This investigation drew upon data from the 2020 China Family Panel Studies (CFPS), a nationally representative survey funded by the National Natural Science Foundation of China and administered by Peking University’s Institute of Social Science Survey (53). Our sample selection process was as follows:

Inclusion criteria:

1. Age between 16-18 years old

Exclusion criteria:

1. Missing data for important variables or total missing data for a respondent reaching 20% or more

From the initial 28530 samples, we selected respondents aged 16-18 years (n=910). We excluded cases with missing data for key variables or where total missing data exceeded 20% per respondent. The final sample consisted of 691respondents (493 aged 16-17 and 198 aged 18), with balanced distribution of gender and urban/rural residency.

Parental Democratic Communication. We assessed this construct based on Baumrind’s (54) theory of democratic parenting using six items that measured the frequency of parents’ democratic communication styles over the preceding 12 months. Including “Parents ask for reasons”, “Parents encourage you to try to do things”, “Parents talk to you kindly”, “Parents encourage you to think independently”, “Parents tell you why” and “Parents like to talk to you.” Participants responded on a 5-point Likert scale ranging from 1 (never) to 5 (always), with higher scores reflecting stronger democratic parental communication. The scale demonstrated good internal consistency (Cronbach’s α = .836 for 16-17-year-olds;. 809 for 18-year-olds).

Societal Trust. With reference to the key dimensions of social trust identified in previous research (55), and based on the results of principal component factor analysis, we selected three items - trust in neighbors, local government officials, and doctors - as measures of societal trust. These items best reflect social trust in everyday life. ¹ This construct was measured using three items assessing trust in neighbors, local government officials, and doctors. Responses were recorded on a 10-point scale from 0 (very distrustful) to 10 (very trusting), with higher scores indicating greater community trust. The internal consistency of the three-item societal trust measure was assessed using Cronbach’s alpha. For 16-17-year-olds, α = .661, and for 18-year-olds, α = .620 (57). This is consistent with findings on the multidimensional nature of social trust in China (58).

Negative Emotion and Pleasant Life. The CFPS2020 used a modified version of the Center for Epidemiologic Studies Depression Scale (CESD-8) (59). Based on its core characteristics (60) and supported by factor analysis, we conceptualized the scale as having two dimensions: pleasant life and negative emotions (61, 62). The Pleasant Life construct was measured using two items from the CESD-8: “I feel happy” and “I live a happy life.” Responses were recorded on the same 4-point scale as the Negative Emotions subscale, ranging from 1 (less than a day) to 4 (5-7 days). Higher scores on this subscale indicate a greater sense of life satisfaction. The two items showed moderate correlation (r = .572, p <.01 for 16-17-year-olds; r = .659, p <.01 for 18-year-olds), supporting their use as a composite measure.

To refine the negative emotions subscale, we conducted a series of analyses:

1. Factor Analysis: Our initial factor analysis supported a two-factor structure, aligning with previous research on emotion dynamics in major depressive disorder (60).

2. Item Reduction: Within the negative emotions factor, we examined the communalities of the six original items. Two items, “sleep difficulties” and “feeling life is unmanageable,” were removed due to common factor variance extraction rate below 0.5 (63), indicating they shared less variance with other items ( Supplementary Material 4 ).

3. Model Refinement: Using AMOS software, we examined modification indices for the remaining items. The covariance between error terms for “I felt everything I did was an effort” and “I felt depressed” was notably high (MI = 7.83) in the 18-year-old group, suggesting potential redundancy or inconsistency with the model.

4. Final Item Selection: Considering these statistical results and aiming to improve model fit, parsimony, and explanatory power while maintaining theoretical integrity, we further removed the item “I felt everything I did was an effort.”

The final negative emotions subscale comprised three items: “I felt depressed,” “I felt lonely,” and “I felt sad.” Participants were asked to report the frequency of these negative emotions over the past week. Responses were recorded on a 4-point scale ranging from 1 (less than a day) to 4 (5-7 days), with higher scores indicating more frequent negative emotions. This refined subscale balances statistical considerations with the core theoretical construct of negative emotions in depression. The scale demonstrated good internal consistency, with Cronbach’s alpha of.744 for 16-17-year-olds and.708 for 18-year-olds, surpassing the conventional.70 threshold for acceptable reliability.

Subjective Well-being. We assessed this construct using three items that measured happiness, life satisfaction, and future confidence. Due to the mixed scale format, items were standardized prior to reliability analysis. The scale showed acceptable internal consistency (Cronbach’s α = .689 for 16-17-year-olds;.666 for 18-year-olds).

