Abstract
Against the backdrop of deepening population aging in China, the phenomenon of empty-nest elderly in rural areas has become increasingly prominent. This study explores the impact of Douyin usage on the subjective well-being (WB) of rural empty-nest elderly, constructing a research model based on the stimulus-organism-response (SOR) framework. The model integrates seven constructs: interactivity, entertainment, relative advantage, compatibility, habit, flow experience, and WB. Data were collected through questionnaires from 407 elderly respondents in rural Sichuan Province, analyzed using a hybrid approach combining partial least squares structural equation modeling (PLS-SEM) and artificial neural network (ANN). PLS-SEM results show that interactivity and entertainment value significantly enhance WB through direct and indirect effects mediated by usage habit and flow experience. Entertainment value exerts a stronger influence on flow experience, while interactivity plays a more pivotal role in cultivating usage habits. The ANN analysis validates the model’s robustness, reveals nonlinear interactions among constructs, and ranks the relative importance of predictors, aligns broadly with the findings of PLS-SEM. The findings not only expand the theoretical understanding of digital technology use and elderly well-being but also offer empirical support for digital solutions to rural aging governance.
Similar content being viewed by others
Introduction
The phenomenon of the “empty nest” is rapidly proliferating across Asian societies amid accelerating urbanization, with China exhibiting a particularly striking trajectory1. As millions of working-age individuals migrate from rural areas to urban centers, the rural landscape is becoming increasingly “hollowed out,” leaving behind a disproportionately large aging population. Empty nesters now account for more than half of China’s elderly, and 65% in rural regions alone2. This demographic shift has triggered profound emotional and social dislocations, particularly among elderly individuals who live alone due to their adult children’s long-term absence3. These individuals are not merely deprived of physical care—they also suffer from deep emotional loneliness and reduced social participation4, significantly elevating their risks of psychological disorders. This emerging mental health crisis poses a dual challenge for both families and the broader public health system5,6. Empty nest life reveals a widening gap in emotional support among rural elders, manifesting in pervasive experiences of loneliness7, anxiety, and depression8. The intergenerational separation driven by socio-structural change fundamentally undermines the traditional model of familial eldercare. Meanwhile, TikTok, one of the world’s fastest-growing social media platforms9, is rapidly expanding into rural markets, emerging as a new digital frontier for older adults10. Enabled by widespread mobile internet access, short-form video platforms now offer not only diverse content consumption but also user-friendly tools for content creation and personal expression11. This digital intervention is reshaping rural elders’ social lives by providing a virtual conduit for emotional expression and informational exchange, thus challenging geographic isolation. Indeed, growing evidence suggests a significant positive association between social media use and improved well-being among older adults12,13. However, existing studies on quality of life improvements among empty nesters remain largely confined to health behaviors14 and familial support mechanisms15, with limited focus on how digital media may influence their subjective life experiences. Indicators of better mental health, including happiness, self-esteem, and reduced loneliness, are critically underexplored within the context of aging and digital engagement16. Furthermore, empirical research on how older adults use platforms like TikTok to renegotiate aging remains in its infancy10.
As a representative digital media platform, TikTok utilizes an algorithm-driven content recommendation mechanism to deliver content tailored to the interests of rural empty-nest elderly. Notably, research has established a relationship between TikTok usage and mental health impacts17. This platform possesses one of the most sophisticated algorithmic systems, rendering it more addictive than other social media platforms18. For rural empty-nest elderly, this technological affordance enables rapid immersion into video-generated atmospheres, facilitating temporary escape from loneliness and solitude during viewing sessions, which provides immediate emotional comfort. Simultaneously, the platform’s interactive features, such as liking, commenting, and sharing, facilitate social connections between empty-nest elderly and other users. Through these engagements, this demographic perceives enhanced self-value and experiences the warmth of social care, thereby alleviating loneliness and strengthening self-identity. Current scholarly discourse on short-video platforms for elderly populations primarily examines: user behavior characteristics19, interface design adaptability20, and usability parameters. Building upon existing research while addressing a critical gap, this study investigates how TikTok’s experience design extends its influence to psychological dimensions.
In response, this study adopts the Stimulus–Organism–Response (SOR) framework, integrating Diffusion of Innovations Theory and Flow Theory to examine how TikTok use reshapes life experiences among rural empty nesters. Specifically, we introduce TikTok’s platform-specific attributes as stimuli, investigate how the interplay between entertainment content and rural cultural elements facilitates psychological flow, and assess their downstream effects on loneliness alleviation, life satisfaction, and cultural identity reinforcement. The model also tests moderating factors such as intergenerational support and neighborhood imitation. By constructing a rural-contextualized SOR model, this research aims to uncover the mechanisms by which digital engagement transforms aging experiences and inform future designs of inclusive digital interventions for aging populations in under-resourced regions.
Literature review and research hypotheses
Literature
Originally proposed by Mehrabian and Russell21, the stimulus–organism–response (SOR) model has become a foundational theory in environmental psychology and has since been widely adopted in fields such as marketing, consumer behavior, and digital technology research22,23. The model posits that external environmental stimuli (S) influence an individual’s internal cognitive and emotional states (O), which in turn lead to specific behavioral responses (R). Particularly in the context of consumer psychology, the SOR framework underscores the importance of understanding how external cues activate internal psychological mechanisms that ultimately shape decision-making and behavioral outcomes24. Recent studies have extended the SOR paradigm to digital environments, including applications in artificial intelligence services25 and short-form video platforms. For instance, prior research has investigated how features of short video apps influence users’ willingness to share content26,27 and drive purchase intentions28. These studies typically examine motivations for media use, behavioral continuance among college students29, and the impact of platform interactivity and algorithmic recommendation on e-commerce engagement30.
However, research on social media usage among older adults has largely focused on barriers such as the digital divide, technology resistance, or addictive behaviors31,32. The use of technology for communication and learning correlates with self-efficacy, indicating that digital literacy not only enhances quality of life but also improves functional performance among older adults33. Digital self-efficacy specifically plays a pivotal role in modulating elderly users’ responses to digital stimuli. Notably, older users with higher digital self-efficacy are more likely to perceive short-video platforms as accessible tools. While the relationship between self-efficacy and usage intention is non-significant in middle-aged cohorts, a robust positive effect exists among the elderly34. This heightened self-efficacy mitigates anxiety and fosters deeper engagement. Furthermore, enhanced social media self-efficacy improves elderly users’ self-esteem and reduces loneliness35.
TikTok, one of China’s most popular short video platforms, offers a highly stimulating digital environment through diverse content ecosystems and accessible interaction features. Although originally branded as a virtual playground for younger users, TikTok has seen increasing engagement from older adults, some of whom have become prominent creators with millions of followers10. Elderly users engage with content aligned with their interests, such as health advice, daily living tips, or traditional opera performances36. For empty-nest elders—often isolated and emotionally underserved—such emotionally resonant content serves as a potent psychological stimulus. These stimuli (S), when aligned with users’ affective and cognitive needs, activate internal mediators (O), including emotional flow and habitual engagement. This ultimately leads to observable behavioral outcomes (R), such as enhanced well-being and life satisfaction. The SOR model thus provides a comprehensive framework to map the dynamic process from external digital stimulation to internal processing and emotional response.
Based on this theoretical foundation, the present study employs the SOR framework to explore the mechanisms by which TikTok usage affects subjective well-being among rural empty-nest elderly users.
Stimulus (S): External platform stimuli, including interactivity, entertainment value, relative advantage, and compatibility, serve as key environmental triggers.
Organism (O): Internal psychological states, usage habits, and flow experience function as mediating variables that bridge stimulus and response.
Response (R): The outcome variable, well-being, reflects the emotional and cognitive responses resulting from sustained engagement with platform stimuli.
This integrative approach contributes a novel, context-sensitive application of the SOR model in aging research, offering both theoretical and practical insights into how digital platforms like TikTok may enhance the life experience of isolated elderly populations in rural China.
