Social media disclosure intention refers to individuals’ behavioral tendency to reveal personal information on social media platforms. It is a key predictor of actual disclosure behavior online, which has received sustained attention in privacy and communication research. According to Norberg et al. (2007), disclosure intention reflects the motivational component of the privacy calculus, revealing how users evaluate the trade-off between perceived benefits and perceived risks. More recently, scholars have emphasized its role in shaping digital identity and online interaction patterns (Rodríguez-Priego et al., 2023).

A number of factors have been identified as influencing social media disclosure intention. Trust in social media, for example, plays a pivotal role in reducing perceived risks (Rodríguez-Priego et al., 2023) and encouraging information sharing (Omar et al., 2024). When users believe that a platform is secure, reliable, and respects their privacy, they are more likely to engage in disclosure (Nemec Zlatolas et al., 2019). Additionally, individual differences, such as personality traits and risk tolerance, further moderate disclosure intention. Studies have shown that users with a higher propensity for risk-taking or with an extroverted personality may exhibit greater disclosure intentions, whereas those with more conservative traits might be more reserved (Chen et al., 2016). The literature on social media disclosure intention underscores the importance of context factors as well. For instance, Bansal et al. (2016) found that the specific nature of the information being shared and the perceived sensitivity of the context can significantly affect users’ decisions to disclose.

However, there is still an inconsistent understanding of the negative role of privacy concerns on social media disclosure (Cain and Imre, 2021). One stream considers this negative effect to be significant [e.g., Lutz et al. (2018), Gruzd and Hernández-García (2018)], while the other considers it to be insignificant [e.g., Barth et al. (2019), Hughes-Roberts (2013)]. Moreover, its interplay with past experiences with privacy invasion and privacy fatigue in influencing social media disclosure intention needs further clarification.

Behavioral decision theory is a framework for understanding how individuals make decisions under uncertainty and in complex situations (Einhorn and Hogarth, 1981). This theory shows that individual behavioral intention is influenced by a variety of factors, including anchoring, heuristics, and loss aversion. The anchoring effect suggests that individuals rely heavily on initial information when making decisions, and prior encounters with privacy breaches can act as a powerful reference point in shaping subsequent perceptions (Furnham and Boo, 2011). Heuristics are simplified decision-making rules that people use to save cognitive resources and time when facing complex problems. Heuristics help people make decisions quickly, but they can also easily lead to cognitive biases (Dale, 2015). Loss aversion refers to the emotional response that people are more sensitive to losses than to gains of equal value (Novemsky and Kahneman, 2005); that is, people are more willing to avoid losses rather than pursue equal gains. Drawing on insights from psychology and cognitive science, scholars posit that decision-making is influenced by cognitive biases, emotional responses, and past experiences (Beldhuis et al., 2021; Naqvi et al., 2006).

Specifically, past experiences always play a pivotal role in influencing future decisions. Privacy invasion experiences refer to individuals’ past encounters with unauthorized access, misuse, or exposure of their personal information in digital environments (Huang and Leu, 2020). When users experience unauthorized access or misuse of their personal data, these events become salient anchors that influence how they evaluate the risks associated with disclosing further information on social media. Drawing on anchoring theory, previous encounters with privacy invasions function as cognitive anchors, influencing users’ perceived vulnerability and decision-making patterns in similar contexts (Acquisti et al., 2015). Recent research confirms that individuals who have experienced privacy invasions are more likely to exhibit heightened risk awareness and lower platform trust (Chen et al., 2024). These anchoring effects create a cognitive frame through which users assess the safety and reliability of disclosing personal information on social media.

Privacy fatigue is conceptually defined as a psychological state of weariness and resignation that arises when individuals feel overwhelmed by the effort required to manage and protect their personal data online (Chen et al., 2023; Choi et al., 2018). This condition is often triggered by increasing data breaches, insufficient control over personal information, and frequent exposure to complex privacy decisions (Acquisti et al., 2015; Smith et al., 2011). As users encounter frequent privacy invasions, they are often required to engage in cognitively demanding tasks, such as navigating dull privacy policies or configuring complex security settings. For many individuals, protecting privacy becomes an exhausting task. The complexity of safeguarding personal information exacerbates feelings of helplessness, inability and exhaustion, ultimately leading to privacy fatigue. This condition reflects heuristic-driven cognitive biases. When individuals face mental overload, they tend to simplify decision-making processes, often adopting avoidance or withdrawal strategies in relation to social media disclosure (Van Der Schyff et al., 2023). Thus, privacy fatigue is not merely a passive response to privacy invasions; rather, it is also a result of psychological overload shaped by heuristic processing and diminished cognitive ability.

