The digital landscape has become an integral component of modern conflict, significantly shaping how individuals and nations navigate and respond to crises [1]. During wartime, civilians increasingly rely on digital channels, including news websites, social media, messaging applications, and institutional platforms, for information, social connection, and emotional regulation; this engagement often involves exposure to emotionally saturated content with implications for digital well-being [2-4]. Real-time social media updates have increasingly displaced traditional broadcast media as the main vehicle for collective exposure to traumatic events [5,6]. The October 7, 2023, attack on Israel exemplified this shift, when platforms such as Telegram and WhatsApp became dominant channels through which civilians encountered graphic wartime content, often involuntarily [5,7,8].

Wartime digital media engagement is heterogeneous, ranging from passive consumption (eg, continuous news monitoring and “doomscrolling”) to active engagement involving information dissemination and seeking social support, and may shape coping resources and mental state [9-13]. Evidence from collective crises indicates that trauma-related media exposure is associated with acute stress and functional impairment [14], with self-reinforcing cycles in which greater concerns lead to more media coverage and higher distress [15-17]. During COVID-19, higher exposure was linked to higher depressive (Patient Health Questionnaire-9 [PHQ-9]) and anxiety (Generalized Anxiety Disorder-7 Scale [GAD-7]) symptom scores [18], and during the Russia–Ukraine war, perceived negative media impact was linked to increased anxiety and reduced resilience, even among uninvolved populations [19,20]. The active vs passive distinction is central to digital well-being research [21-23], although operationalizations vary widely and associations depend on measurement and contextual factors, including platform type [22-25].

Israel has a long-standing research tradition on media-based exposure to terrorism and related mental health consequences [26]. Following the October 7, 2023, attack, a nationwide prospective study documented marked increases in posttraumatic stress disorder (PTSD), generalized anxiety, and depression among Israeli adults [27], and extensive media consumption, particularly graphic Telegram content, was identified as a significant risk factor for PTSD among indirectly exposed Israelis [8]. Psychological stress has been consistently linked to increased media engagement during crises: anxiety related to health-protective behaviors and information seeking is associated with greater media exposure during emergencies [15,16,28], uncertainty-inducing news representations may sustain cycles of worry and excessive consumption [29], and acute distress may override cognitive resources in shaping digital behavior [16,30].

Resilience, conceptualized as the maintenance or rapid recovery of mental health after stressor exposure, may buffer the psychological effects of crisis-related media exposure [31] and moderate associations between crisis-related information exposure and emotional distress [32], including effects of personality vulnerabilities during the COVID-19 pandemic [33]. In Israel, resilience-related factors such as emotion regulation and perceived social support showed stronger associations with mental health outcomes than social media use itself [34].

Protection motivation theory (PMT) provides a conceptual framework, proposing that protective responses depend on appraisals of threat severity and vulnerability and on coping appraisals such as response efficacy and self-efficacy [35,36]. PMT has been applied in crisis contexts; a meta-analysis found positive associations between PMT components and protective behaviors during COVID-19, with coping appraisal variables emerging as particularly strong correlates [36]. Extending PMT to media-related processes, trust in media channels has been shown to predict both threat and coping appraisals [37], and perceived risk and eHealth literacy have been linked to engagement with social media health content and protective behavior [38].

Despite growing evidence, important gaps remain. Most studies focus on overall exposure without differentiating engagement patterns such as active, passive, and institutional modes [22,23]; fewer examine engagement during active armed conflict [19,20]; and resilience has been examined as a moderator in only a limited number of studies, with few examining all 3 engagement patterns simultaneously [32,33]. This study addresses these gaps by empirically characterizing distinct patterns of wartime digital media engagement (active, passive, and institutional) and examining their differential associations with stress, resilience, and well-being. By incorporating institutional engagement, it provides a more granular account of how civilians navigate digital environments during ongoing armed conflict. Based on the framework outlined above, this study addressed the following research questions (RQs):

This study aimed to characterize patterns of digital media engagement during wartime and examine their associations with stress, resilience, and well-being in a cross-sectional survey. We also aimed to identify distinct engagement patterns (including active, passive, and institutional modes) and assess how these patterns relate to psychological outcomes in the context of an acute armed conflict.

