Cyber upward social comparison and well-being among college students: the chain mediating roles of self-esteem and emotional regulation

Abstract

Background

With the rapid rise of social media, college students are increasingly exposed to others’ “highlight moments,” making cyber upward social comparison more common and potentially detrimental to Well-being. This study explores how cyber upward social comparison affects well-being, focusing on self-esteem and emotional regulation as mediators, and examining gender differences.

Methods

A questionnaire survey yielded 500 valid responses, all from Chinese college students. The participants’ ages were primarily concentrated between 18 and 20 years old (accounting for 70.80%). Among them, 234 were male (46.8%) and 266 were female (53.2%). Data were analyzed with Smart-PLS for SEM and multi-group analysis to examine gender differences.

Results

SEM results showed that the model explained 60.8% of the variance in well-being. Self-esteem and cognitive reappraisal sequentially mediated the relationship between cyber upward social comparison and well-being, but expressive suppression showed no significant association. Multi-group analysis revealed a stronger negative association between cyber upward social comparison and cognitive reappraisal in females, while the positive association between cognitive reappraisal and well-being was more pronounced in males.

Conclusion

Cyber upward social comparison reduces college students’ well-being both directly and through a chain of lowered self-esteem and reduced cognitive reappraisal. Gender differences suggest the need for gender-sensitive psychological interventions. This study provides new insights into college students’ psychological adaptation in digital environments, with the goal of offering both theoretical support and practical recommendations for mental health promotion in higher education.

Introduction

Well-being (WB) refers to individuals’ positive, satisfying, and pleasurable subjective experiences regarding their life circumstances [1]. College students, in the transition to adulthood, face academic demands, career choices, interpersonal challenges, and identity construction, making their WB especially vulnerable [2]. Recent surveys show that an increasing number of Chinese college students report low WB, underscoring its importance for mental health and social adaptation [3]. Reduced WB not only increases the risk of anxiety and depression but also undermines academic success, social competence, and career development [4–6]. Therefore, examining the factors that shape WB is of critical significance.

The formation of WB results from the interaction of external and internal factors. Externally, social support, family dynamics, and financial conditions provide essential environmental foundations [7–9]. Among socio-psychological processes, cyber upward social comparison (CUSC) has become increasingly salient, reflecting how individuals evaluate themselves in digital environments and shaping their WB [10]. Internally, psychological resources such as resilience, self-esteem (SE), and emotional regulation (ER) enable individuals to cope with stress and challenges [11–13]. These positive resources stabilize emotions, sustain self-evaluations, and buffer the negative effects of stressors on WB, while low SE has been identified as a risk factor for anxiety and ER difficulties [14, 15].

Although research on college students’ WB has been growing, several gaps remain. First, existing studies have mainly examined traditional factors such as emotional support, personality traits, and financial situation, while paying little attention to how digital social cues, particularly upward social comparisons, shape WB [16]. Second, many works rely on narrow variable selection and linear analytical frameworks, which limits understanding of the dynamic interplay among cognitive, emotional, and social processes in online contexts. Third, most theoretical models have been developed in Western settings, whereas research focusing on Chinese college students is still limited [17]. Considering that Chinese culture stresses collectivism, interpersonal harmony, and sensitivity to face and social evaluation, these traits may magnify negative emotional responses to CUSC, underscoring the importance of culturally grounded investigations.

In digital environments, face culture heightens individuals’ attention to others’ displays of success, which can easily lead to self-deprecation and emotional distress, thereby exerting distinctive effects on SE and ER compared with Western contexts [18]. Regarding gender socialization, Chinese society places different expectations on men and women in terms of emotional expression and social roles. Studies show that female college students are more likely to engage in appearance-based social comparisons and report higher levels of body dissatisfaction than males, which increases their vulnerability in CUSC situations [19, 20]. Other research further indicates that portrait-focused social media images intensify upward comparisons and online social anxiety, with gender differences evident in emotional impact pathways [21, 22]. Furthermore, empirical studies in Chinese samples have directly shown that collectivism strengthens individuals’ sensitivity to peer evaluation, thereby exacerbating the negative impact of upward social comparison on emotional well-being [23]. Similarly, research on face culture has demonstrated that Chinese college students with higher face concern report greater social comparison anxiety and lower self-esteem in online contexts [24]. These findings highlight that collectivism, face culture, and gender socialization jointly shape Chinese students’ emotional responses and regulation in CUSC contexts. Therefore, it is necessary to examine how these cultural factors influence the mechanisms linking CUSC, SE, ER, and WB.

To address these gaps, this study investigates the impact of CUSC on WB among contemporary college students in online environments. By doing so, it expands the scope of WB research in three ways. First, it emphasizes students’ emotional responses and psychological adaptation in digital contexts, underscoring the influence of new media on mental states and extending WB research into a more era-specific domain. Second, it incorporates SE and ER as key internal resources, examining how individuals cognitively appraise and emotionally regulate external information, thereby deepening understanding of the psychological mechanisms underlying WB. Third, by situating the analysis within the Chinese cultural context, this study increases cultural relevance and offers empirical evidence that strengthens the practical applicability of its findings. It also provides theoretical support and guidance for mental health education and intervention programs in Chinese universities.

In summary, this study examines how CUSC influences WB through the mediating role of SE and the regulatory function of ER. By exploring these mechanisms, it not only extends the theoretical boundaries of WB research in digital environments but also demonstrates, through a Chinese cultural lens, that collectivism, face culture, and gender socialization shape sensitivity to social comparison and regulation strategies. The findings have practical implications for mental health education in higher education, particularly in enhancing students’ SE and ER and mitigating the negative emotional effects of online social interactions.

Theoretical framework and hypotheses

Theoretical foundations

According to Social Comparison Theory (SCT) [25], individuals evaluate themselves through comparisons with others when no objective standards are available. This theory identifies social comparison as a fundamental cognitive process that helps individuals assess their abilities, social standing, and self-worth. As social media has grown, the relevance of this theory has been further amplified: digital platforms selectively present highly curated content, exposing users predominantly to others’ “highlight moments,” thereby increasing both the frequency and intensity of upward social comparison [26]. As proposed by the framework of SCT, making comparisons with superior others can lead to two varied mental consequences. On one hand, perceiving others’ advantages as a threat may elicit negative emotions, lower SE, and undermine WB [27]. On the other hand, if individuals interpret the comparison target as a role model, upward comparison may stimulate learning motivation and positive emotions, producing a motivational association [28]. SCT thus offers a key theoretical foundation for understanding how online social comparison behaviors are associated with college students’ psychological states. It has been widely applied in research on SE, self-identity, adolescent development, and the psychological correlates of social media use [29, 30].