Simultaneously, emotions also influence judgments and choices. Privacy concern is defined as individuals’ worries and anxieties regarding the potential misuse, unauthorized access, or secondary use of their personal information (Wang J. et al., 2025). Privacy concerns reflect the anxiety and apprehension stemming from potential privacy invasions, diminishing the disclosure intention by heightening risk perceptions. Theoretically, it align with the loss aversion principle in behavioral decision theory, which posits that individuals are more motivated to avoid potential losses than to pursue equivalent gains. Empirical evidence shows that higher levels of privacy concern negatively impact users’ willingness to disclose information (Kim et al., 2023). In the digital context, users fear potential losses, such as unauthorized access to personal data or identity theft, which heightens their perceived risks and reduces their willingness to disclose information online (Van der Schyff and Flowerday, 2023). Privacy concerns may translate into restrictive disclosure intentions, as the psychological worry of potential losses outweighs the perceived benefits of online disclosure.

Drawing on insights from behavioral decision theory, we employ three core concepts of behavioral decision theory as the key variables of this study, namely, privacy invasion experiences as the anchor in the anchoring effect, privacy fatigue as the cognitive bias caused by heuristics, and privacy concern as the emotional response of loss aversion, to understand the multi-layered disclosure decision-making process users undergo when confronting privacy issues on social media platforms.

In digital environments, trust is often conceptualized as the belief in the integrity, competence, and benevolence of a platform or service provider (Soleimani, 2022). In the current study, we focus on trust in social media platforms. Trust in social media refers to users’ belief that the platform will act in their interest, particularly by safeguarding personal information, ensuring system reliability, and enabling secure interactions (Bansal et al., 2015). When users perceive that a social media platform is competent, ethical, and benevolent in its data handling practices, they are more likely to engage with it and disclose personal information. This trust is built upon users’ assessments of a platform’s past performance and reliability, as well as its perceived commitment to user welfare (Xie et al., 2024).

Drawing on trust theory, this study introduces trust as a central explanatory mechanism. On one hand, it explores how users’ privacy invasion experiences, privacy fatigue, and privacy concerns affect their trust in social media. Prior studies indicated that trust in social media is impacted by multiple factors (Balapour et al., 2020; Bansal et al., 2016; Krishna et al., 2023). On the other hand, it examines how such trust, in turn, influences their intention to disclose on social media platforms. In addition, trust often acts as a mediator between users’ privacy concerns and their willingness to disclose personal information (Rodríguez-Priego et al., 2023; Zhang, 2024). Even when users have strong privacy concerns, a high level of trust in a platform may encourage online disclosure, if they perceive the benefits of disclosure to outweigh the risks.

In summary, this study integrates behavioral decision theory and trust theory, offering a robust framework for understanding social media disclosure intention. Our model operationalizes privacy invasion experiences as an anchor that captures the historical context of users’ encounters with privacy invasions, influencing both their cognitive and emotional responses. Privacy fatigue, as a cognitive bias factor, represents the overload from frequent privacy invasions, which may affect users’ trust and disclosure in social media. Privacy concerns, reflecting the emotional reaction to potential breaches, are supposed to impact trust and disclosure intention. Drawing on trust theory, trust in social media embodies users’ belief in the platform’s good intention, mediating the relationship between the above privacy-related factors and the disclosure intention on social media.

This study utilized an online survey distributed via WJX.com, and employed a network sampling technique. To ensure the reliability and clarity of the questionnaire, a pilot test was conducted prior to formal data collection. Participation in the survey was entirely voluntary and anonymous. All respondents were informed about the academic purpose of the study and assured of data confidentiality. They retained the right to withdraw from the survey at any time. The data collection was carried out in mainland China from June to December 2024. The target population consisted of Chinese social media users aged 20 to 59 years. Adolescents and older adults were excluded from the sample due to their potentially distinct usage patterns of social media (Lei et al., 2024).

This study applied the network sampling technique, a method increasingly adopted in studies focusing on interconnected digital communities (Baltar and Brunet, 2012). In this approach, initial respondents were encouraged to disseminate the survey link within their own social networks. To address potential sampling bias due to homophily in network sampling (Heckathorn and Cameron, 2017), we used demographic benchmarks from the CNNIC (2024) to monitor and guide the sampling process. Key demographic indicators such as age, gender, and residence were continuously assessed. If discrepancies in sample composition emerged, particularly with regard to underrepresented age or gender groups, targeted data supplementation was implemented to improve representativeness.