We conducted a cross-sectional web-based survey to examine patterns of digital media engagement and their associations with stress, resilience, and well-being during a national crisis. Data collection took place in early January 2024, during the fourth month of the ongoing war, using research interfaces built on the Qualtrics platform. Participants were recruited through Panel4all, a professional web-based survey firm, using an online panel. The survey firm distributed invitations via a unique survey link sent to panel members. Recruitment followed a nonprobability quota sampling approach designed to approximate the distribution of Israeli adults aged 18 to 65 years by age, gender, level of religiosity, and district of residence, as specified by the survey firm. Panel members received incentives for survey completion, provided by the survey firm as part of its standard rewards program (eg, points redeemable for online shopping vouchers).

The survey introduction described the study objectives, provided the researcher’s contact information, assured anonymity, and clarified that responses would be used for research purposes. Participants provided electronic informed consent before accessing the questionnaire. Participation was voluntary, and participants could withdraw at any time without penalty or consequences.

The final sample consisted of 512 adults aged 18 to 65 years (mean age 37.54, SD 12.89 years), with 47.1% (241/512) identifying as male. A total of 64% (330/512) were employees, 8.4% (43/512) were managers, 4.5% (23/512) were furloughed, and 9% (46/512) were unemployed. Regarding the sector, 51.2% (262/512) were employed in private companies and 24.6% (126/512) in public organizations; the remaining participants worked in the military or police force. A total of 18.8% (96/512) reported cyber-related knowledge, either through a computer science degree or employment in a related field. Table 1 summarizes participants’ demographic and background characteristics.

The data were collected by an external survey company that does not provide datasets with missing responses; accordingly, the dataset used in this study did not contain missing values.

A total of 11,130 email invitations were sent (1 invitation per panel member); of which 3622 invitations were opened, 585 respondents accessed the survey link, and 512 completed the questionnaire and were included in the analyses. Respondents who accessed the link but did not complete the questionnaire (n=73) were not included (Figure 1).

Statistical analyses were performed using a multistage analytic strategy. First, descriptive statistics were calculated for all media usage and psychological variables, including means, medians, and 95% CIs. To identify patterns of media engagement among participants, a 2-step cluster analysis was conducted. We selected a 2-step cluster analysis because it is appropriate for large datasets and enables automatic determination of the optimal number of clusters. The adequacy of the 2-step cluster analysis was also assessed using the silhouette value. Group differences between the high- and low-media engagement clusters were examined using independent samples 2-tailed t tests for continuous variables and chi-square tests for categorical variables. Effect sizes for between-group differences were calculated using Cohen d, defined as the difference between group means divided by the pooled SD. Effect sizes were interpreted using conventional benchmarks (0.2=small, 0.5=medium, and 0.8=large) [48]. In the clustering analysis, media engagement was treated as a categorical variable (low vs high), as this approach provides clinically and practically meaningful profiles and allows clearer interpretation of behavioral patterns during crisis conditions. To explore the underlying structure of media usage, an exploratory factor analysis (EFA) using principal axis factoring with Promax rotation was performed. Factor retention was based on eigenvalues greater than 1, and sampling adequacy was confirmed via the Kaiser-Meyer-Olkin (KMO) measure and Bartlett test of sphericity. Reliability analyses guided the exclusion of items with low loadings. Subsequently, hierarchical stepwise regression analyses were conducted to assess the contribution of demographic variables, types of media usage, stress, and resilience to individual well-being. Finally, to examine the hypothesized mediation and moderation effects, path analysis using structural equation modeling (SEM) was performed. Model fit was evaluated using standard indices, including chi-square, comparative fit index (CFI), Tucker-Lewis Index (TLI), standardized root-mean-square residual (SRMR), and root-mean-square error of approximation (RMSEA). Indirect effects were estimated with bias-corrected CIs to assess the significance of mediated pathways.