Stimulus variables and hypothesis development
Interactivity
In the context of social media studies, numerous scholars have affirmed that platform attributes, such as interactivity, entertainment, and relative advantage, have significant impacts on users’ psychological states and behavioral outcomes37,38,39. As a prototypical short-form video platform, TikTok exhibits strong interactivity through features such as real-time commenting, bullet chats, and live co-broadcasts. These affordances transform users from passive recipients into active participants, enhancing both emotional value and engagement. According to Xi and Hamari40, interactive environments intensify users’ sense of presence by enabling them to manipulate and influence their digital surroundings. Thus, interactivity, as a stimulus (S), may trigger internal psychological responses (O) and ultimately influence user outcomes (R).
H1a: Interactivity (IN) has a positive effect on users’ habit formation (HB) in TikTok usage.
H1b: Interactivity (IN) has a positive effect on users’ flow experience (FT).
H1c: Interactivity (IN) has a positive effect on users’ perceived well-being (WB).
Entertainment
Entertainment is another critical stimulus within short video platforms, functioning as a central driver of user engagement and retention37. As Negash et al.41 suggest, entertainment aims to induce pleasure and captivate attention. Lo et al.42 further argue that intrinsic enjoyment is vital to consumer satisfaction and continued digital media use. The entertaining nature of TikTok, manifested through humorous content, special effects, and viral challenges, offers emotional relief and immersion, thereby enhancing hedonic value and overall satisfaction43,44.
H2a: Entertainment (EN) has a positive effect on users’ habit formation (HB) in TikTok usage.
H2b: Entertainment (EN) has a positive effect on users’ flow experience (FT).
H2c: Entertainment (EN) has a positive effect on users’ perceived well-being (WB).
Relative advantage
The Diffusion of Innovations Theory45 highlights four core factors that shape innovation adoption: the nature of the innovation itself, communication channels, time, and the surrounding social system46,47. Drawing from Diffusion of Innovations Theory, relative advantage and compatibility are critical innovation attributes that influence adoption rates. For late-adopter elderly users, TikTok’s relative advantage and compatibility reduce adoption barriers, aligning with the theory’s emphasis on innovation characteristics driving behavioral change.
Within this framework, relative advantage refers to the extent to which an innovation is perceived as superior in terms of efficiency, usability, or performance48. TikTok embodies these advantages through real-time interactivity, intelligent algorithmic recommendations, and a highly engaging content ecosystem49,50. The attributes of an innovation are critical determinants of its adoption rate, encompassing relative advantage, compatibility, complexity, trialability, and observability51. This theory has been widely applied in fields such as technology adoption and social media usage52,53,54. Rural empty-nest elderly individuals are categorized as late adopters of digital technologies, and their acceptance of platforms is more heavily reliant on their perception of innovation attributes. Studies have shown that elderly individuals with high motivation can achieve knowledge levels comparable to, or even higher than, those of young people55. However, older adults often encounter more difficulties in adopting technologies, such as a gradual decline in visual and auditory abilities, and reduced flexibility leading to typing challenges56. Thus, the Diffusion of Innovations Theory provides a core theoretical foundation for understanding the selection of relative advantage and compatibility as stimulus variables (S) in this study.
Relative advantage refers to the perceived superiority of an innovation over existing alternative solutions. TikTok demonstrates relative advantages over traditional forms of entertainment and aligns with the elderly’s lifestyle rhythms and technical capabilities. Its advantages compared to traditional entertainment and its compatibility with elderly users’ daily routines and technical proficiency precisely meet the specific needs of this group. Importantly, the app’s design is optimized for diverse age groups and technological abilities. It features a simple interface and algorithmically personalized content, enhancing usability across demographics57. As such, the relative advantage of TikTok may influence both internal cognitive responses and behavioral outcomes.
H3a: Relative advantage (RA) has a positive effect on users’ habit formation (HB) in TikTok usage.
H3b: Relative advantage (RA) has a positive effect on users’ flow experience (FT).
H3c: Relative advantage (RA) has a positive effect on users’ perceived well-being (WB).
Compatibility
Compatibility, defined as the degree to which an innovation aligns with users’ existing values, needs, and experiences, is a critical determinant of technology adoption58. Prior studies have demonstrated the positive influence of compatibility on behavioral intention and user satisfaction59,60. TikTok’s success lies partly in its ability to accommodate diverse cultural expressions and emotional needs, while also supporting self-realization and social interaction61. By simplifying content creation processes and supporting fragmented usage patterns, TikTok has effectively integrated itself into the daily lives of elderly users, including empty nesters. Compatibility between users’ skill levels and task demands during interaction facilitates the emergence of flow experience62. This alignment enhances user engagement and reduces entry barriers.
H4a: Compatibility (CB) has a positive effect on users’ habit formation (HB) in TikTok usage.
H4b: Compatibility (CB) has a positive effect on users’ flow experience (FT).
H4c: Compatibility (CB) has a positive effect on users’ perceived well-being (WB).
Organism variables and hypothesis development
Habit
Habit refers to the degree to which a behavior becomes automatic through repeated practice, enabling individuals to perform actions with minimal cognitive effort63. As Lu et al.64 point out, cultivating healthy digital habits can significantly improve the living experience and quality of life among older adults, particularly those living alone. Habits reduce decision-making uncertainty and enhance individuals’ sense of behavioral control, thereby improving psychological satisfaction65. For older users, who often experience cognitive decline, the formation of stable and routine digital usage patterns is especially critical66. TikTok’s age-friendly interface and rich social features help reinforce these usage patterns among elderly users. By lowering learning costs and aligning with daily behavioral tendencies, the platform facilitates long-term engagement. Previous studies have confirmed that habit plays a pivotal role in the continued use of digital technologies67. Habit automatically manifests an individual’s psychological disposition68, representing an internalized behavioral tendency formed through repeated platform use, rather than an immediate emotional response. The formation of habit reduces cognitive effort and enhances behavioral control, serving as a mediator between platform characteristics and well-being.
H5: Habit (HB) has a positive effect on users’ perceived well-being (WB) in TikTok usage.
Flow experience
The concept of flow describes a mental state in which individuals become fully immersed in an activity, characterized by deep focus, loss of self-consciousness, and a sense of control over the environment69. It represents the intrinsic motivation for a certain activity and can generate satisfying pleasure and enjoyment70. Flow is a powerful predictor of user satisfaction, behavioral intention, and media loyalty71. High-quality interaction design reduces perceived technological complexity, allowing users to achieve immersive engagement72. On TikTok, flow is closely tied to content relevance and the depth of interaction. When rural empty-nest elders encounter content that aligns with their interests—such as traditional opera, health advice, or family-oriented storytelling—they often enter a time-distorted state of concentration and enjoyment73. This immersive state enhances emotional pleasure and may trigger “social contagion effects,” such as sharing content and initiating peer interactions. Although flow theory has been validated in VR, gaming, and online learning contexts74,75, its application to the digital experiences of rural elderly users remains underexplored.
H6: Flow experience (FT) has a positive effect on users’ perceived well-being (WB) in TikTok usage.
Response variable
In the S–O-R model, the response (R) encompasses users’ emotional and behavioral outcomes, including attitudinal shifts and behavioral intentions76. Among these, subjective well-being (WB) has emerged as a key psychological indicator reflecting emotional balance, social confidence, and mental resilience77. Empty-nest elders, those who live alone or only with a spouse, often face a lack of daily care and emotional support5. For rural empty-nest elderly, WB is manifested through reduced loneliness, enhanced life satisfaction, and strengthened cultural identity, which are direct outcomes of sustained engagement with platform stimuli. Well-being may manifest through reductions in stress and loneliness, improvements in emotional regulation and attention, and increased self-esteem78. Currently, some scholars have taken subjective well-being as R79,80. By addressing the emotional void of empty-nest life, TikTok transforms digital engagement into a psychosocial intervention, where WB serves as the measurable endpoint of this process. TikTok, by offering nostalgic content, social interaction opportunities, and age-relevant content streams, enables older users to fulfill emotional and psychological needs. These interactions reduce loneliness and amplify positive emotional experiences, thereby enhancing overall well-being. In this context, TikTok serves not just as a digital entertainment platform but as a psychosocial intervention tool, mediating the pathway from external digital stimuli to internal psychological fulfillment.