Our study initially collected a total of 1,000 responses. After removing responses with unusually short completion times, duplicate IP addresses, and highly uniform answer patterns (i.e., over 70% identical responses), 787 valid responses remained for analysis. Descriptive statistics of the final sample are presented in Table 1. The majority of participants were under the age of 39, with the 30–39 age group comprising the largest proportion (37.4%). The gender distribution was balanced (52.7% male and 47.3% female). In terms of residence, 44.9% of respondents lived in provincial capital cities. Additionally, 36.6% of participants reported spending more than 5 h per day on social media. The sample is broadly representative of Chinese social media users.

Our study includes five constructs, namely privacy invasion experiences, privacy fatigue, privacy concerns, trust in social media, and social media disclosure intention. We adapted the measurement items for these constructs from well-established scales in prior literature and localized them to fit our specific research context. Table 2 shows the operationalized definition of constructs. We adapted the items of privacy invasion experiences in Masur and Trepte (2021) to fit the research topic of this study. The measurement items for privacy fatigue were adapted from Choi et al. (2018), with modifications to align with the social media environment. Privacy concerns were measured using an adapted version of the scale proposed by Xu et al. (2008), ensuring its relevance to users’ privacy perceptions on social media platforms. The trust in social media construct was derived from Lo and Cindy (2010), incorporating elements that reflect trust in the context of social media interactions. Lastly, social media disclosure intention was measured using items from Zhu et al. (2022) and Fan et al. (2021), assessing users’ willingness to share personal information on social media. The measurement items of constructs are shown in Appendix A. All constructs were measured using a five-point Likert scale ranging from “strongly disagree” (1) to “strongly agree” (5). To ensure the validity and reliability of the measurement instruments, these indicators were validated by an expert panel. Additionally, a pilot test was conducted with a sample of social media users to evaluate the measurement items in advance. The pilot test results confirmed that all constructs exhibited good internal consistency and convergent validity, supporting their use for further empirical analysis.

In addition, to assess potential common method variance, a variable unrelated to this study, namely Neighborhood Aesthetics (Carlson et al., 2012), was incorporated as a marker variable. This construct consists of four items and is also measured using a five-point Likert scale. These items were presented alongside the main variables within the questionnaire.

To minimize potential common method variance (CMV) due to data collected from a single questionnaire, this study adopted both procedural design and post hoc detection. At the design stage, we adopted a variety of techniques, including ensuring anonymity and reducing item ambiguity, and shuffling the order of questions. In post hoc, two statistical approaches were employed to assess the presence of CMV, Harman’s single-factor test and the marker variable technique. Firstly, we conducted Harman’s single-factor test. We performed an unrotated principal component analysis in SPSS, including all items from the measurement scales. The results indicated that the first factor load of an interpretable variable accounted for only 29.5% of the total variance, which is well below the 50% threshold recommended by Eichhorn (2003), suggesting that CMV is not a significant concern.

Since Harman’s test is considered a relatively weak diagnostic tool for CMV (Fuller et al., 2016), we further applied the PLS marker variable technique, as suggested by Chin et al. (2013) and Simmering et al. (2015). Regarding the selection of the marker variable, prior research suggests that as few as four items can effectively detect and control over 70% of CMV (Chin et al., 2013). We selected “Neighborhood Aesthetics” from Carlson et al. (2012) as the marker variable since it comprises four items and has no conceptual or theoretical relationship with any of the variables in our study. The marker variable was incorporated into the baseline model and linked to all observed variables as a latent method factor. We then used SmartPLS 4 to compare changes in R² values and path coefficients between the baseline model and marker model (including the marker construct). As shown in Tables 3, 4, the introduction of the marker variable resulted in changes of less than 10% in path coefficients and less than 0.2 in R², consistent with the threshold suggested by Chin et al. (2013). These findings confirm that CMV does not pose a substantial threat in this study.