All measures were derived from the CFPS questionnaire, which has been validated for use in the Chinese context (53). We included gender and urban-rural residence as control variables.

To address the non-consistency in score statistics between questions, we standardized individual question scores before aggregating them to construct final scores for latent variables. We computed descriptive statistics using SPSS 26.0 and examined the factor structure of key variables through exploratory factor analysis (EFA).

We tested hypothesized mediation models and conducted multicohort analyses across age groups using structural equation modeling (SEM) in AMOS 26.0 (64). Following Anderson and Gerbing’s (65) two-step approach, we first estimated measurement models using confirmatory factor analysis (CFA) to assess construct validity and reliability. We then estimated the structural model to test hypothesized relationships and mediating roles of societal trust, negative emotions, and pleasant life.

We use Maximum Likelihood (ML), set the convergence criteria from 1E-05 to 0.001, and limit the number of iterations to 50, comparative fit index (CFI), Tucker-Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR) to evaluate model fit. We tested mediating effects using a bootstrap method (2,000 resamples) to estimate bias-corrected 95% confidence intervals (CIs), with significance determined if the 95% CI did not include zero (66). Using age as the grouping variable, we examined unconstrained, measurement-weighted, structural-weighted, structural covariance, structural residual, and measurement-residual models for the two age groups (29–31) to assess model structure homogeneity, factor loadings, intercepts, and error variances. We determined model stability and invariance by comparing the absolute value of the critical ratio (α = .05 corresponds to a critical value of 1.96) of fit metrics and parameter differences across models (67). Pathways with critical ratios > 1.96 indicated significant differences, determining pathway influence across groups (68).

To rigorously test our model’s robustness, we conducted a comprehensive sensitivity analysis ( Supplementary Material 2 ). This analysis (69) involved varying several key aspects of our modeling approach: Employing different parameter estimation methods; Adjusting convergence criteria; Increasing the number of iterations; Expanding the number of bootstrap replications;and testing theoretically feasible alternative models. The results of this sensitivity analysis demonstrated remarkable consistency and stability in our model’s performance. Across these variations, the key fit indices (CFI, TLI, RMSEA, and SRMR) remained largely unchanged, exhibiting only minor fluctuations. This stability across different analytical conditions provides strong evidence for the robustness of our model, enhancing confidence in the reliability and generalizability of our findings.

In this study, semPower package 2.1.1 of R version 4.3.1 (2023-06-16 ucrt) was used for power analysis to check the reasonableness of the sample size ( Supplementary Material 3 ). Calculations using the RMSEA values and degrees of freedom of both models yielded an actual efficacy of 80.03% in the 16-17-year-old group and 80.13% in the 18-year-old group, which are both greater than 80%. This result proves that the sample sizes of the two models are at a reasonable level, which can provide reliable support for the conclusions of the study and ensure that the results of the study are statistically valid and credible.

We assessed common method bias using Harman’s single-factor method (70). For both age groups, five factors emerged with eigenvalues greater than 1. The first factor accounted for 19.958% and 18.096% of the variance for the 16-17 and 18-year-old groups, respectively. These values fall below the 40% threshold, suggesting no significant common method bias in either group.

For the 16-17 age group, skewness values ranged from -0.62 to 1.23, and kurtosis values ranged from -0.25 to 2.48. The 18-year-old group exhibited skewness values between -0.63 and 0.54, and kurtosis values between -0.37 and 1.38 ( Table 1 ). These values fall within acceptable ranges, indicating normal distribution of the data (71).

Correlation analyses ( Table 2 ) revealed significant relationships between key variables. In both age groups, subjective well-being showed significant positive correlations with parental democratic communication (r = 0.30, p <.01 for 16-17; r = 0.25, p <.01 for 18), societal trust (r = 0.34, p <.01 for 16-17; r = 0.27, p <.01 for 18), and pleasant life (r = 0.25, p <.01 for 16-17; r = 0.14, p <.05 for 18). Conversely, subjective well-being demonstrated significant negative correlations with negative emotions (r = -0.34, p <.01 for 16-17; r = -0.23, p <.01 for 18). In the 16-17 age group, pleasant living was positively correlated with parental democratic communication (r = 0.23, p <.01) and social trust (r = 0.15, p <.01) and negatively correlated with negative emotions (r = -0.28, p <.01). In the 18-year-old group, pleasant life was positively correlated with parental democratic communication (r = 0.21, p <.01) and negatively correlated with negative emotions (r = -0.16, p <.05). Negative emotions were negatively correlated with parental democratic communication (r = -0.19, p <.01 for 16-17) and social trust (r = -0.19, p <.01 for 16-17; r = -0.14, p <.05 for 18). In both groups, social trust was positively correlated with parental democratic communication (r = 0.15, p <.01 for 16-17; r = 0.26, p <.01 for 18).