Based on the above discussion and analysis, the study proposes the following hypotheses, as illustrated in (Fig. 1).
The theoretical model.
Methods
This study employed Partial Least Squares Structural Equation Modeling (PLS-SEM) to validate the hypothesized relationships. A structured questionnaire was designed comprising two sections. The study was approved by the Ethics Committee of Macau Polytechnic University, and all methods were performed in strict accordance with the relevant guidelines and regulations. The first section collected demographic data, while the second focused on measuring core constructs using items adapted from established and validated scales drawn from prior literature to ensure measurement reliability and construct validity81. To tailor the instrument to the research context, several items were linguistically and contextually adjusted. The final instrument included 7 constructs and 21 items, each measured using a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree) (Table 1). To ensure clarity and cultural appropriateness, the instrument underwent two rounds of pre-testing. In the first round, 12 older adults were randomly selected to assess item readability. Based on their feedback, minor revisions were made to enhance comprehensibility. In the second round, a pilot survey was conducted with 13 rural empty-nest elderly individuals from Langzhong City, Sichuan Province. Particular attention was paid to regional dialect variations and local expressions, with further refinements made accordingly.
Employing a door-to-door convenience sampling approach, formal data collection was conducted from January to June 2025. This study focused specifically on rural empty-nest elderly residing in high-density areas of Sichuan Province, including Chengdu, Leshan, and Nanchong. To ensure data integrity, face-to-face interviews formed the primary method, leveraging direct engagement to maximize informational completeness . Despite inherent limitations in generalizability, convenience sampling remains methodologically appropriate for exploratory research90. To mitigate measurement bias, our team refined questionnaire phrasing during the pre-test phase by adapting terminologies to dialectal variations in Langzhong City, Sichuan. This linguistic calibration guaranteed consistent comprehension of digital usage behavior items across Chengdu, Leshan, and Nanchong respondents, reducing cultural-linguistic distortion risks. During formal fieldwork, identify and prioritize remote households using local administrative registers, actively counterbalancing spatial distribution skew inherent to convenience sampling. To ensure respondent eligibility, a brief screening was administered before participation, asking: “Have you used TikTok in the past six months?” “Do you currently live alone (or only with your spouse)?”Only those answering “yes” to both were included in the sample91; others were excluded from further participation. A total of 407 valid questionnaires were ultimately collected, exceeding the recommended subject-to-parameter ratio of 10:1 for structural equation modeling applications92. Of the 407 valid questionnaires collected, the gender distribution was relatively balanced: 50.6% female and 49.4% male. Monthly income was concentrated between 2000 and 5000 CNY, accounting for 44.7% of the sample. In terms of educational background, 86.7% of respondents had received elementary education or below. Furthermore, 84.8% of elderly participants reported that their children visited them no more than five times per year. These sample characteristics are highly consistent with the known demographic and social profiles of rural empty-nest elderly populations in China. Table 2 presents the complete demographic data.
Results
Reliability and internal consistency
To assess the reliability of the measurement model, both factor loadings and cross-loadings were examined. Item loadings exceeding 0.70 were considered acceptable, indicating strong indicator reliability93. All items in this study met or exceeded the threshold, confirming item-level reliability94. Furthermore, both Cronbach’s α and Composite Reliability (CR) values for all constructs ranged between 0.70 and 0.95, satisfying the standard for internal consistency. These results suggest that the constructs reliably measure the intended latent variables without redundancy or overfitting94. To control for common method variance (CMV), we employed Harman’s single-factor test using SPSS 25.0. The unrotated principal component analysis revealed that the first factor accounted for only 29.320% of the total variance, which is well below the critical threshold of 50%. This indicates that common method bias is not a serious concern in this study95. Convergent validity was evaluated using the Average Variance Extracted (AVE) metric. All constructs exhibited AVE values above 0.50, indicating that more than half of the variance in the indicators was captured by the latent constructs, thus confirming sufficient convergent validity96. To test for multicollinearity, the Variance Inflation Factor (VIF) values were computed for all indicators. The VIF values ranged between 1.437 and 3.060, well below the conservative threshold of 5.097. These results demonstrate that no significant multicollinearity exists among the observed variables, satisfying the assumptions required for structural equation modeling. The Standardized Root Mean Square Residual (SRMR) value was assessed to evaluate the overall model fit. The SRMR value was found to be less than 0.08, indicating that the model exhibits an acceptable level of fit and is well-suited for predictive analysis96. As shown in Table 3, the data demonstrate that the measurement model exhibits good internal consistency, item reliability, and convergent validity.
For discriminant validity, we followed the example of Fornell and Larcker93,98, where the square root of the AVE should exceed its highest correlation with items in different constructs (Table 4). To further ensure the robustness of discriminant validity and mitigate potential biases associated with PLS-SEM—such as overestimation of indicator loadings or underestimation of inter-construct correlations—the Heterotrait–Monotrait Ratio (HTMT) was also calculated. The highest HTMT value observed was 0.813, which is below the conservative threshold of 0.85 for conceptually unrelated constructs. According to Hair and Alamer99, a threshold of 0.90 is acceptable for conceptually related constructs. The HTMT results further affirm the absence of discriminant validity concerns in this study.
Structural model and hypothesis test
To assess the predictive relationships between exogenous and endogenous variables, the bootstrap resampling method in PLS-SEM was employed with 5000 bootstrap samples to estimate standard errors and test the statistical significance of path coefficients. As shown in Fig. 2 and Table 5, the model demonstrates strong explanatory power. The coefficient of determination (R2) for the dependent variable well-being (WB) reached 0.721, indicating that over 72% of the variance in well-being can be explained by the model’s predictor variables, well above the threshold for a large effect size (R2 ≥ 0.67)100. Additionally, habit (HB) and flow experience (FT) yielded R2 values of 0.290 and 0.522, respectively, both exceeding the effect threshold (R2 ≥ 0.200), suggesting robust construct-level explanatory power. The predictive relevance (Q2) was evaluated using the blindfolding procedure. The resulting Q2 values were:WB: 0.694, HB: 0.274, FT: 0.509. All values exceeded the recommended benchmarks of 0.02 (small), 0.15 (medium), and 0.35 (large), confirming that the model possesses strong out-of-sample predictive accuracy101. The structural model analysis revealed that all six predictors exerted significant positive effects on users’ perceived well-being. Specifically, interactivity (H1c: β = 0.377, p < 0.001), entertainment (H2c: β = 0.396, p < 0.001), relative advantage (H3c: β = 0.234, p < 0.001), compatibility (H4c: β = 0.280, p < 0.001), habit (H5: β = 0.142, p < 0.001), and flow experience (H6: β = 0.104, p = 0.008) all demonstrated statistically significant and positive relationships with well-being. Regarding habit formation, interactivity (H1a: β = 0.335, p < 0.001), entertainment (H2a: β = 0.299, p < 0.001), and compatibility (H4a: β = 0.252, p < 0.001) were found to significantly promote the development of user habits, whereas the effect of relative advantage (H3a: β = 0.045, p = 0.304) was not statistically supported. As for flow experience, all four stimulus variables—interactivity (H1b: β = 0.382, p < 0.001), entertainment (H2b: β = 0.395, p < 0.001), relative advantage (H3b: β = 0.284, p < 0.001), and compatibility (H4b: β = 0.307, p < 0.001)—exhibited significant positive effects, confirming their importance in facilitating immersive user experiences on the TikTok platform.
The theoretical model.