The study protocol received full ethical approval from the Ethics Committee of Ariel University (institutional review board approval number AU-SOC-MZ-20231203). Before participation, respondents viewed an online information sheet describing the study aims and procedures, expected duration, potential risks and benefits, and the voluntary nature of participation. Participants then provided electronic informed consent before accessing the questionnaire. The survey was administered anonymously, and no directly identifying information was collected. The anonymized data were stored in a restricted-access environment and were accessible only to the research team. Participants received panel-based incentives for survey completion through the survey firm’s standard rewards program. The research team did not provide any direct financial compensation. No images or materials containing identifiable participant information are included in this manuscript or its supplementary files.

This study highlights the complex interplay between digital media engagement, psychological distress, and well-being during a national crisis. The findings show that high media engagement during wartime was associated with increased stress levels. This is consistent with the growing literature on “doomscrolling,” in which compulsive consumption of negative news has been associated with heightened anxiety [53]. Although seeking information is a natural response to uncertainty, the “vicious cycle” of constant monitoring may result in emotional exhaustion rather than relief [29]. Consistent with this broader pattern, higher exposure to COVID-19–related media information was associated with greater distress [18]. This pattern suggests that during times of crisis, media use may serve as a coping mechanism to manage uncertainty and maintain perceived control, a phenomenon also observed during the COVID-19 pandemic and in conflict-related contexts [18,19].

According to PMT, individuals adopt protective behaviors based on a threat- and coping-appraisal process [54]. However, our data suggest that during acute crises, heightened distress may be associated with increased digital engagement even when such engagement does not necessarily translate into psychological relief [15,18-20]. Viewed through PMT, stress can be interpreted as a proxy for heightened threat-related activation; our finding that stress was associated with greater engagement aligns with the general PMT expectation that perceived threat can increase motivation to act or seek coping responses. At the same time, the co-occurrence of elevated distress with both active and passive engagement patterns suggests that threat-related activation alone does not guarantee more adaptive outcomes, highlighting the potential importance of coping resources when interpreting crisis-related engagement. Accordingly, we interpret this pattern as increased engagement under distress, rather than as evidence of improved security behavior or decreased vigilance. Furthermore, prior research during population-level public health crises (eg, the COVID-19 pandemic) has examined links between anxiety and perceived efficacy of health-protective behaviors [16]. Related work has highlighted that exposure to the internet and media sources can be intertwined with health anxiety, health literacy, and information-seeking processes, including mediated pathways to downstream self-care behaviors [55]. Conceptual accounts suggest that media-induced uncertainty may sustain cycles of worry and repeated monitoring [29]. Empirical work using the Doomscrolling Scale indicates that doomscrolling is associated with psychological distress and poorer well-being, with distress mediating the relationship between doomscrolling and well-being indicators [56]. Continuous consumption of negative news, or doomscrolling, can be detrimental to mental health, especially when combined with intolerance of uncertainty and low psychological resilience [53].

The cluster analysis revealed 2 distinct engagement profiles: high and low media users. High engagement was associated with elevated stress levels and increased activity across digital platforms, particularly news consumption, social media, and online financial transactions. This pattern aligns with evidence from population-level crises showing that greater exposure to COVID-19–related media information is associated with higher psychological distress [18], and that social media use during the COVID-19 outbreak was associated with depression and secondary trauma [17]. In health-information contexts, exposure to the internet and media sources has also been linked to information-seeking and health anxiety processes [55]. This intensified reliance is consistent with the observation that the constant stream of updates can generate a sense of urgency and a feeling of permanent connectedness that has been linked to stress experiences [40]. Problematic smartphone-related experiences, such as nomophobia, have also been described among social media users [41]. Finally, qualitative work in individuals affected by social anxiety disorder suggests that unregulated digital engagement (eg, endless scrolling and information overload) may undermine subjective digital well-being [57].

