Global challenges of the recent decade have affected the daily life of many emerging adults (age group: 18–29 years) who are vulnerable for poor mental health. Many of them engage in intensive social media use (SMU) to escape negative offline experiences. This increases their risk for addictive tendencies. The present study investigated whether addictive SMU could contribute to a decline of mental health in this age group. Moreover, we compared emerging adults attending university and employed emerging adults matched by age. Data of 2052 SM users (respectively 1026 students and non-students) from Germany were collected via an online survey. Students reported significantly more daily stressors, higher addictive SMU, loneliness, depressive and anxiety symptoms, and significantly lower perceived offline social support and positive mental health (PMH) than non-students (small effects: .16 ≤ d ≤ .29). In both groups, daily stressors statistically positively predicted addictive SMU in a regression analysis. In the student group, offline social support statistically negatively predicted addictive SMU. In both groups, loneliness mediated the positive association between addictive SMU and depressive/anxiety symptoms, as well as the negative association between addictive SMU and PMH. The present findings reveal that a high level of daily stressors could contribute to addictive SMU that could result in loneliness and a decline in mental health. Students who lack offline social support are at particular risk for these negative consequences. Potential ways to protect this group from the negative effects of daily stressors and online activity are discussed.
Self-branding on professional social media is increasingly common, yet its adoption and benefits remain uneven. This study examines how users construct personal-professional hybrid identities through both posting content and linguistic style, and how these strategies relate to social capital building. Using 9918 LinkedIn profiles and 86,160 associated posts, the study identifies dimensions of personal and professional self-presentation in content and linguistic style through automated text mining, including large language models. The findings show that content and linguistic style hybridity are asymmetric rather than interchangeable. Sharing professional content in a personal style is more conducive to bridging social capital by helping users reach broader networks, whereas combining personal content with a personal style is more helpful for bonding social capital by fostering closer ties. These patterns also vary by users’ online and offline status, including gender. The study advances theory and offers practical insight into visibility, trust, and connection online.
Sustaining participant engagement while maintaining data quality is a central challenge for games with a purpose (GWAPs). In this article we examine whether player orientation (according to Bartle’s taxonomy) shapes contribution volume under competitive and hybrid (intra-team cooperation + light inter-team rivalry) GWAP mechanics. Although the design provisioned a purely cooperative arm, too few cooperative-only sessions materialized for session-level analysis, so the empirical comparison is Competitive versus Hybrid. To test this, we ran a one-week quasi-experimental field trial of JellyfishGO!, an Augmented Reality (AR) GWAP, with 64 participants in 32 two-person teams. Players completed a six-item Player-Type Decision (PTD) at profiling and the 24-item Hexad post-play; high-resolution logs captured captures, missions, badges, team matches and leaderboard dynamics. Across model-free contrasts, robust Ordinary Least Squares (OLS) and multilevel count models (with exposure offsets), Hybrid configurations produced the largest and most robust throughput gains: Hybrid sessions were longer and showed higher captures/minute, with Incidence-Rate Ratio (IRR) estimates for Hybrid vs Competitive in a compact window ≈1.47-1.91 (robust to trimming, negative-binomial re-specification and jackknife). Modeled Arm × Bartle interactions were small and imprecise, giving limited evidence that PTD labels systematically moderate the Hybrid boost; Killers did exhibit pronounced short bursts in Hybrid contexts. Session-level consensus quality was effectively invariant, limiting fidelity inference. We conclude that hybrid team structures are a promising route to raise contribution volume (throughput) in mixed populations—but must be paired with layered quality controls (redundancy, rolling consensus, fraud detection) and larger, preregistered future replications to clarify Mechanic × Orientation × Task interactions.
The evaluation of explainable AI (XAI) approaches often relies on three families of methods: subjective measures (e.g., questionnaires on trust or satisfaction), objective measures (e.g., task performance metrics), and mathematical metrics (e.g., for faithfulness). Yet, it remains unclear how these families align or diverge in practice. In a preregistered between-subjects study (N =166), we use three established saliency map techniques (LIME, Grad-CAM, Guided Backpropagation) as a testbed to examine this issue. We find that each family of methods leads to different conclusions: participants reported no differences in trust or satisfaction, Grad-CAM improved user performance, while mathematical metrics favored Guided Backpropagation. At the same time, mathematical metrics were only partially related to user performance, and these relationships were sometimes counterintuitive. Our findings highlight the methodological importance of comparing subjective, objective, and mathematical approaches when evaluating XAI, illustrating both tensions and aspects that are aligned. We discuss implications for XAI evaluation frameworks.