Mediating effect test
The mediation analysis further revealed the indirect mechanisms by which stimulus variables influence perceived well-being through the organism constructs of HB and FT (Table 6). For EN, the indirect path EN → FT → WB (95% CI [0.011, 0.074]), indicating that entertainment positively enhances user well-being by stimulating immersive flow experiences. In comparison, the EN → HB → WB pathway demonstrated a slightly stronger effect (95% CI [0.025, 0.073]), suggesting that entertainment also improves usage experience by fostering habit formation, highlighting its dual role in both short-term engagement and long-term behavioral integration. For IN, the indirect effect through habit (IN → HB → WB, 95% CI [0.008, 0.053]) was stronger than that through flow (IN → FT → WB, 95% CI [0.018, 0.057]). Both paths were significant and positive (95% CIs excluding 0), indicating that interactivity not only enhances immediate enjoyment via flow states but also reinforces sustainable user engagement through habit development. In the case of CB, the CB → HB → WB path showed a moderate and significant effect (95% CI [0.022, 0.066]), suggesting that compatibility contributes meaningfully to usage experience by facilitating the formation of habitual behaviors. Meanwhile, the CB → FT → WB path demonstrated a weaker effect (p = 0.011) , the practical effect of compatibility on stimulating flow is less stable and relatively limited, though it still plays a non-negligible role in the immersive experience. As for RA, the RA → FT → WB pathway was statistically significant but relatively weak (95% CI [0.009, 0.058]), suggesting that the perceived superiority of the platform contributes to user well-being primarily through momentary flow experiences, with a small but stable positive effect. However, the indirect path RA → HB → WB (95% CI [0.011, 0.072]) was not significant. This implies that relative advantage has extremely limited power to drive long-term user habits, and its perceived benefits are more likely linked to immediate gratification than to routine behavioral reinforcement.
SEM–ANN hybrid analysis
To enhance the explanatory and predictive capabilities of the model, this study integrates Structural Equation Modeling (SEM) with Artificial Neural Networks (ANN) to capture both linear causal relationships and complex nonlinear interactions in the dataset. While SEM is well-suited for testing causal structures in a theory-driven framework, ANN excels at modeling intricate, nonlinear patterns that may not be well-captured through traditional linear approaches102,103,104.
Building on the validated SEM results, a hybrid SEM–ANN model was developed to refine prediction accuracy and explore latent patterns, combining the confirmatory power of SEM with the adaptive learning capability of multilayer feedforward neural networks105. The ANN architecture used in this study comprises three layers: an input layer, a hidden layer, and an output layer106. Three distinct neural network models were constructed to predict key endogenous variables WB、HB, and FT, based on the significant predictors identified through SEM. Model A was designed to predict WB. The input layer consisted of IN, EN, RA, CB, HB, and FT, with WB as the single output node (Fig. 3). Model B aimed to predict HB. Its input variables included IN, EN, and CB, and the output was HB. Model C was developed to predict FT. The input layer included IN, EN, RA, and CB, with FT as the output. Their design leverages the causal validation power of SEM with the nonlinear approximation strength of ANN, enabling a more nuanced understanding of user behavior patterns and deepening insight into the mechanism through which TikTok usage affects the well-being of rural empty-nest elderly users.
ANN model A.
Neural network architecture and predictive accuracy
The hybrid SEM–ANN analysis was implemented using multilayer perceptron (MLP) neural networks, with predictive performance evaluated through tenfold cross-validation106. This approach ensures the model’s generalizability while minimizing the risk of overfitting. The full dataset was randomly partitioned into ten equal folds: 90% of the data was used for model training, and the remaining 10% was reserved for independent testing107,108. Both the hidden and output layers employed the sigmoid activation function, enabling the network to model nonlinear mappings between predictors and outcomes. Before training, all input variables were normalized to a [0,1] range to improve convergence efficiency and ensure compatibility across features. The training process was governed by backpropagation, and model fit was assessed through the Standardized root mean square residual (SRMR), confirming adequate model specification109. The root mean square error (RMSE) values were used to evaluate prediction accuracy (Table 7). The average RMSE for the training sets ranged between 0.079 and 0.145, while the RMSE for the testing sets remained between 0.080 and 0.143, indicating strong convergence and robust predictive performance110. These results validate the use of the SEM–ANN hybrid model as a complementary analytical technique capable of capturing both the theoretically grounded causal structure and the complex nonlinear relationships embedded within user behavioral data.
Sensitivity analysis
To assess the relative importance of each input factor in predicting the target outcomes, a sensitivity analysis was conducted for the ANN models (Tables 8, 9). This technique quantifies the impact of each input neuron or factor on the output neuron by evaluating their contribution to the prediction process111. Specifically, determinant importance was used to calculate the normalized importance, which represents the ratio of each variable’s influence relative to the most influential predictor, expressed as a percentage112. The analysis was performed across all three neural network models—Model A (predicting well-being), Model B (predicting habit), and Model C (predicting flow experience). Based on the results, the predictors were ranked by their normalized importance to determine which factors played the most critical role in driving each target construct. This ranking provides valuable insights into the nonlinear explanatory power of each stimulus and organism variable.
Comparison of ANN and PLS-SEM results
A comparative analysis between the PLS-SEM and ANN models was conducted to examine the consistency of variable importance rankings across methodologies (Table 10). In Model A (predicting well-being), noticeable differences were observed in the ranking of entertainment (EN) and interactivity (IN) between the two methods. This discrepancy suggests the potential presence of latent or interactional factors influencing the effects of EN and IN—factors that may be obscured in real-world complexity and not fully captured by the linear modeling approach of PLS-SEM113. These findings indicate that ANN’s nonlinear learning capability can uncover variable interdependencies that may be undetected in traditional SEM frameworks114. In contrast, for Model B (habit) and Model C (flow experience), the ranking of independent variables showed strong alignment between PLS-SEM and ANN results, reinforcing the stability and reliability of the structural model. This consistency affirms that the causal pathways validated through SEM are also supported by ANN’s predictive strength. Such convergence is particularly important because ANN accounts for both linear and nonlinear relationships, thus providing a higher-resolution validation of causal structure115. Overall, the close agreement between the two approaches—especially in Models B and C—demonstrates the robustness of the PLS-SEM model and supports its suitability for explaining behavioral mechanisms in the context of TikTok usage among rural elderly. Moreover, the hybrid SEM–ANN approach enriches the interpretation of causal influence by offering complementary perspectives on the strength, stability, and interaction of key predictors116.
Discussions
The findings of this study underscore the profound impact of short-form digital media, particularly TikTok, on the psychological well-being of empty-nest elderly individuals. Existing research has shown that this population is significantly more prone to depressive symptoms than their non-empty-nest counterparts117, and digital entertainment platforms offer promising avenues for improving their mental health118. Enjoyment-oriented media consumption is a key driver of technology adoption among older adults, as it satisfies emotional needs and enhances pleasure, thereby promoting sustained usage119. From a product attribute perspective, interactivity plays a critical role. Rich interactive experiences not only enhance user enjoyment (H1c) but also foster habit formation (H1a). Features such as comments, likes, and video-sharing mechanisms facilitate repeated usage, while real-time, responsive interaction forms the foundation for immersive flow states (H1b). This finding suggests that platform designers should prioritize real-time feedback and personalized interaction interfaces to elicit stronger flow experiences and emotional engagement88. Entertainment also directly enhances users’ emotional states (H2c), and its impact on flow (H2b) exceeds that of interactivity, suggesting that engaging content more effectively draws users into focused attention. Furthermore, entertainment promotes habitual use (H2a), indicating that localized and interest-based content ecosystems are essential to maintaining user retention among elderly users. Relative advantage, while positively associated with well-being (H3c), plays a more limited role in shaping habit (H3a), reflects the functional primacy principle governing elderly technology adoption: older adults systematically prioritize core functionality and usability over advanced features, with perceived advantage contributing to psychological outcomes only after these basic functional needs are met120,121. Crucially, this context-dependent nature of technological determinants extends to compatibility, as evidenced by divergent empirical findings: while Liao et al.122 documented compatibility’s significant positive effect on satisfaction in goal-directed health device contexts, Choi et al.123 observed no significant impact of TikTok usage on habit formation in entertainment-driven environments. This fundamental divergence originates from task nature variance, outcome variable disparity, and elderly engagement patterns, collectively underscoring how gerontechnology effects are contingent upon alignment between technological attributes and population-specific motivational frameworks. Relative advantage does positively influence flow (H3b), suggesting that innovation can still stimulate curiosity and engagement. Elderly users, particularly those new to digital media, are more likely to embrace platforms that offer emotionally resonant, intuitive experiences124. Compatibility strongly correlates with enhanced well-being (H4c), and by lowering usage barriers, it also promotes habit formation (H4a). However, poor compatibility can disrupt flow (H4b). These results emphasize the importance of age-friendly design features, such as interface simplicity and offline accessibility, especially for rural users facing unstable network conditions. Habituation (H5) significantly predicts well-being, consistent with prior findings that habitual behavior strengthens positive affective responses125. Similarly, flow (H6) positively affects well-being, though its impact is most potent when balanced with appropriate levels of content challenge and feedback. When digital experiences align with users’ lifestyles, they generate a sense of control and technological self-efficacy126. Routine use also reduces cognitive effort, reinforcing satisfaction, similar to patterns observed in social gaming127.
