The study, conducted by Yuna Lee of Kyungwoon University and Sang-Soo Lee of Pusan National University, drew on two related but distinct constructs. Digital stress refers to the pressures that arise specifically from digital environments, including the sense of being constantly available, the anxiety of missing what peers are doing, and the strain of managing social feedback online. Social media fatigue, by contrast, describes the cognitive, behavioral, and emotional exhaustion that accumulates from sustained platform use, a state that prior research has linked to compulsive checking, fear of missing out, anxiety, and depressed mood. Rather than averaging these experiences into one composite score, the researchers treated them as a multidimensional configuration, asking whether certain combinations of stress and fatigue form recognizable patterns across the adolescent population.

Methodologically, the study was unusually careful about measurement. Before comparing groups, the authors confirmed measurement invariance, a statistical property ensuring that the survey instruments measure the same constructs in the same way across different subgroups of students. Without this step, apparent differences between elementary, middle, and high school students could reflect artifacts of how the questions function rather than genuine developmental change. The team then applied latent profile analysis, a person-centered mixture modeling approach that identifies unobserved subgroups within a population. Unlike variable-centered methods, which estimate how variables relate to each other on average across everyone, person-centered approaches follow a tradition in developmental psychopathology that treats the individual pattern as the meaningful unit of analysis. The researchers used established fit criteria and simulation-based guidance to decide how many profiles best described the data.

The headline finding is that a six-profile solution emerged for the full sample, and that these profiles differed not only in severity, meaning how much digital burden students carried overall, but also in shape, meaning the relative balance among the different burden indicators. The most consequential dimension differentiating the profiles was emotional fatigue. In other words, two students could report similar overall levels of digital strain while occupying meaningfully different risk configurations, one dominated by exhaustion and emotional depletion and another by different combinations of stress and fatigue. This shape-versus-level distinction matters because a single composite score would flatten exactly the differences that may be most clinically relevant, obscuring subgroups who look average on paper but are struggling in specific, identifiable ways.

The study also examined self-compassion, the capacity to treat oneself with kindness during failure or distress, as an external covariate using the R3STEP three-step approach, a method for relating auxiliary variables to latent class membership without distorting the underlying profile structure. Self-compassion has a substantial evidence base in adolescent mental health, with meta-analytic findings linking it to lower psychological distress, and prior work showing it predicts emotional well-being in both early and older adolescents. In this study, after adjusting for gender, higher self-compassion was associated with lower relative odds of belonging to any of the non-low-burden profiles compared with the low-burden profile. The odds ratios ranged from .215 to .546, meaning that adolescents higher in self-compassion were substantially less likely to fall into every profile characterized by elevated digital stress or social media fatigue.

That pattern held across all five elevated profiles, which is notable because it suggests self-compassion is not merely buffering one particular type of digital strain. Instead, it appears to function as a broad protective characteristic associated with membership in the low-burden group regardless of the specific shape that burden takes. The authors are careful about interpretation, as is standard for cross-sectional data: the analysis cannot establish whether self-compassion prevents digital burden, whether heavy digital burden erodes self-compassion, or whether both are shaped by other factors. But the consistency of the association across every elevated profile strengthens the case that self-compassion is a meaningful correlate worth targeting in future longitudinal and intervention research.

Perhaps the most striking result is the developmental heterogeneity the researchers uncovered when they repeated the analysis separately by school level. Among elementary school students, five profiles emerged. Among middle school students, the number rose to seven, the most complex configuration of any school level. Among high school students, the solution settled back to five. This non-linear pattern implies that the transition into middle school, a period coinciding with early adolescence and profound neurological, social, and emotional change, is also when the structure of digital burden becomes most differentiated. Adolescence is widely recognized as a window of heightened vulnerability and opportunity in brain development, and the social media context is known to transform peer relations in ways that amplify visibility, comparison, and social feedback during exactly these years.

