and political concerns, including surveillance, data rights, and state repression. The study surveyed 374 Kenyan Gen Z participants, primarily in Nairobi, and used Partial Least Squares Structural Equation Modeling (PLS-SEM) to analyze the connections among AI use, tool appropriation, and political participation. Results show that ChatGPT use alone did not directly increase offline activism; its effect appeared when combined with Protest_GPT_KE and online participation. This study is one of the first to document how youth in the Global South are creatively using GAI for grassroots mobilization, demonstrating that GAI’s political influence depends on user innovation and context.