The Emotion Regulation Model (ERM), proposed by Gross [31], systematically explains how individuals regulate and manage their emotional experiences through specific strategies. The model emphasizes that ER is not a passive response but an active, goal-directed process initiated by individuals in specific contexts. Cognitive reappraisal (CR) and expressive suppression are two prominent strategies [32]. CR involves altering one’s perception in ways that are associated with reduced emotional intensity, and is generally regarded as a positive ER strategy. Unlike CR, expressive suppression refers to holding back emotional expressions to manage feelings, which may offer short-term benefits but can harm psychological health over time [33]. This model elucidates the internal mechanisms by which individuals cope with emotional stimuli and have been widely applied in studies on anxiety, depression, stress coping, and WB [34, 35]. It provides a framework to study how individuals manage their emotions during CUSC.

Integrating SCT and the ERM facilitates the construction of a systematic framework that captures the pathway from external social information stimuli to internal psychological responses and ultimately to WB. Specifically, when college students engage in CUSC on online platforms, their self-evaluation systems are first activated, potentially leading to fluctuations in SE and the emergence of negative emotional experiences. Subsequently, the ER strategies employed by individuals determine the trajectory of their emotional responses and their subsequent impact on WB. CR may help transform comparison situations into opportunities for self-improvement, whereas expressive suppression may lead to emotional inhibition and diminished WB [36, 37]. On this basis, the present study further introduces SE and ER strategies as chain mediators to enhance the explanatory power of the theoretical model. SE, a key aspect of self-evaluation, is both a direct psychological result of social comparison and an important predictor of WB. ER reflects individuals’ internal management of emotional experiences elicited by social comparison, playing a particularly vital role in the digital social context. This study proposed a chain mediation model in which CUSC is associated with SE, which in turn influences ER, ultimately is associated with WB. The model aims to systematically uncover how college students psychologically respond to social comparison on social media. In sum, the integration of SCT and the ERM offers strong theoretical complementarity and substantial practical relevance, providing a solid theoretical foundation for understanding how online social comparison experiences are associated with the WB of college students.

Cyber upward social comparison and well-being

Upward social comparison refers to the process by which individuals, in the absence of objective standards, evaluate themselves against those perceived as more accomplished [25]. In digital environments, this process is intensified, as people often compare themselves with curated portrayals of success on social media [26]. According to SCT [25], people tend to evaluate their abilities and beliefs through comparisons with others, particularly when no objective standards are available. Studies have widely examined how upward social comparison on social media affects WB [38–40]. For example, Bazine et al. [41] found that SNS usage negatively influenced career satisfaction through upward social comparison and self-depletion. Li and Liu [42] showed that upward social comparison on social media significantly reduced WB. Some evidence also suggests potential benefits when upward comparisons are viewed as role-model learning opportunities [28, 43]. In Chinese contexts, CUSC appears especially detrimental due to cultural factors. Xu and Li [44] found that stronger face concerns predicted greater social anxiety during online comparisons, while Tian et al. [21] reported that collectivist orientations increased sensitivity to others’ online success. Gender norms further shape these effects, with female students more likely to engage in appearance-based comparisons and experience stronger emotional reactions [45]. Similar studies in other East Asian societies have also found that collectivist tendencies and social evaluation pressures amplify the negative impact of CUSC on WB [46, 47]. Overall, CUSC can be considered a double-edged sword. However, existing evidence indicates that its negative association with WB is more pronounced, a phenomenon particularly evident among student populations with relatively insufficient psychological resources.

The mediating role of self-esteem

Self-esteem (SE) is a key indicator of psychological health, reflecting individuals’ overall sense of self-worth [48]. Prior research has shown that SE functions as a protective resource, mitigating psychological stress and facilitating positive psychological adjustment. Elevated SE tends to correspond with decreased depressive symptoms [49–51], improved interpersonal relationships [52, 53], and greater life satisfaction [54, 55]. Many studies have shown a strong positive relationship between SE and WB [56–59]. For example, Katsantonis et al. [60] found that adolescents with high SE reported greater life satisfaction and positive emotions, while Kara and Aslan [61] showed that SE enhances perceptions of social support, further promoting WB. According to SCT [25], SE shapes individuals’ emotional reactions during comparisons with others. Those with high SE are more likely to sustain positive self-perceptions and buffer the negative emotional consequences of upward comparisons, which supports WB [12]. In contrast, individuals with low SE often interpret external events negatively, leading to more negative emotions and weaker WB [62].

Empirical research has shown that CUSC is closely linked with reduced SE [63–65]. Exposure to idealized online portrayals often triggers negative self-assessments and diminishes SE [66, 67]. For instance, Aubry et al. [68] found that CUSC increased perceptions of self–other discrepancies, threatening self-worth, while Midgley et al. [27] showed that upward comparisons with superior but similar peers particularly harm those with already low SE. SE has also been identified as a mediator between CUSC and mental health outcomes [69, 70]. For instance, Yuan et al. [71] demonstrated that SE mediates the link between CUSC and depressive symptoms in adolescents. Evidence from East Asian contexts further strengthens this pathway. In China, Hai and Yang [72] reported that collectivist values, by heightening sensitivity to social evaluation, indirectly reduced WB through lowering SE. Research also suggests that students with stronger face concerns are more vulnerable to SE decline after CUSC exposure. In Korean, Lee [73] found that social comparison pressure significantly undermined SE and reduced WB. These findings indicate that cultural factors such as collectivism, face culture, and gender socialization may intensify the mediating role of SE, making students with stronger social evaluation concerns more susceptible to self-worth erosion. In sum, SE mediates the link between CUSC and WB. Exposure to idealized online images can lower SE and reduce WB, whereas stable SE helps students maintain positive self-perceptions and emotional balance. Recognizing this mechanism clarifies how social media influences WB and highlights the need to strengthen psychological resources among Chinese college students.