The practical implications cut against the grain of popular screen-time debates. If adolescent digital burden were adequately captured by hours of use or a single summed score, then one-size-fits-all guidance might suffice. Instead, the finding that profile structure itself changes across school levels suggests that screening tools, school counseling programs, and digital wellness curricula need to be developmentally calibrated. A middle school intervention built around the profile landscape of elementary students, or a high school program modeled on middle school patterns, may miss the configurations that actually characterize the students in front of them. The authors argue that their findings support assessment that goes beyond digital exposure or a single composite score, a recommendation with direct consequences for how schools and researchers operationalize digital well-being.

The study also contributes to a rapidly maturing research literature on digital stress as a construct in its own right. Recent years have seen the development and psychometric validation of multidimensional digital stress scales and social media fatigue measures grounded in cognitive, behavioral, and emotional frameworks, along with longitudinal work linking adolescent digital stress to subsequent depressive symptoms. Researchers in this area have repeatedly cautioned that the relationship between screens and adolescent mental health is more nuanced than early public fears suggested, and that addictive use patterns, motives for use, and content type matter alongside raw time online. The present study extends this nuance to the population level, showing that the internal architecture of digital burden, not just its quantity, varies systematically across a very large and diverse student sample.

Some limitations deserve mention. The data come from a secondary analysis of a de-identified, institutionally administered school survey in a single Korean province, and the dataset is not publicly available, held instead by the data-holding educational institute under institutional authorization. The cross-sectional design precludes causal claims, and the authors received no external funding for the work while declaring no competing interests. Still, the scale of the sample, the rigor of the invariance testing, and the use of modern mixture modeling methods give the findings considerable weight. For parents, educators, and clinicians anxious about the mental toll of adolescent connectivity, the message is both sobering and actionable: digital stress is real, multidimensional, and developmentally shifting, and the adolescents most immersed in it may be distinguishable, and potentially reachable, through the psychological resources they bring to their own inner lives, with self-compassion emerging as one of the most consistent protective signals in the entire analysis.

Subject of Research: Latent profiles of digital stress and social media fatigue among adolescents and their association with self-compassion

Article Title: Multidimensional Latent Profiles of Digital Stress and Social Media Fatigue Among Adolescents: Associations with Self-Compassion and Heterogeneity Across School Levels

Article References: Lee, Y., & Lee, S.-S. (2026). Multidimensional Latent Profiles of Digital Stress and Social Media Fatigue Among Adolescents: Associations with Self-Compassion and Heterogeneity Across School Levels. International Journal of Mental Health and Addiction. https://doi.org/10.1007/s11469-026-01732-1

Image Credits: AI Generated

DOI: 10.1007/s11469-026-01732-1

Keywords: digital stress, social media fatigue, adolescents, self-compassion, latent profile analysis, school-level heterogeneity, mental health, cyberpsychology, screen time, South Korea, emotional fatigue, measurement invariance

The phenomenon at the center of the study is known as information systems discontinuance, the point at which individuals or organizations stop using a technology they have already adopted. While decades of research have explored why people embrace new systems, far less attention has been paid to the reverse journey. This gap matters commercially as well as scientifically: industry benchmark reports cited by the authors show that mobile app retention rates drop steeply within days of installation, and organizations routinely struggle with costly decisions about decommissioning legacy platforms. Understanding the mechanics of disengagement, the authors contend, is essential for anyone who designs, deploys, or regulates digital technology.

To build their synthesis, Ghelani and Pandey applied the Gioia methodology, a well-established approach to qualitative research that imposes systematic rigor on literature synthesis. The method requires researchers to move from open-coded first-order concepts drawn directly from the source studies, to broader second-order themes, and finally to aggregate dimensions that form a coherent conceptual structure. The 97 studies they reviewed spanned social media platforms such as Facebook, WeChat, Twitter, and Weibo; consumer applications including mobile banking, gamification fitness apps, and short video services; and organizational systems ranging from cloud computing deployments to legacy enterprise software and even mental health chatbots.

The organizing lens for the synthesis is Leavitt’s socio-technical systems framework, a model dating to 1965 that views any organized system as the joint product of four interacting elements: People, Technology, Task, and Structure. The framework has deep roots in information systems research, tracing back to foundational work in the 1970s that explained many system failures as mismatches between social and technical components. By mapping discontinuance research onto these four dimensions, the review reveals patterns that single-context studies could never show on their own.