We evaluated the measurement model using exploratory factor analysis (EFA), followed by confirmatory factor analysis (CFA). We conducted these analyses separately for both age groups.

For the 16-17 age group ( Table 4 ), the total indirect effect accounted for 43.46% of the total effect, while the direct effect accounted for 56.75%. The specific effect sizes for each mediating pathway were:

1) PDC→ST→SWB: 15.40% of total effect

2) PDC→NE→SWB: 9.70% of total effect

3) PDC→PL→SWB: 9.49% of total effect

4) PDC→ST→NE→SWB: 2.32% of total effect

5) PDC→PL→NE→SWB: 6.33% of total effect

For the 18-year-old group: only one mediating pathway showed a significant effect size: PDC→ST→SWB: 58.39% of total effect.

The multi-group analysis demonstrated good overall model fit across different invariance levels. The measurement-weighted model showed stability, while other models exhibited non-significant differences. This suggests that the measurement structure is consistent across age groups, but there may be some differences in structural relationships ( Tables 5 , 6 ).

Path Coefficients Comparison ( Table 7 ):

1) Parental Democratic Communication (PDC) to Societal Trust (ST), both significant, slightly stronger for 18-year-olds.

2) PDC to Pleasant Life (PL), significant and similar for both groups.

3) PDC to Negative Emotion (NE), significant only for 16-17 years.

4) ST to NE, significant only for 16-17 years.

5) PL to NE, significant only for 16-17 years.

6) ST to Subjective Well-being (SWB), significant for both, stronger for 18-year-olds.

7) PL to SWB, significant only for 16-17 years, notable difference in direction. The critical ratio is 2.769 and greater than 1.96.

8) NE to SWB, significant for both, slightly stronger for 16-17 years.

9) PDC to SWB (Direct effect), significant only for 16-17 years.

As highlighted in Table 8 , our study findings suggest that the influence of parental democratic communication on subjective well-being, as well as the mediating mechanisms, differ between the two age groups. Our findings highlight the complex interplay between PDC, societal trust, negative emotions, and pleasant life experiences in shaping adolescent well-being.

Our findings reveal significant age-related differences in the impact of Parental Democratic Communication (PDC) on Subjective Well-Being (SWB) among Chinese adolescents. The direct effect of PDC on SWB was significant for mid-adolescents (16-17 years) but not for late adolescents (18 years). This shift likely reflects the changing dynamics of parent-child relationships as adolescents gain autonomy (73), prioritize independent self-construal (74), and focus more on peer relationships and identity development (75–77). These results suggest that supportive and open parent-child communication may be particularly crucial for younger adolescents (78, 79). Our findings are consistent with recent research in China that emphasizes that adolescents’ autonomy increases with age. Li (80) notes that high school students in China experience enhanced psychological resilience and autonomy as they mature, which may explain the changing dynamics in our study. Similarly, Wang and Zhang (81) emphasize the growing importance of peer relationships for Chinese adolescents, supporting our observation of diminishing direct parental influence on well-being for 18-year-olds.

However, our findings extend beyond these observations by revealing that parental influence persists, albeit through different mechanisms. The mediating role of social trust remained significant across both age groups, indicating that even as Chinese adolescents transition to young adulthood, parental communication continues to shape their perceptions of the social world. This observation challenges the notion that parental influence necessarily diminishes with increasing adolescent autonomy, a concept often emphasized in Western literature (82). Our results align with and extend previous research linking societal trust to psychological adjustment and well-being among Chinese adolescents (83, 84).