The mediating role of emotional regulation

Emotional regulation (ER) involves adopting cognitive and behavioral techniques to adjust emotional responses, thereby facilitating adaptive functioning and preserving psychological WB [31]. According to the process model of ER [32], cognitive reappraisal (CR) and expressive suppression (ES) are the two primary strategies. CR, which involves reinterpreting situations more positively, is generally linked to reduced negative emotions, better psychological health, and higher life satisfaction. In contrast, ES, which inhibits emotional expression, is often associated with emotional exhaustion, weaker social support, and poorer adjustment [74, 75]. A growing body of research shows that ER strategies play a decisive role in WB. CR is consistently positively related to WB, whereas ES is negatively related [76, 77]. For example, Sadraei et al. [78] found that teachers using CR reduced negative emotions and enhanced WB by reframing challenges as growth opportunities. In contrast, ES tends to predict emotional strain and lower life satisfaction [79]. For college students, adopting adaptive strategies like CR is particularly important for coping with academic and social stressors and maintaining emotional stability.

Evidence indicates that CUSC influences the use of ER strategies [80–82]. Negative self-evaluations during upward comparisons may reduce CR and increase reliance on ES, thereby undermining WB [83, 84]. CR helps transform negative emotions into positive experiences, thereby enhancing WB, while ES is linked to emotional buildup and reduced WB [85]. Evidence from Chinese and East Asian contexts supports these mechanisms. Li and Liu [86] found that CUSC reduced the use of CR and increased reliance on ES, which in turn predicted lower WB. Similarly, Tamir et al. [87] reported that among Japan university students, face consciousness heightened negative emotional reactions to online comparisons, indirectly reducing WB through maladaptive ER strategies. Gender norms influence ER use as well, with female students more likely to engage in both CR and ES when facing comparison-related distress [88]. Building on this foundation, the present study introduces a dual-path ER model to examine how CR and ES function as mediators, aiming to clarify the psychological mechanisms through which CUSC influences WB and to provide insights for supporting students’ emotional adjustment.

The chain mediating role of self-esteem and emotional regulation

Research has shown a strong link between SE and ER [15], with low SE identified as a risk factor for anxiety and poor ER [14]. According to the ERM, ER refers to how individuals respond to and manage emotional stimuli [32]. SE, as a reflection of self-worth, directly influences which ER strategies individuals use [89]. Individuals with high SE typically hold more positive self-evaluations, which enables them to adopt more effective ER strategies, such as CR, when facing stress and negative emotions [90]. In contrast, those with low SE are more prone to self-doubt and negative emotions, and tend to rely on maladaptive strategies like ES [91].

Building on this, CUSC may lower SE, which then shapes ER and ultimately affects WB. When college students compare themselves with peers online, they may feel inferior, reducing SE and increasing reliance on maladaptive strategies [68]. As an effective regulation strategy, CR allows individuals to reinterpret negative emotions in a more positive light, thereby easing emotional distress [32]. However, individuals with low SE often lack the confidence and sense of self-worth needed to apply this strategy effectively, which makes it harder for them to manage negative emotions and maintain WB [92]. Evidence from Chinese and East Asian contexts further supports these pathways. Studies on Chinese students indicate that strong face concerns and collectivist values increase vulnerability to SE decline and reduce the use of CR [90]. Similarly, Korean students with high face consciousness show stronger negative reactions to CUSC, which lowers WB through maladaptive ER [93]. Gender differences have also been reported: males report higher SE and ES, while females report greater shame and more CR use [15]. In sum, CUSC may reduce SE, which in turn influences ER strategies and thereby WB. This chain mediation highlights the psychological pathway of social comparison and underscores its cultural relevance in Chinese and East Asian contexts.

The present study

Based on relevant literature and previous empirical findings, the theoretical model is illustrated in Fig. 1, and the following hypotheses are proposed:

Fig. 1.

Research model diagram

• H1: CUSC negatively affects WB.

• H2a: CUSC is negatively related to SE.

• H2b: SE is positively related to WB.

• H2c: SE mediates the relationship between CUSC and WB.

• H3a: CUSC is negatively related to CR.

• H3b: CR is positively related to WB.

• H3c: CR mediates the relationship between CUSC and WB.

• H4a: CUSC is positively related to ES.

• H4b: ES is negatively related to WB.

• H4c: ES mediates the relationship between CUSC and WB.

• H5a: SE and CR jointly mediate the association between CUSC and WB.

• H5b: SE and ES jointly mediate the association between CUSC and WB.

Methodology

Sample and data collection procedure

We collected data via the online survey platform (https://www.wjx.cn/) between January and April 2025. We recruited participants using a random sampling strategy from several universities in Dalian and Shenyang, two major cities in Liaoning Province, northeastern China. The specific procedure was as follows: First, we contacted several universities and, upon obtaining approval from the relevant departments, acquired the complete student rosters for each class. Then, using a computer-based random number generator (such as the RAND function in Excel), all students were assigned a number, and participants were randomly selected to ensure that each student had an equal chance of being chosen. Counselors assisted the research team by explaining the purpose of the study during class meetings, emphasizing the voluntary nature of participation and the anonymity of the data. Questionnaire links were then sent to the randomly selected students only. Each questionnaire included a unique identification code to ensure that only selected students could complete it, thereby preventing non-target respondents from affecting data quality.

Following the sample size estimation guideline suggested by Kline [94], at least 10 participants are recommended for each questionnaire item. Based on the 37 survey items and an expected 20% dropout rate, a minimum of 444 participants was needed. To ensure sufficient data, 555 questionnaires were administered, resulting in 500 valid responses. During data cleaning, 55 questionnaires were excluded based on two criteria: more than 20% of items left unanswered, or over 80% of items answered with extreme responses. First, according to Schafer and Graham [95] regarding the proportion of missing data, when more than 20% of a questionnaire’s items are left unanswered, it may compromise the structural stability and analytical validity of the data. As such, the 20% threshold is widely regarded as an important criterion for determining the validity of a questionnaire. Second, concerning the proportion of extreme responses, Wang et al. [96], in their review on method bias in behavioral research, noted that a high degree of uniform extreme responses may result from inattentive responding, response sets, or other non-genuine answering behaviors, leading to floor or ceiling effects that severely impair measurement reliability and validity. Based on the analysis of our pilot data, we found that when extreme responses exceeded 80%, response patterns became highly uniform and failed to accurately reflect participants’ true psychological states. Therefore, this was set as an exclusion criterion. 500 valid questionnaires were collected, with a response rate of 90.10%.