TikTok’s interactivity and entertainment attributes positively influence psychological outcomes through the mediating roles of habit and flow. These results provide actionable design strategies for age-friendly digital platforms: enhancing interactive functionality, focusing on interest-based entertainment, and optimizing system compatibility. The ANN-based sensitivity analysis further confirmed these priorities. In Model A, IN and EN ranked highest, confirming that both social engagement features and content enjoyment are central to enhancing well-being. Other variables (CB, RA, FT) followed in decreasing order of importance, suggesting that emotional connectivity and ease of use take precedence over deep immersion for this demographic. In Model B, IN ranked highest with 100% relative importance, underscoring that repeated social feedback strongly reinforces behavioral regularity, while EN(85.5%) and CB(83.3%) followed as secondary influencers. Notably, in Model C, EN (100%), IN (96.7%), and CB (73.4%) emerged as key drivers of immersive flow experiences, with RA contributing moderately (65.4%).
Conclusions
This study integrates the Stimulus–Organism–Response (S–O-R) framework, Diffusion of Innovations theory, and Flow theory to offer a multidimensional understanding of digital media usage among empty-nest elderly users. It confirms that interactivity and entertainment, as core external stimuli, enhance life satisfaction through the mediating mechanisms of habit and flow. These results expand the theoretical scope of the S–O-R model to address the unique dynamics of intergenerational separation and social disconnection in aging populations. PLS-SEM validated linear causal relationships, while ANN uncovered nonlinear synergies, threshold effects and compensatory interactions between TikTok’s attributes, innovation diffusion factors, and empty-nesters’ emotional gratification. This dual approach models nonlinear dynamics in elderly users’ digital behavioral responses, enhancing theoretical insights. Practically, the findings offer a roadmap for improving digital platforms targeting rural older adults. Developers should prioritize real-time interaction and design intergenerational communication features to capitalize on the habit-forming potential of interactivity. Additionally, localized, culturally relevant content ecosystems should be created to heighten engagement. Compatibility should be optimized across devices and network conditions, particularly in underdeveloped areas. Finally, platform design should shift from highlighting technical specifications to emphasizing intuitive functionality and emotional value, aligning with older users’ preference for simplicity and usefulness.
This study has several limitations, future research could conduct cross-regional and cross-population comparative studies. Second, the study adopts a cross-sectional design, which does not capture the long-term effects of TikTok usage on psychological well-being. Longitudinal studies are recommended to track these dynamics over time. Finally, future work may build on the integrated S–O-R–Innovation Diffusion model to explore digital media’s role in reducing loneliness, incorporating constructs such as social capital and digital empowerment. Neuroscientific tools could also be employed to investigate the physiological mechanisms underlying flow experiences in older adults.
Data availability
The data supporting the findings of this study are available from the corresponding author, Junping Xu, upon reasonable request.
References
Hartanto, A., Sim, L., Lee, D., Majeed, N. M. & Yong, J. C. Cultural contexts differentially shape parents’ loneliness and wellbeing during the empty nest period. Commun. Psychol. 2 (1), 105 (2024).
Tao, T., Jin, G. Z. & Guo, Y. L. Empty-nest elderly households in China: Trends and patterns. Popul. Res. 47, 58–71 (2023).
Zheng, C. & Zhang, H. Latent profile analysis of depression among empty nesters in China. J. Affect. Disord. 347, 541–548 (2024).
Gao, Y. et al. Social participation and health in middle-aged and older empty nesters: A study on gender differences. SSM – Popul. Health 25, 101641 (2024).
Fang, E. F. et al. A research agenda for aging in China in the 21st century. Ageing Res. Rev. 24, 197–205 (2015).
Li, X., Fan, L. & Leng, S. X. The aging tsunami and senior healthcare development in China. J. Am. Geriatr. Soc. 66 (8), 1462–1468 (2018).
Cheng, P. et al. Disparities in prevalence and risk indicators of loneliness between the rural empty nest and non-empty nest older adults in Chizhou, China. Geriatr. Gerontol. Int. 15 (3), 356–364 (2015).
Zhai, Y. et al. Association of empty-nesters with depressive symptom in a Chinese elderly population: A cross-sectional study. J. Affect. Disord. 187, 218–223 (2015).
Vaterlaus, J. M., & Winter, M. TikTok: An exploratory study of young adults’ uses and gratifications. Soc. Sci. J. 1–20 (2021).
Ng, R. & Indran, N. Not too old for TikTok: How older adults are reframing aging. Gerontologist 62 (8), 1207–1216 (2022).
Wang, Y. Humor and camera view on mobile short-form video apps influence user experience and technology-adoption intent: An example of TikTok (DouYin). Comput. Hum. Behav. 110, 106373 (2020).
Cotten, S. R., Schuster, A. M. & Seifert, A. Social media use and well-being among older adults. Curr. Opin. Psychol. 45, 101293 (2022).
Harriger, J. A., Thompson, J. K. & Tiggemann, M. TikTok, TikTok, the time is now: Future directions in social media and body image. Body Image 44, 222–226 (2023).
Zhang, C. et al. Exploring the influencing factors of quality of life among the empty nesters in Shanxi, China: A structural equation model. Health Qua. Life Outcomes 19 (1), 156 (2021).
Yu-Ting, Y., Miao, Y., Yong-Wei, Y., Qiong, Y. & Ting, L. Relationships between children-related factors, basic psychological need satisfaction, and multiple happiness among urban empty-nesters in China: A structural equation modeling. BMC Geriatr. 22 (1), 925 (2022).
Ryan, R. M. & Deci, E. L. On happiness and human potentials: A review of research on hedonic and eudaimonic well-being. Annu. Rev. Psychol. 52 (1), 141–166 (2001).
Conte, G. et al. Scrolling through adolescence: a systematic review of the impact of TikTok on adolescent mental health. Eur. Child Adolescent Psychiatr. 34 (5), 1511–1527 (2025).
Rogowska, A. M. & Cincio, A. Procrastination mediates the relationship between problematic TikTok use and depression among young adults. J. Clin. Med. 13 (5), 1247 (2024).
Liao, S. H., Widowati, R. & Lee, C. Y. Data mining analytics investigation on TikTok users’ behaviors: social media app development. Library Hi Tech. 42 (4), 1116–1131 (2024).
Zhang, S. & Song, J. An empirical investigation into the preferences of the elderly for user interface design in personal electronic health record systems. Front. Digit. Health 5, 1289904 (2024).
Mehrabian, A. & Russell, J. A. An Approach to Environmental Psychology (The MIT Press, UK, 1974).
Alam, M. M. D. & Noor, N. A. M. The relationship between service quality, corporate image, and customer loyalty of generation Y: An application of SOR paradigm in the context of superstores in Bangladesh. SAGE Open 10 (2), 2158244020924405 (2020).
Sultan, P., Wong, H. Y. & Azam, M. S. How perceived communication source and food value stimulate purchase intention of organic food: An examination of the stimulus-organism-response (SOR) model. J. Clean. Prod. 312, 127807 (2021).