The measurement model assessment demonstrated satisfactory reliability and validity. The Cronbach’s α (CA) for all constructs ranged from 0.743 to 0.846, while the composite reliability (CR) values ranged from 0.854 to 0.890, indicating strong internal consistency and reliability (Hair et al., 2021). Additionally, the outer loadings of all indicators exceeded the recommended threshold of 0.7, confirming adequate indicator reliability. The AVE for all constructs was above the threshold of 0.5, with values ranging from 0.588 to 0.664, suggesting satisfactory convergent validity (Fornell and Larcker, 1981). The results are presented in Table 5. To assess discriminant validity, the Heterotrait-Monotrait (HTMT) ratio technique was applied (Henseler et al., 2015). As shown in Table 6, all HTMT values were below the critical threshold of 0.85, confirming that discriminant validity was established.

After establishing the reliability and validity of the research instrument, we employed a bootstrapping procedure with 5,000 resamples using the one-tailed test to estimate the significance of the path coefficients (Hair et al., 2021). First, we assessed the VIF of the structural model, as path coefficients between constructs are calculated based on regression analysis. This step ensures that multicollinearity does not distort the regression results. The VIF values of all predictors did not exceed the recommended threshold of 3.3 (Hair et al., 2021), indicating that multicollinearity was not a concern (see Table 7). Next, we evaluated the structural model using key statistical indicators, including path coefficients (β), t-values, p-value, confidence interval, effect size (f²), predictive relevance (Q²), and the coefficient of determination (R²). The results provide insights into the relationships between constructs and the explanatory power of the structural model.

The Importance-Performance Map Analysis (IPMA) provides insights beyond traditional path coefficient analysis by considering both importance (total effects) and performance (average latent variable scores). It was shown to be of great use in explaining which factors have the greatest impact on the target construct and which improvements would be most effective (Streukens et al., 2017). We conducted IPMA to further investigate whether the positive effect of trust in social media on disclosure intention outweighs the negative effects of privacy-related factors (privacy invasion experiences, privacy fatigue, and privacy concerns). Additionally, this analysis helps to determine which of the negative factors exert the greatest influence on social media disclosure intention (Figure 3).

Table 9 reveals that trust in social media is the most influential positive factor (importance = 0.323), surpassing all negative factors. This reinforces its role as a key driver in users’ online behavior. Privacy concerns exert the strongest negative impact (importance = −0.312), indicating that users who perceive high privacy risks are significantly less likely to disclose personal information. Furthermore, privacy fatigue exhibits the lowest importance, suggesting that while some users feel overwhelmed or resigned regarding privacy risks, this fatigue alone does not substantially deter disclosure compared to other factors. Although trust in social media demonstrates the highest importance, it records the lowest performance score (performance = 71.146), indicating that users currently exhibit a relatively low level of trust in these platforms. In other words, cultivating users’ trust is not only critical but also presently underdeveloped. The gap between the high importance and low performance of trust in social media underscores that enhancing users’ trust could substantially improve their intention to disclose on social media. This finding offers practical insights for social media platforms that strengthening users’ trust may prove to be a more effective strategy than merely addressing negative factors such as privacy fatigue or privacy concerns.

This study advances the theoretical development of social media privacy research by integrating behavioral decision theory and trust theory into the context of social media disclosure. While prior research has predominantly conceptualized disclosure decisions as rational cost–benefit calculations, behavioral decision theory emphasizes the influence of psychological heuristics such as anchoring effects, cognitive biases, and loss aversion (Gilovich et al., 2002; Tversky and Kahneman, 1974). Drawing from this framework, this study conceptualizes privacy invasion experiences as anchoring events that serve as cognitive reference points for future disclosure decisions. Privacy fatigue is interpreted as a manifestation of cognitive overload resulting from the continual demands of privacy management, and privacy concerns are framed as expressions of loss aversion, reflecting users’ fear of potential harm from personal information disclosure.

By applying these constructs to the digital disclosure context, this study extends the application of behavioral decision theory beyond its traditional domains in economics and consumer behavior to the emerging field of digital privacy. Simultaneously, the incorporation of trust theory allows for the examination of trust as a psychological mechanism that mediates the relationship between privacy-related emotional responses and users’ disclosure intentions. Findings demonstrate that trust in social media significantly mediates the effects of privacy concerns and privacy invasion experiences on social media disclosure intention, highlighting its pivotal role as a psychological bridge between emotional perception and behavioral intention.

Using IPMA, this study compares the positive and negative factors influencing social media disclosure intention and finds that trust in social media emerges as the most influential positive predictor, surpassing all negative factors, including privacy concerns, which exert the strongest negative effect. However, we argue that the positive effect of trust do not offset the negative influence of privacy concerns. This is because these two constructs operate through distinct psychological mechanisms. Trust in social media functions as emotional reassurance rooted in perceived credibility and platform reliability, whereas privacy concerns stem from evaluations of risk and control. They influence user behavior in parallel but non-compensatory ways. This finding enriches the theoretical understanding of the trust-privacy concern framework.