As shown in Table 1, among the 500 respondents, 234 (46.80%) were male and 266 (53.20%) were female. The majority (n = 354, 70.80%) were aged 18–20. In terms of academic standing, half of the respondents (n = 250, 50.00%) were first-year undergraduate students. social sciences (44.40%), humanities (43.00%), and natural sciences (12.60%), ensuring disciplinary diversity. Additionally, 276 participants (n = 55.20%) reported coming from urban areas. For monthly living expenses, most participants (n = 211, 42.20%) reported a range between 1601 and 2500 RMB.

Table 1.

Demographic characteristics of the participants by gender, age, grade level, academic major, place of origin, and monthly living expenses

Demographic Characteristics	Category	n	Percentage (%)
Gender	Male	234	46.80
	Female	266	53.20
Age	Under 18	12	2.40
	18–20 years	354	70.80
	20–22 years	109	21.80
	Above 22	25	5.00
Grade Level	Freshman	250	50.00
	Sophomore	132	26.40
	Junior	97	19.40
	Senior	21	4.20
Academic Major	Humanities	215	43.00
	Social Sciences	222	44.40
	Natural Sciences	63	12.60
Place of Origin	Urban	276	55.20
	Town	122	24.40
	Rural	102	20.40
Monthly Living Expenses (RMB/month)	0–800	20	4.00
	801–1200	52	10.40
	1201–1600	141	28.20
	1601–2500	211	42.20
	2501–3500	51	10.20
	3501–4500	10	2.00
	4501–10,000	15	3.00

Measures

Cyber upward social comparison

CUSC was measured using the upward comparison subscale of the Iowa-Netherlands Comparison Orientation Measure (INCOM), originally developed by Gibbons and Buunk [97] and later adapted for China by Lian et al. [98]. This subscale contains six items that assess the frequency and intensity of comparisons with individuals perceived as better off on social media (e.g., “I often compare myself to more successful individuals on social media.“). Responses were recorded on a five-point Likert scale from 1 (strongly disagree) to 5 (strongly agree), where higher values represented stronger CUSC tendencies. This scale has been validated among Chinese populations and has demonstrated good reliability and validity [10]. In the present study, the Cronbach’s α coefficient for the scale was 0.930, indicating excellent internal consistency.

Well-being

WB was assessed with the Satisfaction with Life Scale (SWLS) by Diener et al. [99]. The SWLS consists of five items designed to measure individuals’ overall cognitive evaluations of their life satisfaction. Each participant rated the statements on a seven-point Likert scale (e.g., “1 = strongly disagree” to 7 = “strongly agree”), with sample items such as “I am satisfied with my life” and “My life conditions are excellent.” Scores were averaged across all items, with higher values representing higher levels of subjective WB. Evidence from Chinese populations supports the scale’s high reliability and validity [100]. In the present study, the Cronbach’s α coefficient for the scale was 0.893, indicating good internal consistency.

Self-esteem

The Chinese Rosenberg Self-Esteem Scale (RSES), developed by Rosenberg [48] and revised by Yang and Wang [101], was used to measure SE. This scale contains 10 items (e.g., “I feel that I have a number of good qualities”) that assess individuals’ self-evaluations of their worth. Reverse coding was applied to items 3, 5, 8, 9, and 10. Each item was rated on a four-point Likert scale (e.g., “1 = strongly disagree” to 4 = “strongly agree”), with total scores reflecting participants’ SE levels. The scale has been validated in Chinese samples and has shown strong psychometric properties [102]. In the present study, the Cronbach’s α coefficient for the scale was 0.925, indicating excellent internal consistency.

Emotional regulation

The Emotion Regulation Questionnaire (ERQ) by Gross and John [32] was used to measure how individuals regulate their emotions in stressful situations. The ERQ comprises two dimensions: CR and Expressive suppression (ES), containing a total of 10 items. A seven-point Likert scale was used, with responses ranging from 1 (strongly disagree) to 7 (strongly agree). The CR dimension includes six items (e.g., “I manage my emotions by changing my perspective”), whereas the ES dimension consists of four items (e.g., “I manage my emotions by suppressing them”). CR and ES scores were calculated separately, with higher scores representing greater use of each strategy. Validated for Chinese populations, the scale has shown strong reliability and validity [103]. In the present study, the Cronbach’s α coefficients were 0.929 for CR and 0.894 for ES, indicating excellent internal consistency for both dimensions.

Statistical analysis

A two-stage analysis using SPSS 26.0 and Smart PLS 4.0 was conducted. SPSS provided descriptive statistics and Pearson correlations for examining dataset characteristics. Smart PLS assessed the measurement model’s validity and reliability, with CFA used to check for common method bias. SEM explored the relationships between predictors and latent constructs. Following Sarstedt et al. [104], PLS-SEM was chosen for its predictive modeling advantages. After evaluating the models, multi-group analysis (MGA) was conducted to examine gender differences [105], using Henseler et al. [106] non-parametric MGA approach. Measurement invariance was tested with the MICOM procedure [107] before conducting MGA.

Results

Descriptive statistics analysis

As shown in Table 2, the descriptive statistics and correlation coefficients are presented for all study variables, including the mean (M), standard deviation (SD), kurtosis (Kur), skewness (Sk), and intercorrelations. Skewness and kurtosis absolute values were below 2 and 7, respectively, suggesting that the data satisfied the normality assumption. CUSC was significantly negatively correlated with WB (r = −0.42, p < 0.001), SE (r = −0.49, p < 0.001), and CR (r = −0.45, p < 0.001), but significantly positively correlated with ES (r = 0.63, p < 0.001). WB was significantly positively correlated with SE (r = 0.72, p < 0.001) and CR (r = 0.70, p < 0.001), and significantly negatively correlated with ES (r = −0.36, p < 0.001). Moreover, SE was significantly positively correlated with CR (r = 0.72, p < 0.001) and significantly negatively correlated with ES (r = −0.54, p < 0.001). CR and ES were also significantly negatively correlated (r = −0.40, p < 0.001).

Table 2.