Jacoby, J. Stimulus-organism-response reconsidered: An evolutionary step in modeling (consumer) behavior. J. Consum. Psychol. 12 (1), 51–57 (2002).
Vafaei-Zadeh, A., Nikbin, D., Wong, S. L. & Hanifah, H. Investigating factors influencing AI customer service adoption: An integrated model of stimulus–organism–response (SOR) and task-technology fit (TTF) theory. Asia Pac. J. Mark. Logist. 37 (6), 1465–1501 (2024).
Shi, R., Wang, M., Liu, C. & Gull, N. The influence of short video platform characteristics on users’ willingness to share marketing information: Based on the SOR model. Sustainability 15 (3), 2448 (2023).
Jiang, S., Wang, Z., Sun, Z. & Ruan, J. Determinants of buying produce on short-video platforms: The impact of social network and resource endowment—Evidence from China. Agriculture 12 (10), 1700 (2022).
Wu, T., Jiang, N., Sharif, S. P. & Chen, M. Explaining subscription intention for video streaming platforms in China: Integrating the UTAUT2 model, perceived value theory, and SOR theory. PLoS One 20 (5), e0322860 (2025).
Mou, X., Xu, F. & Du, J. T. Examining the factors influencing college students’ continuance intention to use short-form video APP. Aslib J. Inf. Manag. 73 (6), 992–1013 (2021).
Wang, S., Esperança, J. P. & Wu, Q. Effects of live streaming proneness, engagement and intelligent recommendation on users’ purchase intention in short video community: Take TikTok (DouYin) online courses as an example. Int. J. Human Comput. Interact. 39 (15), 3071–3083 (2023).
Xu, J., Ramayah, T., Arshad, M. Z., Ismail, A. & Jamal, J. Decoding digital dependency: Flow experience and social belonging in short video addiction among middle-aged and elderly Chinese users. Telematics Inform. 96, 102222 (2025).
Hu, H. & Huang, M. How stress influences short video addiction in China: An extended compensatory internet use model. Front. Psychol. 15, 1470111 (2024).
Lozoya, S. V. M., Guirado, M. A. Z., González, A. Z. & López, A. B. M. Use of technologies and self-efficacy in older adults. IEEE Revista Iberoamericana de Tecnol. Del Aprendizaje 17 (2), 125–130 (2022).
Shin, H. R., Kim, S. K. & Kim, Y. S. Effect of self-efficacy of middle-aged and elderly on the intention to use digital health devices: Focusing on the difference between middle-aged and elderly. J. Digit. Converg. 18 (10), 13–22 (2020).
Chen, Y. & Gao, Q. Effects of social media self-efficacy on informational use, loneliness, and self-esteem of older adults. Int. J. Human-Comput. Interact. 39 (5), 1121–1133 (2023).
Zhao, S., Kinshuk, Y. Y. & Ya, N. Adoption of mobile social media for learning among Chinese older adults in senior citizen colleges. Educ. Technol. Res. Dev. 69, 3413–3435 (2021).
Liu, C., Jiang, M. & Muhammad, Z. A. The impact of TikTok short video factors on tourists’ behavioral intention among Generation Z and Millennials: The role of flow experience. PLoS ONE 19 (12), e0315140 (2024).
Chen, C. C. & Lin, Y. C. What drives live-stream usage intention? The perspectives of flow, entertainment, social interaction, and endorsement. Telematics Inform. 35 (1), 293–303 (2018).
Sallaku, R. & Vigolo, V. Predicting customer loyalty to Airbnb using PLS-SEM: The role of authenticity, interactivity, involvement and customer engagement. TQM J. 36 (5), 1346–1368 (2024).
Xi, N. & Hamari, J. Shopping in virtual reality: A literature review and future agenda. J. Bus. Res. 134, 37–58 (2021).
Negash, S., Ryan, T. & Igbaria, M. Quality and effectiveness in web-based customer support systems. Informat. Manag. 40 (8), 757–768 (2003).
Lo, M. C., Mohamad, A. A., Wan Ibrahim, W. H. & Thong, J. Z. Unveiling the Dynamics of Malaysian Viewers’ Continuance Usage Intention of OTT Platforms: Insights from the Stimulus-Organism-Response (SOR) Paradigm (Memory and Communication. Advance online publication, 2024).
Baccarne, B., Evens, T. & Schuurman, D. The television struggle: An assessment of over-the-top television evolutions in a cable dominant market. Commun. Strateg. 92, 43–61 (2013).
Jiang, S., Zhang, Z., Xu, H. & Pan, Y. What influences users’ continuous behavioral intention in cultural heritage virtual tourism: Integrating experience economy theory and stimulus–organism–response (SOR) model. Sustainability 16 (23), 10231 (2024).
Lundblad, J. P. A review and critique of Rogers’ diffusion of innovation theory as it applies to organizations. Organ. Dev. J. 21 (4), 50–64 (2003).
Lu, X. J. & Hsiao, K. L. Effects of diffusion of innovations, spatial presence, and flow on virtual reality shopping. Front. Psychol. 13, 941248 (2022).
Khan, A. J. et al. Adoption of sustainability innovations and environmental opinion leadership: A way to foster environmental sustainability through diffusion of innovation theory. Sustainability 14 (21), 14547 (2022).
Islam, A. K. M. N. The role of perceived system quality as educators’ motivation to continue E-learning system use. AIS Trans. Human-Comput. Interact. 4 (1), 25–43 (2012).
Montag, C., Yang, H. & Elhai, J. D. On the psychology of TikTok use: A first glimpse from empirical findings. Front. Public Health 9, 641673 (2021).
De Leyn, T., De Wolf, R., Vanden Abeele, M. & De Marez, L. In-between child’s play and teenage pop culture: Tweens, TikTok & privacy. J. Youth Stud. 25 (8), 1108–1125 (2022).
Rogers, E. M. Diffusion of Innovations 5th edn. (Free Press, 2003).
Jiang, Q., Zhang, J., Wang, P. H. & Zhang, Y. Technology acceptance and innovation diffusion: Are users more inclined toward aigc-assisted design?. Int. J. Human-Comput. Interact. 41 (13), 8370–8384 (2025).
Lin, K. Y. & Huang, T. K. Shoping in the digital world: How augmented reality mobile applications trigger customer engagement. Technol. Soc. 77, 102540 (2024).
Kamboj, S. & Sharma, M. Social media adoption behaviour: Consumer innovativeness and participation intention. Int. J. Consum. Stud. 47 (2), 523–544 (2023).
Martínez-Alcalá, C. I. et al. The effects of Covid-19 on the digital literacy of the elderly: norms for digital inclusion. In Frontiers in Education (eds Martínez-Alcalá, C. I. et al.) (Frontiers Media, 2021).
Song, Y., Yang, Y. & Cheng, P. The Investigation of adoption of voice-user interface (VUI) in smart home systems among Chinese older adults. Sensors 22 (4), 1614 (2022).
Weimann, G. & Masri, N. Research note: Spreading hate on TikTok. Stud. Conflict Terror. 46 (5), 752–765 (2023).
Qader, G., Shahid, Z. A., Junaid, M., Shaikh, I. M. & Qureshi, M. A. The role of diffusion of innovation theory towards the adoption of halal meat supply chain. J. Islamic Market. 14 (5), 1211–1228 (2023).
Folorunso, O., Vincent, R. O., Adekoya, A. F. & Ogunde, A. O. Diffusion of innovation in social networking sites among university students. Int. J. Comput. Sci. Secur. 4 (3), 361–372 (2010).
Jiang, Y., Wang, X. & Yuen, K. F. Augmented reality shopping application usage: The influence of attitude, value, and characteristics of innovation. J. Retail. Consum. Serv. 63, 102720 (2021).
Almaiah, M. A. et al. Measuring institutions’ adoption of artificial intelligence applications in online learning environments: Integrating the innovation diffusion theory with technology adoption rate. Electronics 11 (20), 3291 (2022).