For mid-adolescents (16-17 years), our study revealed a complex serial mediation involving negative emotions, pleasant life experiences and societal trust. This finding aligns with existing research showing that supportive parental communication is associated with lower levels of anxiety and depression (85, 86), as well as higher levels of life satisfaction and positive affect (87, 88). Such communication appears to provide a secure foundation for building societal trust and navigating social relationships (89), while also minimizing negative emotions (46). However, our study extends these findings by demonstrating a serial mediation relationship, offering a more nuanced understanding of how parental democratic communication (PDC) influences well-being through multiple pathways simultaneously. In the Chinese context, deep family values and an emphasis on filial piety often lead parents to provide more instruction and discipline (90). Consequently, adolescents place significant weight on parental communication and advice (91–93). In addition, with the trend of delayed adulthood becoming increasingly prevalent (94), adolescents may rely more heavily on parental guidance for developing social trust, experiencing pleasant life events, and regulating emotions, all of which influence their subjective well-being.

In late adolescence, we observe a significant developmental shift where societal trust becomes the primary mediator between parental democratic communication and well-being for 18-year-olds. This transition aligns with previous research indicating that during this period, 18-year-olds diversify their social relationships, placing greater value on interactions with peers, teachers, and community members, which significantly impact their psychology and behavior (95). This shift underscores the increasing importance of social connections during the transition to adulthood, extending theories of emerging adulthood (75).

Our findings demonstrate how parental influence adapts and persists into this developmental period, particularly in the Chinese context, by identifying societal trust as a key mediator. This perspective builds upon previous studies on parenting and adolescent well-being in China, which have emphasized the enduring influence of family due to deep-rooted cultural values. These studies have highlighted that parental education and care tend to be present throughout their children’s development (96), with factors such as parental expectations and family atmosphere significantly impacting adolescents’ academic achievement and psychological well-being (97).

Our study reveals the paramount importance of social trust in adolescent well-being, surpassing traditionally emphasized factors such as negative emotions and pleasant life experiences. This finding led us to propose the concept of “trust-mediated well-being.” For 16-17-year-olds, societal trust significantly influences subjective well-being through both direct (β=0.359) and indirect pathways. Indirectly, societal trust reduces negative emotions (β=-0.169), which in turn affects subjective well-being (β=-0.32). The effect of pleasant life experiences (β=0.14) is considerably smaller than that of societal trust. Among 18-year-olds, the importance of societal trust increases dramatically. Its direct effect on subjective well-being (β=0.481) accounts for 58.39% of the total effect, while the impact of negative emotions decreases (β=-0.246), and pleasant life experiences cease to have a significant effect. This concept is particularly relevant in addressing the global trend of increasing loneliness and social isolation, exacerbated by “lonelygenic environments” (13), although the specific mechanisms may differ by age.

These findings challenge the traditional emphasis on emotional regulation and hedonic experiences in well-being research, suggesting a more complex and age-specific relationship between societal trust and well-being. Our results indicate that the ability to trust and feel connected to society plays a crucial role in well-being throughout adolescence, but its mechanisms appear to evolve with age. For mid-adolescents (16-17 years), our findings partially align with recent research highlighting social trust’s protective role against negative emotions, including loneliness (98–100). In this age group, societal trust not only directly enhances well-being but also shows a small but significant negative relationship with negative emotions. However, for late adolescents (18 years), while societal trust remains a significant direct contributor to well-being, its relationship with negative emotions becomes non-significant. This suggests a shift in how societal trust operates during the transition to adulthood, maintaining its importance for overall well-being but potentially becoming decoupled from immediate emotional experiences.

In the Chinese context, where rapid social changes, including the increasing prevalence of online interactions and intense academic pressure (101, 102), as well as the COVID-19 pandemic have strained traditional social bonds (103, 104), understanding the role of societal trust in well-being becomes even more critical. Our findings suggest that while societal trust consistently contributes to well-being, its relationship with negative emotions and potential loneliness may be more complex and age-dependent than previously thought.

Trust-mediated well-being aligns with emerging concepts in the third wave of positive psychology, that emphasizes the interconnectedness of individual and collective well-being (105, 106). It extends this idea by demonstrating how individual well-being is fostered through one’s connection and trust in the larger social fabric, especially evident in our findings for 18-year-olds. In the Chinese context, where collective harmony is culturally valued, our findings offer a bridge between traditional collectivist values and contemporary approaches to individual well-being. This trust-mediated well-being model provides a framework for understanding how positive family dynamics contribute to both individual and societal well-being through the cultivation of social trust.

Our study makes significant contributions to adolescent development and well-being research, aligning with the third wave of positive psychology:

1. “Societal Trust-Mediated Well-Being” Pattern:

We reveal a pattern where societal trust mediates the relationship between parental democratic communication and adolescent well-being, with this mediation strengthening with age. This framework bridges individual, familial, and societal levels of analysis, resonating with the holistic approach of third-wave positive psychology.