Descriptive statistics, skewness, kurtosis, and pearson correlations of core study variables

	M ± SD	Sk	Kur	1	2	3	4	5
CUSU	2.98 ± 1.02	−0.23	−0.80	1
WB	4.34 ± 1.42	−0.44	−0.49	−0.42***	1
SE	2.91 ± 0.65	−0.70	0.31	−0.49***	0.72***	1
CR	27.58 ± 8.03	−0.84	0.37	−0.45***	0.70***	0.72***	1
ES	15.52 ± 5.87	0.02	−0.91	0.63***	−0.36***	−0.54***	−0.40***	1

CUSC Cyber upward social comparison WB Well-being, SE Self-esteem. CR Cognitive reappraisal, ES Expressive suppression, Sk Skewness, Kur Kurtosis

Path significance: ***p < 0.001

SEM analysis

Measurement model

Following the recommendations of Hair Jr et al. [108], the reliability and validity of the measurement model were assessed. Reliability indicators include the outer loadings of items and the CR of the scale. According to the guidelines, the outer loadings of items must exceed 0.70 [108]. In this study, the outer loadings for all constructs met or exceeded this threshold, ranging from 0.707 to 0.889, demonstrating high indicator reliability. Internal consistency reliability was assessed using CR, and all constructs exhibited CR values well above the required 0.708, with values of 0.945 for CUSC, 0.922 for WB, 0.937 for SE, 0.944 for CR, and 0.926 for ES. Additionally, convergent validity was confirmed as the average variance extracted (AVE) for all constructs surpassed the minimum criterion of 0.5, with AVE values of 0.740 for CUSC, 0.704 for WB, 0.600 for SE, 0.739 for CR, and 0.758 for ES [108]. Discriminant validity was evaluated using the heterotrait-monotrait (HTMT) ratio of correlations standard and the traditional Fornell-Larcker criterion [108]. All HTMT values were below the critical threshold of 0.85, confirming discriminant validity across constructs. Furthermore, the inter-construct correlations were lower than the square root of the AVE for each construct, as per the Fornell-Larcker criterion. For example, the square root of the AVE for CUSC, WB, SE, CR, and ES was 0.860, 0.839, 0.774, 0.860, and 0.871, respectively, all exceeding the corresponding inter-construct correlations [108]. These results confirm that the measurement model exhibits robust reliability, convergent validity, and discriminant validity.

Structural model

The structural model was assessed using collinearity diagnostics, path coefficients, and R² values. Multicollinearity was evaluated based on variance inflation factor (VIF), requiring values below 3.3 [109]. All VIF values (1.000–2.609) met this criterion, confirming the absence of multicollinearity.

Confirmatory factor analysis (CFA)

CFA was used to evaluate the measurement model, ensuring the questionnaire’s factor structure matches theoretical expectations and accurately measures the latent constructs. Confirmatory Factor Analysis (CFA) was conducted to assess the measurement model, ensuring that the factor structure of the questionnaire aligned with theoretical expectations and accurately captured the latent constructs. Parameter estimation was performed using the Maximum Likelihood Estimation (MLE) method, which is suitable for normally distributed data and yields stable and efficient estimates. As shown in Table 3, the model’s fit was evaluated using multiple indices including CMIN/DF, RMSEA, GFI, AGFI, CFI, IFI, and TLI. All values met the recommended thresholds, indicating a good model fit. Based on widely accepted criteria in the literature [110], a good model fit is indicated by CMIN/DF < 5, RMSEA < 0.05, and values exceeding 0.90 for GFI, AGFI, CFI, IFI, and TLI.

Table 3.

Confirmatory factor analysis results: model fit indices and recommended thresholds

Fit Index	Recommended Value	Final Model
CMIN/DF	< 5	1.658
RMSEA	< 0.05	0.036
GFI	> 0.9	0.919
AGFI	> 0.9	0.903
CFI	> 0.9	0.977
IFI	> 0.9	0.977
TLI	> 0.9	0.974

CMIN Chi-square value, DF Degrees of freedom, RMSEA Rroot mean square error of approximation, GFI Goodness-of-fit index, NFI Normed fit index, CFI Comparative fit index, IFI Incremental fit index, TLI Tucker- Lewis index

Direct effects

The baseline model tested the direct association between CUSC and WB, showing good fit indices (PLS-SEM: NFI = 0.918, SRMR = 0.038). CUSC explained 19.6% of WB’s variance, confirming the model’s explanatory strength. CUSC significantly negatively predicted WB (β = −0.418, p < 0.001). Therefore, H1 is supported.

Mediation effects

Figure 2 shows SE, CR, and ES in the baseline model to test the mediating roles of SE and ER. The model fit indices of the PLS-SEM structural model indicated good fit (NFI = 0.868, SRMR = 0.048). Regarding explanatory power, CUSC accounted for 23% of the variance in SE, 54.7% in CR, and 46.8% in ES. Additionally, the four core variables collectively explained 60.8% of the variance in WB, demonstrating strong explanatory power.

Fig. 2.

Path coefficients of the research model

Path coefficient analysis revealed that CUSC significantly negatively predicted SE (β = −0.480, p < 0.001), supporting H2a; significantly negatively predicted CR (β = −0.121, p = 0.005); and significantly positively predicted ES (β = 0.491, p < 0.001), supporting H3a and H4a, respectively. SE significantly positively predicted WB (β = 0.480, p < 0.001), supporting H2b. It also significantly positively predicted CR (β = 0.673, p < 0.001) and significantly negatively predicted ES (β = −0.295, p < 0.001), further highlighting its mediating role in ER strategy formation. The associations of ER strategies with WB were differentiated: CR significantly negatively predicted WB (β = 0.353, p < 0.001), supporting H3b, while ES did not significantly predict WB (β = 0.086, p = 0.083), thus failing to support H4b. Notably, the direct path from CUSC to WB was not statistically significant (β = −0.086, p = 0.113), suggesting its influence on WB is primarily indirect via SE and ER. Among the control variables, none showed significant associations with WB.

As shown in Table 4, the bootstrap analysis with 5,000 samples confirmed the mediating effects.SE served as a significant mediator between CUSC and WB (β = −0.230, p < 0.001, 95% CI [−0.291, −0.172], excluding zero). CR also showed a significant mediating effect (β = −0.043, p < 0.001, 95% CI [−0.075, −0.014], excluding zero), whereas ES did not function as a significant mediator (β = 0.042, p > 0.05, 95% CI [−0.006, 0.091], including zero). Furthermore, the chain mediation pathway “CUSC → SE → CR→WB” was significant (β = −0.114, 95% CI [−0.160, −0.076], excluding zero), while the alternative pathway “CUSC → SE → ES→WB” was not significant (β = 0.012, 95% CI [−0.002, 0.028], including zero). In summary, H2c, H3c, and H5a were supported, whereas H4c and H5b were not. These findings highlight the key mediating and chain roles of SE and CR in the link between CUSC and WB, while ES did not form a significant mediating mechanism in this process.