Keller, J., Ringelhan, S. & Blomann, F. Does skills–demands compatibility result in intrinsic motivation? Experimental test of a basic notion proposed in the theory of flow-experiences. J. Posit. Psychol. 6 (5), 408–417 (2011).
Venkatesh, V., Thong, J. Y. L. & Xu, X. Consumer acceptance and use of information technology: Extending the unified theory of acceptance and use of technology. MIS Q. 36 (1), 157–178 (2012).
Lu, Y. et al. Research on designing context-aware interactive experiences for sustainable aging-friendly smart homes. Electronics 13 (17), 3507 (2024).
Wiangkham, A. & Vongvit, R. Exploring the drivers for the adoption of metaverse technology in engineering education using PLS-SEM and ANFIS. Educ. Inf. Technol. 29 (6), 7385–7412 (2024).
Li, S., Yin, Y. & Yunus, M. M. Elderly people in the digital cultural environment: Social media use, environmental stress, and intergenerational interaction. Environ. Behav. Proc. J. 10 (31), 433–439 (2025).
Hu, L., Wang, H., & Xin, Y. Factors influencing Chinese pre-service teachers’ adoption of generative AI in teaching: An empirical study based on UTAUT2 and PLS-SEM. Education and Information Technologies, 1–23 (2025).
Yaqub, M. Z., Badghish, S., Yaqub, R. M. S., Ali, I., & Ali, N. S. Integrating and extending the SOR model, TAM and the UTAUT to assess M-commerce adoption during COVID times. J. Econ. Administrat. Sci. (2024).
Csikszentmihalyi, M. Flow and the Foundations of Positive Psychology: The Collected Works of Mihaly Csikszentmihalyi (Springer, 2014).
Hyun, H., Thavisay, T. & Lee, S. H. Enhancing the role of flow experience in social media usage and its impact on shop**. J. Retail. Consum. Serv. 65, 102492 (2022).
Lu, T., Zhou, B. & Wang, Y. Exploring Chinese users’ acceptance of instant messaging using the theory of planned behavior, the technology acceptance model, and the flow theory. Comput. Hum. Behav. 25 (1), 29–39 (2009).
Chen, C. C., Liu, C. C., Chiu, T. H., Lee, Y. W. & Wu, K. C. Role of perceived ease of use for augmented reality app designed to help children navigate smart libraries. Int. J. Human-Comput. Interact. 39 (13), 2606–2623 (2023).
Goh, T. T. & Yang, B. The role of e-engagement and flow on the continuance with a learning management system in a blended learning environment. Int. J. Educ. Technol. High. Educ. 18 (1), 49 (2021).
Liang, Y., Zhang, X., Wang, H. & Liu, M. Users’ willingness to adopt metaverse drawing on flow theory: An empirical study using PLS-SEM and FsQCA. Heliyon 10 (13), e33394 (2024).
Fang, Y. & Chung, Y. Interplay of immersion, absorption, and flow in virtual reality: Based on experience economy theory and flow theory. J. Glob. Fash. Market. 16 (2), 212–230 (2025).
Xia, Y. et al. Digital art exhibitions and psychological well-being in Chinese Generation Z: An analysis based on the S-O-R framework. Humanit. Soc. Sci. Commun. 11, 266 (2024).
Britton, E., Kindermann, G., Domegan, C. & Carlin, C. Blue care: A systematic review of blue space interventions for health and wellbeing. Health Promot. Int. 35, 50–69 (2020).
Lu, P., Sani, N. M., Li, Y. & Wang, Y. How does urban blue space affect human well-being? A study based on the stimulus-organism-response theory. Front. Psychol. 16, 1553296 (2025).
Wang, J. et al. Exploring the impacts of urban community leisure on subjective well-being during COVID-19: A mixed methods case study. Int. J. Environ. Res. Public Health 19 (14), 8514 (2022).
Attiq, S., Hamid, A. B. A., Shah, H. J., Khokhar, M. N. & Shahzad, A. Impact of brand hate on consumer well-being for technology products through the lens of stimulus organism response approach. Front. Psychol. 13, 946362 (2022).
Elshaer, I. A., AlNajdi, S. M. & Salem, M. A. Sustainable AI solutions for empowering visually impaired students: The role of assistive technologies in academic success. Sustainability 17 (12), 5609 (2025).
Etemad-Sajadi, R. The impact of online real-time interactivity on patronage intention: The use of avatars. Comput. Hum. Behav. 61, 227–232 (2016).
Barrett, A. J., Pack, A. & Quaid, E. D. Understanding learners’ acceptance of high-immersion virtual reality systems: Insights from confirmatory and exploratory PLS-SEM analyses. Comput. Educ. 169, 104214 (2021).
Mustafi, M. A. A. & Hosain, M. S. The role of online advertising on purchase intention of smartphones: Mediating effects of flow experience and advertising value. J. Contempor. Market. Sci. 3 (3), 385–410 (2020).
Amini, M. & Jahanbakhsh Javid, N. A multi-perspective framework established on diffusion of innovation (DOI) theory and technology, organization and environment (TOE) framework toward supply chain management system based on cloud computing technology for small and medium enterprises. Int. J. Informat. Technol.Innovation Adopt. 11, 1217–1234 (2023).
Su, D. N. et al. Identifying must-have factors and should-have factors affecting the adoption of electric motorcycles—A combined use of PLS-SEM and NCA approach. Travel Behav. Soc. 33, 100633 (2023).
Al Breiki, M., Al Abri, A., Al Moosawi, A. M. & Alburaiki, A. Investigating science teachers’ intention to adopt virtual reality through the integration of diffusion of innovation theory and theory of planned behaviour: The moderating role of perceived skills readiness. Educ. Inf. Technol. 28(5), 6165–6187 (2023).
Tuncer, I. The relationship between IT affordance, flow experience, trust, and social commerce intention: An exploration using the SOR paradigm. Technol. Soc. 65, 101567 (2021).
Kang, W. & Shao, B. The impact of voice assistants’ intelligent attributes on consumer well-being: Findings from PLS-SEM and fsQCA. J. Retail. Consum. Serv. 70, 103130 (2023).
Rahman, M. M. & Sadik, N. Nexus of digital financial service adoption, decent work and environmental sustainability: results from SEM-NCA approaches. Discover Sustain. 6 (1), 1–19 (2025).
Hameed, I. et al. Decoding willingness to buy in live-streaming retail: The application of stimulus-organism-response model using PLS-SEM and SEM-ANN. J. Retail. Consum. Serv. 84, 104236 (2025).
Liu, H., Ma, L., & Park, K. S. The effect of the red cultural atmosphere on tourists’ subjective well-being: An application of the stimulus–organism–response model. Curr. Issues Tour. 1–20 (2025).
Hair, J., Hollingsworth, C. L., Randolph, A. B., & Chong, A. Y. L. An updated and expanded assessment of PLS-SEM in information systems research. Industrial management & data systems, 117(3), 442-458 (2017).
Hair, J. F. Jr. et al. Partial Least Squares Structural Equation Modeling (PLS-SEM) Using R: A Workbook (Springer, 2021).
Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y. & Podsakoff, N. P. Common method biases in behavioral research: A critical review of the literature and recommended remedies. J. Appl. Psychol. 88 (5), 879–903 (2003).
Hair, J. F., Risher, J. J., Sarstedt, M., & Ringle, C. M. When to use and how to report the results of PLS-SEM. European business review, 31(1), 2-24 (2019).
Hair, J. F., Ringle, C. M., & Sarstedt, M. PLS-SEM: Indeed a silver bullet. Journal of Marketing theory and Practice, 19(2), 139-152 (2011).
Fornell, C. & Larcker, D. F. Evaluating structural equation models with unobservable variables and measurement error. J. Mark. Res. 18 (1), 39–50 (1981).
Hair, J. & Alamer, A. Partial least squares structural equation modeling (PLS-SEM) in second language and education research: Guidelines using an applied example. Res. Methods Appl. Linguist. 1 (3), 100027 (2022).