2. Extension of Emerging Adulthood Theory:

Traditionally, theories of adolescent development, including some interpretations of Arnett’s emerging adulthood theory (75), have posited that individuals become less influenced by their parents and more independent as they approach and enter adulthood around age 18. However, our study reveals a more nuanced picture. We find that the influence of parental democratic communication does not simply diminish at age 18; instead, its pathway of influence evolves. For 18-year-olds, parental influence manifests more prominently through the mechanism of societal trust, which in turn affects well-being.

3. Cultural Sensitivity:

Our research provides a nuanced understanding of how family relationships evolve and maintain their importance during this critical developmental period. This insight is particularly valuable in the Chinese context, where family ties traditionally remain strong even as young people enter adulthood, offering a culturally sensitive extension to existing theories of adolescent development and emerging adulthood.

4. Addressing Global Youth Negative Emotions including Loneliness:

In response to growing global concerns about youth loneliness and negative emotions (3, 13), our study offers a novel theoretical framework. This framework elucidates how positive family interactions can serve as a buffer against these issues through the cultivation of trust-mediated well-being. While previous research has predominantly focused on individual-level interventions to mitigate loneliness and negative emotions, our study breaks new ground by integrating family dynamics and societal trust into the equation. By examining the interplay between parental democratic communication, societal trust, and adolescent well-being, we address a significant theoretical gap in tackling these pressing social issues. Our approach provides a more comprehensive understanding of the protective factors against youth loneliness and negative emotions, emphasizing the crucial role of family relationships and broader social connections in fostering resilience and emotional well-being among adolescents and young adults.

Our findings suggest the need for integrated, age-tailored approaches in promoting adolescent well-being. For younger adolescents, interventions might focus on enhancing PDC to directly impact SWB, in addition to societal trust, emotion regulation and pleasant life experiences. For older adolescents, the focus might shift to fostering trust-mediated well-being alongside parental communication. Our results suggest the need for culturally and age-appropriate in-person interactive activities in schools and communities. Youth well-being professionals should consider incorporating societal trust-mediated well-being as part of their objectives in assessments and interventions, particularly for older adolescents.

While our study utilized data from a large-scale survey, enhancing reliability and generalizability (107), several limitations should be noted:

1. Measurement Issues:

The use of secondary data limited available measures and constructs and some measures showed suboptimal internal consistency reliability. However, this can be considered acceptable due to:

a) Short scale length, which often results in lower Cronbach’s alpha values (108).

b) Construct breadth: Social trust is a broad construct involving trust in different social entities, which may lead to lower internal consistency but improve construct validity (109).

c) Age-specific considerations, with slightly lower alpha for 18-year-olds potentially reflecting developmental changes.

d) Mean inter-item correlations (ranging from.23 to.51) may be more appropriate for short scales of societal trust (109).

2. Methodological Limitations:

Reliance on self-reported measures introduces potential response bias. The cross-sectional design precludes causal inferences. Potential confounding variables (e.g., socioeconomic status, family structure) were not adequately controlled for. In addition, the data collection during the COVID-19 period may have introduced confounding effects. The interpretation of age differences should consider cohort effects and other potential confounds. Despite using data from the nationally representative CFPS, our sample selection process may have introduced some selection bias. The exclusion of respondents with missing data, particularly those who indicated parental communication was not applicable (n=142), may have systematically removed adolescents with non-traditional family structures or unique living situations. This could potentially skew our understanding of parental influence.

3. Conceptual Limitations:

Our focus on the hedonic approach to well-being may not capture all aspects of adolescent development (110).

Future research should address these limitations by:

a) Validating the paths and mechanisms in diverse ethnic groups with more comprehensive measurement scales. Employing multiple methods such as behavioral observations, physiological measures, reports from others to reduce response bias and improve data objectivity.

b) Investigating the characteristics and outcomes of adolescents for whom traditional measures of parental communication may not apply. Exploring potential boundary conditions (e.g., social-emotional competence, social media use, family structure, socioeconomic status). Utilizing parceling techniques and advanced predictive models for feature selection.

c) Conducting longitudinal studies to examine relationships between variables over time and employing experimental designs to identify causal effects.

d) Utilizing mixed-methods approaches to gain deeper insights into adolescents’ lived experiences and the nuanced influences of societal trust on well-being.