Table 4.

Bootstrap results for indirect, direct, and chain mediation effects in the relationship between CUSC and WB

Indirect Pathway	Estimated Value	Standard Error	t	95% CI
CUSC→SE→WB	−0.230***	0.030	7.655	[−0.291, −0.172]
CUSC→CR→WB	−0.043**	0.016	2.747	[−0.075, −0.014]
CUSC→ES→WB	0.042	0.024	1.720	[−0.006, 0.091]
CUSC→SE→CR	−0.323	0.035	9.344	[−0.393, −0.258]
CUSC→SE→ES	0.142	0.023	6.082	[0.097, 0.188]
SE→CR→WB	0.238	0.038	6.258	[0.167, 0.317]
SE→ES→WB	−0.025	0.015	1.635	[−0.059, 0.003]
CUSC→SE→CR→WB	−0.114	0.022	5.201	[−0.160, −0.076]
CUSC→SE→ES→WB	0.012	0.007	1.646	[−0.002, 0.028]

CUSC Cyber upward social comparison, WB Well-being, SE Self-esteem, CR Cognitive reappraisal, ES Expressive suppression

t represents the two tailed t-test value; Path significance: ***p < 0.001

Explanatory power and predictive relevance

R² values of endogenous constructs and Stone-Geisser Q² values indicate the model’s explanatory power and predictive relevance, respectively [111]. As shown in Table 5, the R² values demonstrate satisfactory explanatory strength, with predictors accounting for 61.3% of the variance in WB, 54.7% in CR, 46.8% in ES, and 23.0% in SE. All Q² values were above zero, confirming strong predictive relevance. Additionally, the model fit indices were acceptable, with NFI = 0.868 and SRMR = 0.048.

Table 5.

Explanatory power (R²), predictive relevance (Q²), and model fit indices of endogenous variables

	R2	R2Adjusted	Q2	Model fit
CR	0.547	0.545	0.193	SRMR:0.048
ES	0.468	0.466	0.398	NFI:0.868
SE	0.230	0.229	0.226
WB	0.613	0.605	0.166

R² represents the explanatory power of the model; R²_Adjusted is the adjusted value of R²; Q² represents the predictive relevance of the model

CUSC, Cyber upward social comparison, WBWell-being, SESelf-esteem, CRCognitive reappraisal, ESExpressive suppression

Multi-group analysis

Measurement invariance analysis

This study conducted multi-group analysis to compare and examine differences in path coefficients between male and female groups, a method widely adopted in prior research [112]. Before testing for moderation effects, measurement invariance was assessed using the Measurement Invariance of Composite Models (MICOM) procedure to identify any potential issues [107].

According to Hair [111], establishing measurement invariance requires both structural invariance (i.e., identical parameters and estimation methods) and compositional invariance (i.e., equal indicator weights across groups). In Smart-PLS 4.0, structural invariance is automatically maintained when the same path model and algorithm are used for both groups, which was the case in this study. Compositional invariance was tested using the permutation approach, and is considered established when the correlation between construct scores exceeds the 5% quantile of the empirical distribution. As shown in Table 6, compositional invariance was supported for all constructs except SE, which failed to meet the criterion (correlation = 0.999, P = 0.020 < 0.05). This result suggests that the measurement of SE may differ between male and female participants. Therefore, although measurement invariance is generally supported across gender groups, interpretations of multi-group analysis results involving SE should be made with caution, as direct comparisons may be confounded by measurement non-equivalence.

Table 6.

Results of MICOM step 2: compositional invariance testing across gender groups

Construct	Correlations among construct scores(H0 = 1)	5% Quantile of Empirical Distribution of cu	P	Compositional Invariance?
CR	1.000	1.000	0.246	Yes
CUSC	1.000	1.000	0.631	Yes
ES	1.000	0.999	0.378	Yes
WB	1.000	1.000	0.932	Yes
SE	0.999	0.999	0.020	No

CUSC Cyber upward social comparison, WB Well-being, SE Self-esteem, CR Cognitive reappraisal, ES Expressive suppression

Multi-group analysis

After assessing measurement invariance, multi-group analysis (MGA) was conducted to examine potential gender differences in structural relationships. Gender was coded using a dummy variable to distinguish male and female participants [113], and the analysis compared corresponding path coefficients between the two groups following established procedures [112].

As shown in Table 7, significant gender differences were observed in two key pathways. The positive relationship between CR and WB was stronger for males (β = 0.463) than for females (β = 0.247), with a statistically significant difference (p = 0.046). Similarly, the negative association between CUSC and CR was more pronounced among females (β = −0.264) and non-significant among males (β = 0.025), with a significant difference (p = 0.001). These findings suggest that female students may be more susceptible to experiencing reduced cognitive reappraisal ability when exposed to CUSC, whereas male students benefit more from CR in terms of WB. Other paths—such as “CUSC → ES”, “CUSC → SE”, “CUSC → WB”, and “ES → WB”—showed no significant gender differences (p > 0.05). Notably, the path comparisons involving SE (e.g., CUSC → SE, SE → CR, SE → WB) did not show statistically significant gender differences. However, as reported in Sect. 4.6.1, SE did not achieve compositional invariance across gender groups, suggesting that male and female participants may interpret or respond to SE items differently. Therefore, gender-based comparisons involving SE should be interpreted with caution, as observed differences or similarities may be partially influenced by measurement non-equivalence rather than true structural effects.

Table 7.

Multi-group analysis: comparison of path coefficients across gender groups

	Male	Female	Original difference	2.5%CI	97.5CI%	p	Hypotheses
CR → WB	0.463	0.247	0.216	−0.201	0.217	0.046	Supported
CUSC → CR	0.025	−0.264	0.289	−0.178	0.174	0.001	Supported
CUSC → ES	0.552	0.424	0.128	−0.158	0.163	0.133	Not Supported
CUSC → SE	−0.404	−0.537	0.132	−0.158	0.159	0.112	Not Supported
CUSC → WB	−0.064	−0.129	0.064	−0.217	0.215	0.570	Not Supported
ES → WB	0.066	0.112	−0.046	−0.204	0.199	0.662	Not Supported

CUSC Cyber upward social comparison, WB Well-being, SE Self-esteem, CR Cognitive reappraisal, ES Expressive suppression

Overall, while most pathways were consistent across gender, the results underscore meaningful gender differences in the formation and impact of CR. These findings highlight the importance of developing gender-sensitive psychological interventions that target ER strategies in digital contexts.