Furthermore, this study contributes novel insights by introducing privacy fatigue as a predictive factor. Although prior literature has recognized privacy fatigue as a product of sustained privacy threats (Chen et al., 2023), its role in shaping trust and disclosure intention remains underexplored. By establishing privacy fatigue as a negative predictor of disclosure intention, this study sheds light on how emotional exhaustion impacts digital behaviors. In an era of increasingly complex digital privacy challenges, the findings offer a new theoretical lens to understand the psychological mechanisms linking privacy and trust to disclosure on social media.

This study offers actionable insights for platform designers, users, and policy makers aiming to foster a more secure and trust-oriented social media environment. First, the negative impacts of privacy fatigue and privacy concerns on users’ disclosure intention underscore the urgency for platforms to reduce privacy-related burdens and increase transparency in data practice. By simplifying privacy controls and minimizing users’ management load, platforms can mitigate privacy fatigue and privacy concerns to restore users’ intention to disclose on social media.

Given that trust in social media emerged as the strongest predictor of disclosure intention with a low performance, platforms must prioritize trust-building of users. Specifically, transparent data practices, verified security certifications, and robust user feedback mechanisms should be implemented to enhance users’ perceived reliability of the platform. Importantly, trust-building should extend beyond system design to encompass user-oriented communication policies. Educating users about how their data is protected, and offering personalized privacy settings, can enhance users’ trust while respecting individual comfort levels with disclosure, which should be put a premium on by policy makers and platform designers. For users, psychological interventions such as mindfulness training (Gu et al., 2022) and the improvement of online privacy literacy (Neves et al., 2024) may contribute to reducing users’ privacy fatigue and alleviating their concerns about online disclosure. As privacy challenges continue to evolve, platform strategies that address users’ emotional and psychological privacy problems will be better positioned to foster sustainable user relationships in digital environments.

Although the contribution of this study is clear, it still has limitations for future research to explore. First, while this research aims to explain how individuals make disclosure decisions under privacy uncertainty, it does not account for broader social, cultural, or personality-related factors that may influence user behavior. Factors such as individual risk tolerance, cultural values, and personality traits (e.g., openness or neuroticism) may moderate users’ privacy perceptions and trust formation. Future research could integrate these factors to develop a more comprehensive understanding of privacy-related decision-making. Second, the study treats social media as a homogeneous environment, without differentiating between platform-specific characteristics. However, different platforms (e.g., TikTok, WeChat, Weibo) have varying privacy settings, affordances, and user interaction patterns, which may influence both trust and disclosure behaviors. Future research should consider platform-level variables or conduct comparative studies across different social media environments. Third, methodological limitations must be acknowledged. This study employs a cross-sectional design and network sampling, which constrain the ability to draw causal inferences and generalize findings beyond the current sample. Longitudinal or experimental designs would provide stronger evidence of causal relationships among privacy experiences, trust, and disclosure intention and allow for continued investigation of the trust-privacy concern relationship. Additionally, expanding sampling strategies to include a more diverse population could enhance the external validity of future studies. Finally, since privacy norms, risk perceptions, and trust formation are deeply influenced by cultural and regulatory environments (Wang et al., 2020), the findings may not be generalizable to other cultural contexts. Future research should adopt a cross-cultural comparative approach to explore how cultural and institutional differences influence the privacy-trust-disclosure process in different regions.

This study integrates behavioral decision theory and trust theory to examine how privacy invasion experiences, privacy fatigue, and privacy concerns influence users’ trust and disclosure intention on social media. Results from a sample of Chinese users reveal that while all three privacy-related factors negatively affect disclosure intention, trust in social media serves as the strongest positive predictor and mediates the effects of both privacy invasion experiences and privacy concerns.

By conceptualizing privacy fatigue and privacy concerns as cognitive and emotional responses, and identifying trust as a central psychological mechanism, this study expands the theoretical application of behavioral decision theory in digital privacy contexts. The inclusion of privacy fatigue offers deeper insight into how emotional exhaustion influences online disclosure behavior. These findings contribute to understanding the psychological processes underlying online disclosure and provide practical implications for designing trustworthy and privacy-sensitive platforms.