Our findings further demonstrate that the type of media usage plays a critical role in associations with psychological outcomes. Active media use, characterized by interactive behaviors such as sharing posts or joining online groups, was positively associated with well-being, while passive media use (eg, browsing without interaction) showed a negative trend. These results align with previous studies suggesting that active media engagement may support coping and emotion regulation, whereas passive use may be associated with greater distress and emotional disengagement [58-60]. Specifically, active social media use has been linked to perceived online social support in some contexts [44]. Conversely, passive consumption has been associated with anxiety and negative self-image due to an increased likelihood of social comparison with peers [58,60]. More broadly, exposure to distressing or graphic crisis-related media content may amplify distress [7]. Institutional media usage (eg, accessing official or organizational channels) had a neutral to slightly positive effect, likely reflecting its role in fulfilling informational needs without causing emotional overload; however, the benefits of social networking services for information use can be attenuated by social media stress [61].

Path analysis confirmed that stress was significantly associated with all 3 types of media engagement. However, the indirect effects indicated that resilience was linked to well-being primarily through lower stress, while the specific indirect paths through media engagement types were small. These findings support the view that resilience is associated with better well-being and may buffer the effects of stress during crises, while engagement patterns may be differentially related to psychological outcomes [31,33,62]. Resilience is broadly defined as the fundamental human ability to “bounce back” or adapt successfully after exposure to a crisis or stressful situation [43]. Resilience has been implicated as a factor that may mitigate maladaptive engagement patterns, including doomscrolling, particularly under conditions of high anxiety and intolerance of uncertainty [53]. Moreover, strengthening resilience at individual and societal levels is recognized as an important strategy for coping with large-scale crises and disasters [31,63].

Importantly, cyber-related knowledge did not significantly differentiate between the 2 engagement clusters. This suggests that during acute crises, emotional states (particularly stress) may play a more prominent role than technical background variables in shaping digital media engagement patterns. Consistent with scholarship on anxiety and decision-making, intense anxiety or panic may impair deliberative decision-making and coherent preference ordering [30], and high-arousal negative emotions (eg, fear and anger) can bias risk judgments and subsequent choices [64]. A critical implication of this emotional overload is the onset of “cybersecurity fatigue,” a state of mental and emotional exhaustion that contributes to burnout and increased psychological strain, potentially leading users to consciously or unconsciously neglect essential security protocols [65,66]. Similar findings have been reported in the cybersecurity behavior literature, where perceived vulnerability and emotional responses influenced behavior more strongly than technical knowledge alone [67,68]. This underscores the importance of integrating emotional and psychological components into digital literacy programs, especially for populations at risk of high exposure to crisis media.

From a theoretical perspective, these findings can be interpreted using PMT as a conceptual lens. While PMT traditionally focuses on cognitive threat and coping appraisals [69], our results highlight associations between stress and digital media engagement, showing that higher stress was associated with greater engagement across types, whereas engagement patterns showed different associations with well-being. Accordingly, we avoid concluding that stress motivates protective or maladaptive behaviors and instead describe these as observed associations that may vary by context. Prior PMT-informed cybersecurity research illustrates that protective behavior is typically framed as something that can be shaped through modifiable organizational levers, such as clear policies and security education, training, and awareness programs, which are intended to strengthen coping appraisals and translate perceived threat into protective action [70]. PMT reviews further emphasize that threat appraisal alone is insufficient; protective responding is expected when coping appraisal, such as perceived efficacy and self-efficacy, is strong relative to perceived costs [69]. Interpreted through this lens, our finding that higher stress co-occurred with both active and passive engagement patterns suggests that crisis-related arousal may broadly increase engagement, whereas coping resources, such as resilience, may be particularly important in determining whether engagement is linked to more favorable psychological outcomes. More broadly, crisis-specific frameworks emphasize the role of information and communication technology–mediated information environments in crisis management and communication [71]. In this study, resilience was associated with lower stress and higher well-being, suggesting that coping resources are relevant when interpreting crisis-related engagement and outcomes [43]. In the context of doomscrolling, intolerance of uncertainty and resilience-related processes have been discussed as relevant correlates [53]. Overall, these findings support a cautious, theory-informed interpretation without implying causal mechanisms or model refinement beyond the variables measured in this study.