Discussion

This study explored how CUSC, SE, CR, and ES affect WB in college students. SEM revealed the complex connections among these factors. Additionally, the mediating roles of SE, CR, and ES between CUSC and WB were examined. The hypotheses proposed that CUSC would significantly negatively predict WB, with SE and CR serving as mediators, while ES would not show a significant mediating role. Multi-group analysis indicates that in most pathways, the predictive effects of CUSC on SE, CR, ES, and WB did not significantly differ by gender. However, significant gender differences emerged in the paths from “CUSC → CR” (Δβ = 0.289, p = 0.001) and “CR → WB” (Δβ = 0.216, p = 0.046). Specifically, CUSC significantly negatively predicted CR to a greater extent among females, while CR significantly positively predicted WB more strongly among males. These findings underscore gender differences in ER mechanisms within the digital environment and highlight the importance of gender-sensitive psychological interventions. A detailed discussion of these results follows.

Cyber upward social comparison and well-being

This study found that CUSC significantly negatively predicted WB among college students, which is consistent with the findings of Li and Liu [42], indicating that frequent engagement in upward social comparison in online environments leads to lower levels of WB. According to SCT [25], individuals evaluate themselves by comparing with others. In online environments, upward comparisons with idealized images of peers often elicit feelings of inferiority, reducing self-evaluations and lowering WB [114]. The curated and selective nature of social media content, such as achievements or appearance highlights, makes upward comparisons more frequent and salient, thereby intensifying negative reactions [10]. From a cultural perspective, Chinese collectivist and face-oriented values may heighten the sensitivity to external evaluations, amplifying the negative effects of CUSC on WB [115]. Although the direct predictive effect showed no significant gender differences, females reported lower WB under high comparison pressure, possibly because gendered socialization encourages greater emotional sensitivity and concern for social connectedness [116]. Overall, these findings underscore the explanatory power of SCT in clarifying how CUSC undermine WB in the Chinese cultural context. Additionally, educators and parents should pay attention to students’ online behaviors and provide emotional support. Future research could further optimize personalized intervention programs based on gender socialization mechanisms to enhance psychological adaptability and WB across different gender groups.

Mediating role of self-esteem

SE mediates the relationship between CUSC and WB, consistent with Piko et al. [65] and Shang et al. [58]. Frequent upward social comparison leads to negative self-evaluations, which reduce SE and lower WB. According to SCT [25], individuals tend to compare themselves with similar peers who outperform them. In digital contexts, college students often use classmates and “familiar others” as reference groups, making comparisons more personally relevant and threatening [21]. These upward comparisons, based on others’ “highlight moments” such as academic excellence or attractive appearance, weaken students’ affirmation of their own worth [53]. Within the Chinese cultural context, where face (mianzi) and social image are highly emphasized, such comparisons extend beyond personal inadequacy to relational threats, often accompanied by shame when social expectations are not met [117]. This cultural sensitivity to external evaluation amplifies the negative impact of CUSC on SE, particularly among students whose self-worth is closely tied to others’ opinions. As a result, SE serves as a key psychological bridge between CUSC and WB, demonstrating how social comparison undermines well-being through culturally shaped self-evaluations.

Mediating role of emotion regulation

CR significantly mediates the relationship between CUSC and WB, according to the results of this study. This finding supports St-Louis et al. [74] and Onorato et al. [118], showing that frequent upward social comparison leads to negative self-evaluations, weakening CR ability and reducing WB. From SCT [25], peers with similar backgrounds are especially threatening targets, while the ERM [32] suggests that CR is an antecedent-focused strategy that becomes less effective under repeated self-threats. In the Chinese cultural context, where academic success and social image are highly valued, such comparisons further intensify pressure and erode WB [2]. In practice, constant exposure to others’ achievements on social media makes college students more self-critical and less able to use CR effectively. Rather than reframing negative experiences, they often ruminate on personal shortcomings, which increases stress and reduces WB [35]. Over time, this cycle may lead to emotional exhaustion, lower life satisfaction, and greater vulnerability to anxiety and depression.

However, the study found no significant mediating role of ES between CUSC and WB. This finding contrasts with the results of Mendia et al. [85] and Sia and Aneesh [79]. According to the ERM [32], ES is classified as a response-focused strategy, which operates after the ER has been generated and targets the external expression of emotions rather than the cognitive origins of emotional experience. Therefore, in the context of frequent CUSC, even if individuals can suppress the external expression of negative emotions, the inner self-doubt, frustration, and negative emotional experiences persist, thereby failing to significantly improve WB [119]. As a response-focused strategy, it suppresses outward expression without reducing internal distress, which may even accumulate negative emotions and impair adaptation [120]. Gender analysis showed that CUSC more strongly reduced CR among females, while CR had a stronger positive effect on WB among males. This reflects gender socialization and Chinese cultural norms that encourage emotional restraint, leading female students to rely more on suppression strategies that are less beneficial for WB.

The chain mediating role of self-esteem and emotional regulation

This study found that SE and ER jointly exerted a significant chain mediating effect on the relationship between CUSC and college students’ WB. This suggests that frequent upward social comparison on platforms may lower SE, which in turn reduces the use of CR, an adaptive ER strategy, ultimately leading to decreased subjective WB. This finding is consistent with the study by Fernandes et al. [15], who reported that lower SE significantly predicted less effective emotion regulation and poorer mental health outcomes. Among the two ER strategies [32], only CR was significant, while ES showed no effect. CR, as an antecedent-focused strategy, helps reframe stressful experiences, whereas ES merely suppresses outward expression without reducing internal distress, which may explain its lack of impact. Beyond theoretical distinctions, previous studies in Chinese and East Asian samples similarly report that suppression often fails to buffer against comparison-related stress, as it preserves social harmony but does not alleviate internal distress [100]. Gender differences were also found: CR had a stronger positive effect on WB among males, while the negative impact of CUSC on CR was greater among females. These results are consistent with gender socialization patterns [116], where males are encouraged to adopt problem-focused strategies, while females are more vulnerable to comparison-related strain. In sum, SE and CR form a key chain mediating pathway linking CUSC to WB, highlighting the limited role of ES and the importance of gender and cultural factors in understanding students’ emotional adaptation in digital contexts.