Within the People dimension, the review identifies a dense cluster of psychological drivers. Overload is among the most heavily documented: information overload, communication overload, and social overload on social networking sites consistently push users toward exhaustion and eventual withdrawal. Technostress, social media fatigue, privacy concerns, cyberbullying, and envy all appear as strains that accumulate until users disengage. The authors also highlight the role of emotions and self-regulation, including guilt, cognitive dissonance, regret, and addiction-related coping, showing that pathological use patterns can paradoxically drive both overuse and the eventual decision to quit. Inhibitors on this dimension include habit, inertia, and satisfaction, which anchor users to incumbent systems even when alternatives beckon.

The Technology dimension captures how the characteristics of systems themselves shape abandonment. Poor system quality, technical failures, and unmet expectations fuel negative disconfirmation, a mismatch between what users anticipated and what they received. Studies of chatbots, wearable self-tracking devices, and mobile shopping apps document how response failures and technostress translate directly into discontinuance intentions. Yet technology can also inhibit departure: switching costs, architectural embeddedness, and the sheer weight of legacy systems keep users locked in. Research on organizational legacy system replacement shows how accumulated technical debt traps firms in outdated platforms, making discontinuance a socio-technical negotiation rather than a simple upgrade choice.

The Task dimension reflects what users are actually trying to accomplish. When a system no longer fits the task, when it delivers irrelevant information, or when competing technologies serve the same purpose better, discontinuance pressure rises. The review documents cases where users intensified their use of a failing system before finally abandoning it, a counterintuitive pattern suggesting that people often test a system’s limits before giving up on it. Task-related drivers interact closely with the fourth dimension, Structure, which encompasses organizational norms, governance arrangements, social networks, and institutional pressures. Critical mass dynamics, for example, determine whether a social platform survives: when enough connections leave, remaining users follow in cascades of network collapse.

Perhaps the review’s most significant conceptual contribution is its treatment of discontinuance as a process with distinct forms. The authors distinguish intermittent discontinuance, in which users take temporary breaks and may return; replacement, in which one system is swapped for another; and permanent discontinuance, in which use ends for good. Which form materializes depends on how the four socio-technical dimensions interact. A dissatisfied user with strong habits and high switching costs may opt for an intermittent pause rather than a permanent exit, while an organization facing obsolescence pressures and institutional mandates may pursue full replacement. Framing these outcomes as points on a continuum of socio-technical interdependence, rather than isolated choices, is the study’s central theoretical advance.

The practical implications reach system designers, organizational leaders, and policymakers alike. For designers, the findings suggest that retention depends not only on polished features but on managing user load, protecting privacy, and aligning the system with genuine task needs. For organizations, the review underscores that replacing or retiring systems requires attention to legacy habits, governance structures, and workforce adaptation, not merely technical migration. For policymakers navigating digital transitions, including the shift away from long-standing public infrastructure, the message is that discontinuance decisions ripple across all four socio-technical dimensions simultaneously. As individuals and institutions continuously reassess their relationships with technology, this synthesis offers a shared map of the terrain, showing that walking away from a system is never just a click of a delete button but the end point of a complex, multidimensional negotiation between people and the socio-technical worlds they inhabit.

Subject of Research: A qualitative review of information systems discontinuance drivers and inhibitors across four socio-technical dimensions

Article Title: A Multi-Dimensional View of IS Discontinuance: Insights Using Socio-Technical Systems

Article References: A Multi-Dimensional View of IS Discontinuance: Insights Using Socio-Technical Systems. (n.d.). https://doi.org/10.1007/s10796-026-10808-1

Image Credits: AI Generated

DOI: 10.1007/s10796-026-10808-1

Keywords: information systems discontinuance, socio-technical systems, post-adoption behavior, social media fatigue, technostress, Gioia methodology, Leavitt framework, system replacement, user abandonment, digital detox, legacy systems, qualitative literature review