The findings have practical implications for designing public health and digital communication strategies during emergencies. Campaigns should encourage intentional and participatory media use, such as verifying sources, seeking community connection, and sharing factual information, while raising awareness about the mental health risks of excessive passive consumption. Platforms could be adapted to promote active engagement by prompting users toward verified resources or facilitating moderated peer-support communities. Educational and organizational programs that combine emotion regulation, media literacy, and resilience-building may serve as protective interventions, particularly for populations frequently exposed to distressing crisis-related media content. Evidence syntheses suggest that media literacy programs can strengthen critical evaluation of information and reduce susceptibility to misinformation [72]. A recent meta-analysis found that media literacy interventions improved fake news credibility assessment, suggesting that strengthening credibility judgment skills through structured media literacy training may support safer digital decision-making during crises, including decisions about sharing, trusting, or acting on rapidly circulating claims [73]. The development of digital tools, including mobile apps to support mental health response in disasters and remotely delivered mental health services, may strengthen mental health support and continuity of care when in-person services are disrupted by extending access to support across preparedness, response, and recovery phases, particularly when demand is high and resources are constrained [74,75].

This study presents several important strengths along with acknowledged limitations. A key strength is its focus on the understudied intersection of cybersecurity-related behavior, media engagement, and psychological well-being during an active national crisis. Moreover, the contribution of this study lies in its unique context, which enabled the observation of digital behaviors during an acute and ongoing national crisis. While previous research has explored media consumption during public health crises such as the COVID-19 pandemic [58] and has documented the psychological toll of exposure to graphic war-related content [7], this study extends this literature in several important ways.

First, rather than treating media usage as a unidimensional construct, we empirically identified 3 distinct patterns of digital engagement: active, passive, and institutional usage, each with different implications for psychological well-being. Second, by embedding these patterns within a stress-resilience framework, we demonstrated that the relationship between media engagement and well-being is not uniform but depends critically on the type of engagement. Specifically, active participation in digital platforms was associated with better well-being, consistent with the view that such behaviors may serve as a form of emotion-focused coping and social support-seeking during acute crisis [76]. In contrast, passive consumption showed a negative trend, aligning with prior findings on doomscrolling and emotional exhaustion [53]. These findings highlight the importance of moving beyond aggregate measures of screen time or media exposure and instead attending to the qualitative nature of digital engagement during wartime.

A further contribution of this study lies in its broader conceptualization of digital media usage. Unlike previous studies, which have typically focused on social media platforms (eg, Abbas et al [77]; Huckins et al [78]; Gao et al [79]), this study also examined institutional digital behaviors, including online banking and access to government and public authority websites. In an era of accelerating digitalization, digital applications are no longer limited to entertainment and social interaction, they have increasingly become the primary interface between individuals and institutional agents, such as financial institutions and governmental services. This shift may have important implications for cybersecurity, as greater reliance on digital institutional channels during crisis periods may heighten individuals’ vulnerability to cyberattacks. Enhanced cybersecurity awareness and digital literacy are therefore particularly critical during times of crisis, such as war or a pandemic, when the availability of in-person institutional contact is restricted and individuals are compelled to rely more heavily on digital alternatives. In this study, we therefore expand the scope of inquiry beyond social media, examining whether individuals turn to digital applications not only to seek information and social connection but also to manage practical and institutional needs during wartime.