Implications and limitations

This study used SEM to explore how CUSC predicts university students’ subjective WB, with a particular focus on the mediating roles of SE and ER. The model explained 61.3% of the WB variance, highlighting the substantial impact of online social comparison on students’ psychological functioning. The results offer valuable theoretical and practical insights.

Theoretical implications

First, this study theoretically extends the application of social comparison theory and emotion regulation models in the context of online environments. By constructing a dual-path mediation model of CUSC, SE, ER and WB, the study systematically presents how external social stimuli, through individuals’ internal cognitive and emotional mechanisms, jointly affect their psychological adaptation outcomes. This integrated pathway reveals various psychological mechanisms associated with WB and moves beyond the linear relationships and unidimensional variables common in traditional social comparison research, offering a more ecologically valid framework for understanding mental health issues in the social media context.

Second, SE and ER were confirmed to serve as mediators in the predictive relationship between CUSC and WB. The significant mediating role of SE suggests that when individuals are confronted with others’ “highlight moments” online, their cognitive evaluation of self-worth is closely associated with their WB experience. This supports the implicit pathway explaining how cyber social comparison influences WB. The significant mediating role of CR further emphasizes the importance of ER in social contexts. Specifically, individuals with strong ER abilities can reduce negative emotions and enhance WB by reinterpreting negative social cues.

Third, the study found that the direct predictive effect of ES on WB was not significant, which contrasts with some findings in the existing literature. This suggests that ES may not be a decisive factor directly influencing WB, but could exert an indirect impact through mechanisms such as emotional accumulation over time or the erosion of perceived social support. This offers theoretical insights for future research on classifying emotion regulation strategies and exploring their underlying mechanisms.

Furthermore, multi-group analysis revealed gender differences in path relationships. Specifically, the negative predictive effect of CUSC on CR was stronger among females, whereas the positive predictive effect of CR on WB was more pronounced among males. These findings emphasize the relevance of gender socialization theory in emotion regulation differences and offer a basis for future gender-specific psychological interventions.

Practical implications

The study’s findings indicate that college students should enhance their self-awareness and emotional responses to CUSC on online platforms. They should avoid developing negative self-evaluations or emotional distress due to frequent upward comparisons. Additionally, students should actively cultivate healthy levels of SE and effective ER strategies. For example, self-affirmation exercises, positive attribution training, and other methods can be used to improve their sense of self-worth, while adaptive emotion regulation strategies such as CR and problem-solving can be applied to alleviate negative emotions arising from social comparison. Moreover, the study results suggest that college students should establish reasonable internet usage habits. They should focus less on others’ curated “perfect images” and more on their own growth and real-life experiences to lessen the negative associations of upward social comparison on WB.

From the perspective of school education, universities should integrate online mental health education into their regular curriculum, with a particular focus on guiding students’ understanding of social comparison psychology and fostering emotion regulation skills. Specifically, schools can help students properly view online information and enhance self-identity and ER abilities by offering media literacy education, SE enhancement workshops, and emotion regulation skills training courses. Furthermore, universities should encourage offline social activities and practical experiences, providing students with diverse platforms to showcase their authentic selves and gain a sense of achievement, thus reducing their reliance on online social comparisons. Additionally, psychological counseling centers can design targeted interventions for student groups who are more vulnerable to social comparison, offering personalized psychological support to promote their mental health development.

At the level of the social support system, online platforms, educational institutions, and society at large should collaborate to create a healthy online environment. First, social media platforms should optimize content recommendation mechanisms to reduce the dissemination of extreme, idealized content that may trigger negative social comparisons, while promoting authentic and diverse value guidance, such as setting up “anti-perfectionism” labels or pushing psychological adjustment tips. Second, educational departments and media organizations can jointly carry out online mental health awareness campaigns to raise public awareness of the potential risks associated with upward social comparison and encourage a rational approach to online information. Additionally, there should be a strengthened psychological support network involving family, schools, and communities. By promoting parent education, community lectures, and other outreach activities, society can enhance attention to college students’ mental health and foster a more inclusive and supportive developmental environment. This will help to systematically enhance their subjective WB and psychological adaptation.

Limitations and future research directions

While this study provides valuable preliminary insights into the relationships among CUSC, SE, ER, and WB, several limitations should be noted. First, the reliance on self-report surveys may introduce bias, and future studies could combine interviews or case studies to obtain richer data. Second, the cross-sectional design prevents capturing long-term changes, highlighting the need for longitudinal research. Third, the sample was limited to Chinese college students, which restricts the generalizability of findings; future research should include larger and more diverse samples across cultures. Finally, the MICOM analysis indicated that SE did not achieve compositional invariance across gender groups, raising concerns about the robustness of gender-based comparisons. Although the observed patterns remain theoretically meaningful, future studies should refine SE measurement, adopt item response theory (IRT), or explore alternative models to improve cross-group comparability and enhance the validity of gender-specific inferences.

Conclusion

Drawing on social comparison theory and the emotion regulation model, this study examines how CUSC influences college students’ subjective WB, focusing on the chain mediating roles of two key internal resources: SE and ER strategies. This study systematically explores the multiple mechanisms influencing college students’ WB in the digital social environment from the dual-path perspective of cognition and emotion by constructing a pathway model of “social comparison, psychological mechanisms, and WB. The path model, analyzed using Smart-PLS 4.0 SEM, explained 61.3% of WB variance, demonstrating strong explanatory strength. Multi-group analysis revealed gender differences in the associations involving CR, highlighting the importance of gender-specific interventions. This research provides a theoretical framework for understanding social comparison’s influence on WB and practical insights for psychological support in higher education. However, due to the cross-sectional design, causal conclusions are limited. Future research should integrate moderating variables and employ longitudinal methods to assess long-term associations.

Authors’ contributions

Conceptualization: Fengliang Zuo; Methodology: Fengliang Zuo; Formal analysis and investigation: Fengliang Zuo; Writing - original draft preparation: Fengliang Zuo; Writing - review and editing: Fengliang Zuo; Supervision: Qi Zan. All the authors have read and agreed to the published version of the manuscript.

Funding

No Funding.

Data availability

The data that support the findings of this study are available on request from the corresponding author.

Declarations

Ethics approval and consent to participate

The researchers confirms that all research was performed in accordance with relevant guidelines/regulations applicable when human participants are involved (e.g., Declaration of Helsinki or similar). This study was approved by the Ethics Committee of Liaoning University. The participants received oral and written information and provided written informed consent before participating in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.