Chin, W. W. The partial least squares approach to structural equation modeling. In Modern methods for business research (pp. 295-336). Psychology Press (1998).
Sarstedt, M., Ringle, C. M. & Hair, J. F. Partial least squares structural equation modeling. In Handbook of Market Research (eds Sarstedt, M. et al.) (Springer, 2021).
Di, K., Chen, W., Shi, Q., Cai, Q. & Liu, S. Analysing the impact of coupled domestic demand dynamics of green and low-carbon consumption in the market based on SEM-ANN. J. Retail. Consum. Serv. 79, 103856 (2024).
Alkawsi, A. et al. A hybrid SEM-neural network method for identifying acceptance factors of the smart meters in Malaysia: Challenges perspective. Alex. Eng. J. 60, 227–240 (2021).
Al-Sharafi, A. et al. Determinants of cloud computing integration and its impact on sustainable performance in SMEs: An empirical investigation using the SEM-ANN approach. Heliyon 9, e16299 (2023).
Liébana-Cabanillas, F., Marinković, V. & Kalinić, Z. A SEM-neural network approach for predicting antecedents of m-commerce acceptance. Int. J. Inf. Manag. 37 (2), 14–24 (2017).
Sohaib, O., Hussain, W., Asif, M., Ahmad, M. & Mazzara, M. A PLS-SEM neural network approach for understanding cryptocurrency adoption. IEEE Access 8, 13138–13150 (2019).
Leong, L.-Y., Hew, T.-S., Ooi, K.-B. & Wei, J. Predicting mobile wallet resistance: A two-staged structural equation modeling-artificial neural network approach. Int. J. Inf. Manag. 51, 102047 (2020).
Foo, P. Y., Lee, V. H., Tan, G. W. H. & Ooi, K. B. A gateway to realising sustainability performance via green supply chain management practices: A PLS–ANN approach. Expert Syst. Appl. 107, 1–14 (2018).
Wong, L.-W., Leong, L.-Y., Hew, J.-J., Tan, G.W.-H. & Ooi, K.-B. Time to seize the digital evolution: Adoption of blockchain in operations and supply chain management among Malaysian SMEs. Int. J. Inf. Manag. 52, 101997 (2020).
Taufiq-Hail, G. A. M., Sarea, A. & Hawaldar, I. T. The impact of self-efficacy on feelings and task performance of academic and teaching staff in Bahrain during COVID-19: Analysis by SEM and ANN. J. Open Innovat. Technol. Market Complex. 7 (4), 224 (2021).
Yin, Y. C., Ahmed, J., Nee, A. Y. H. & Hoe, O. K. The rural consumer adoption of sustainable energy: A PLS-SEM-ANN approach of conceptual model development and cross-country validation of Pakistan and Malaysia. Environ. Sci. Pollut. Res. 30 (3), 5881–5902 (2023).
Liébana-Cabanillas, F., Marinkovic, V., De Luna, I. R. & Kalinic, Z. Predicting the determinants of mobile payment acceptance: A hybrid SEM-neural network approach. Technol. Forecast. Soc. Chang. 129, 117–130 (2018).
Bellary, S., Bala, P. K. & Chakraborty, S. Exploring cognitive-behavioral drivers impacting consumer continuance intention of fitness apps using a hybrid approach of text mining, SEM, and ANN. J. Retail. Consum. Serv. 81, 104045 (2024).
Wang, G., Tan, G. W. H., Yuan, Y., Ooi, K. B. & Dwivedi, Y. K. Revisiting TAM2 in behavioral targeting advertising: A deep learning-based dual-stage SEM-ANN analysis. Technol. Forecast. Soc. Chang. 175, 121345 (2022).
Al Karim, R., Kabir, M. R., Rabiul, M. K., Kawser, S. & Salam, A. Linking green supply chain management practices and environmental performance in the manufacturing industry: A hybrid SEM-ANN approach. Environ. Sci. Pollut. Res. 31 (9), 13925–13940 (2024).
Ibrahim, A. N. H. et al. Modelling of passenger satisfaction and reuse intention with monorail services in Kuala Lumpur, Malaysia: A hybrid SEM-ANN approach. Mathematics 11 (15), 3361 (2023).
Nayak, I., Siddhanta, A. & Panda, B. K. Does empty nest elderly experience more depressive symptoms than non-empty nest elderly? Evidence from longitudinal aging study in India. Hosp. Top. 102 (2), 96–109 (2024).
Vivekananthan, K. & Ponnusamy, R. Mental health of the empty nest elderly. In Handbook of Aging, Health and Public Policy: Perspectives from Asia (eds Vivekananthan, K. & Ponnusamy, R.) (Springer, 2023).
Dogruel, L., Joeckel, S. & Bowman, N. D. The use and acceptance of new media entertainment technology by elderly users: Development of an expanded technology acceptance model. Behav. Informat. Technol. 34 (11), 1052–1063 (2015).
Rexfelt, O. & Hiortaf Ornäs, V. Consumer acceptance of product-service systems: Designing for relative advantages and uncertainty reductions. J. Manufact. Technol. Manag. 20, 674–699 (2009).
Bhattacherjee, A., Limayem, M. & Cheung, C. M. User switching of information technology: A theoretical synthesis and empirical test. Informat. Manag. 49, 327–333 (2012).
Liao, X., Wu, D., Zhang, Q. & Han, G. How to improve users’ loyalty to smart health devices? The perspective of compatibility. Sustainability 13 (19), 10722 (2021).
Choi, Y., Wen, H., Chen, M. & Yang, F. Sustainable determinants influencing habit formation among mobile short-video platform users. Sustainability 13 (6), 3216 (2021).
Li, Y., Liu, Q. & Wang, Y. The research on the usage behavior of TikTok short video platform in the elderly group. J. Educ. Human. Soc. Sci. 5, 189–197 (2022).
Hudgins, B. L. & Maher, J. P. Does habit strength moderate affective responses to movement-related behaviors among older adults?. Appl. Psychol. Health Well Being 17 (1), e12643 (2025).
Gumasing, M. J. J. Exploring factors influencing E-bike adoption among Filipino commuters: An integrated diffusion of innovation and technology acceptance model. World Electr. Vehicle J. 16 (2), 113 (2025).
Azman Ong, M. H. & Ibrahim, N. S. Towards the formation of we-intention to continue playing mobile multiplayer games: Importance of gamification design elements and social play habit roles. Inf. Technol. People 38 (3), 1281–1304 (2025).
Acknowledgements
This research was supported by 2026 Zhejiang Provincial Philosophy and Social Science Planning Project“Image Co-creation Empowered by Generative AI: Research on User Interaction Mechanism and Aesthetic Experience” [Grant No. 26NDJC316YB]; and the following funds ”Digital Learning, Psychosocial Orientation and Education for Sustainable Development System Construction” (Project No.: RP/FCHS-01/2023), by Macao Polytechnic Institute.
Funding
Macao Polytechnic University, RP/FCHS-01/2023.
Author information
Authors and Affiliations
Contributions
X.L. Li designed the research framework, analyzed data, interpreted results, wrote and revised the paper, and supervised the research. N. Tang and H.F. Zhang contributed to research design, data collection, and paper improvement. J.P. Xu participated in data—related work, provided technical support, reviewed relevant content,set the research direction, guided the study, and offered theoretical and resource support.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Ethics approval and consent to participate
This study was approved by 2026 Zhejiang Provincial Philosophy and Social Science Planning Project [Grant No. 26NDJC316YB]; and the Macao Polytechnic University’s ethics review body (Approval Number: [No.: RP/FCHS - 01/2023/E01]).
Informed consent statement
Informed consent was obtained from all subjects involved in the study.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Li, X., Xu, J., Tang, N. et al. The Interaction Mechanism of Short Video Platform Usage Experience on the Subjective Well-Being of China's Rural Elderly. Sci Rep 15, 39185 (2025). https://doi.org/10.1038/s41598-025-25279-8
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s41598-025-25279-8