While the findings provide important insights, several limitations should be considered when interpreting the results. First, the cross-sectional design precludes causal inference [80]. Although the path analysis model implies directionality, the temporal order between stress, media engagement, and well-being cannot be established from a single time-point measurement. All associations reported in this study should therefore be interpreted as correlational rather than causal. Longitudinal or experimental designs are needed to establish whether stress causally drives media engagement or whether intensive media use itself amplifies stress over time.

Second, cybersecurity awareness and knowledge were assessed using a single self-report item. In addition, media usage was operationalized directly and did not rely on previously validated instruments. While this approach captures behavior directly, this operationalization may not fully reflect the multidimensional nature of media usage awareness, which may also involve more implicit or less visible patterns of engagement. Future studies should incorporate validated measures to enable more rigorous testing of their role in shaping digital behavior during crises.

Third, the study did not measure potential confounding variables that may influence both media engagement and well-being, such as prior mental health history, social support networks, proximity to conflict zones, or direct exposure to wartime events. The observed associations between stress, media use, and well-being may therefore partly reflect the influence of unmeasured third variables rather than the relationships among the constructs themselves.

Lastly, participants were recruited through an online panel using quota sampling, which, while designed to approximate the demographic distribution of the Israeli population, does not constitute a probability sample. Self-selection bias [81] may have resulted in the overrepresentation of individuals who are more digitally engaged or more willing to report their media behaviors and psychological states. In addition, previous studies suggest that cybersecurity behaviors differ across cultures [82,83]. Accordingly, the findings should be interpreted with caution and may not be fully generalizable to populations with lower digital access or different media consumption patterns. In this vein, it should also be noted that the study was conducted within a single national context during the early months of the Israel–Hamas war, which limits the external validity of the findings. The specific cultural, political, and media environment of this conflict may not be representative of other crisis contexts, such as natural disasters, pandemics, or conflicts in different geopolitical settings. Replication across diverse crisis contexts and cultural settings is therefore required before broader conclusions regarding the relationship between stress, media engagement, and well-being can be drawn. Future research should build on these findings by examining the dynamic relationships among stress, media engagement, and well-being using longitudinal designs and more comprehensive measures of digital behavior.

Future research should test the direction and mechanisms underlying the observed associations by using longitudinal and experience-sampling designs that capture within-person changes in stress, resilience, and engagement over time and across crisis phases. Studies that combine self-reports with objective indicators of digital behavior (such as passive sensing or platform-provided usage metrics) and more detailed characterization of content (such as exposure to graphic material, misinformation, or institutional alerts) may help distinguish when engagement is adaptive vs burdensome. It is also important to examine heterogeneity across subgroups (ie, age, prior trauma, frontline exposure, and baseline mental health) and across platforms and engagement modes, including institutional channels that are often overlooked. Finally, intervention-focused research should evaluate targeted, real-world strategies (ie, media literacy and resilience-building components embedded in crisis communication, platform-based prompts toward verified sources, and moderated peer-support pathways) to determine which approaches most effectively promote healthier engagement and protect well-being during prolonged crises.

This study offers new evidence on wartime digital media engagement by moving beyond time-based exposure measures and empirically identifying distinct engagement patterns, including active, passive, and institutional modes, which are rarely examined together in crisis-media research. Compared with previous work that often focuses on overall media exposure or social media alone, our findings offer a more detailed account of how different forms of engagement relate to stress, resilience, and well-being during an ongoing armed conflict. This contribution advances the field by presenting an empirically grounded framework for interpreting when engagement may be supportive vs potentially burdensome in prolonged crises. Practically, the results can inform public health and organizational communication by emphasizing targeted guidance that encourages adaptive, intentional engagement (ie, seeking verified information and social support) while reducing risks from excessive passive exposure. They can also help platforms and service providers direct users to reliable institutional information and supportive resources.