Psychosocial predictors of adolescent stress: insights from a school-going cohort

Participants

The sample consisted of 1,156 school-going adolescents aged between 11 and 18 years (mean age=14.59) from nine educational institutions in rural and urban regions of Telangana, India. After accounting for subject attrition and missing data, the final sample size was 1,104 students. Missing data, which reduced the sample from 1,156 to 1,104 participants, were handled via listwise deletion, as the missingness was minimal (4.5 %) and random, minimizing bias (Little’s MCAR test, p>0.05).

The participants were distributed across different educational levels: Grade 6 (11.6 %), Grade 7 (12.6 %), Grade 8 (12.9 %), Grade 9 (12.4 %), Grade 10 (12.1 %), 11th grade (14.6 %), 12th grade (12.5 %), and first-year under-graduates (11.3 %). The gender distribution was nearly balanced, with 48.9 % girls and 51.1 % boys. Participants were grouped into early adolescence (11–14 years), which comprised 46.6 % of the sample, and late adolescence (15–18 years), comprising 53.4 %. Geographically, 46.4 % of participants resided in urban areas, 30.2 % in semi-urban areas, and 23.5 % in rural areas. In terms of economic background, 0.7 % came from economically poor backgrounds, 5.8 % from the lower middle class, 68.5 % from the middle class, 22.9 % from the upper middle class, and 2.1 % from the upper class.

Tools

The study used the Adolescence Stress Scale as its primary instrument, along with 11 additional psychological tools to measure various psychosocial factors contributing to stress.

Procedure

Ethical approval for the study was granted by the Institutional Ethics Committee, University of Hyderabad (Ref. No. UH/IEC/2021/174). Following this, written administrative approval was obtained from the participating educational institutions, along with informed consent from parents and informed assent from the students. Data collection was then initiated. The participants completed two tests per day in their respective classrooms. Tests requiring a re-test to establish reliability were re-administered to the participants within a fortnight.

Hierarchical multiple regression

Pearson’s correlation was calculated to identify psychosocial factors significantly associated with the overall stress levels. Of the 16 variables analysed, 12 showed significant correlations with stress levels, as presented in Table 1. Hierarchical multiple regression was employed, where predictor variables were entered into the regression equation in a predetermined order to assess the unique contribution of each predictor to the dependent variable, while controlling for other variables [25]. Table 2 provides a summary of the hierarchical multiple regression analysis for various psychosocial variables predicting stress in adolescents.

Each psychosocial factor added hierarchically to the models, resulted in nine models. Model 9 was significant, F(12, 1091)=6.77, p<0.001, explaining an additional 0.6 % of the variance (R² change=0.006, p<0.01), with a total of 6 % of the variance in stress levels accounted for (adjusted R²=0.059), indicating a modest but significant contribution of the identified predictors. While this suggests that other factors, such as genetic or environmental influences, also play a role, low explained variance is common in psychological research due to the complexity of human behavior [26]. The analysis revealed that ill health experiences (β=0.11, p<0.01), conscientiousness (β=0.12, p<0.01), emotional instability (β=0.12, p<0.001), and psychosocial support (β=0.10, p<0.01) were significant predictors of stress – offering valuable insights for designing targeted interventions, such as emotional regulation programs, despite the modest overall impact.

Serial mediation

A serial mediation model was developed using path analysis to evaluate the direct and indirect effects of psychosocial factors on stress. This process involved several steps: confirmatory factor analysis (CFA), testing a hypothetical model, and model confirmation.

Confirmatory factor analysis. The steps involved in this stage are described below in two sections: (i) measurement model, and (ii) structural model.

1. Measurement model: The measurement model is essential for structural equation modeling (SEM) and was used to validate the indicators for each construct [27]. CFA was performed on a sample of 1,104 participants, separate from the sample used for exploratory factor analysis. variables with factor loadings below 0.3, and in some cases below 0.4, were considered insufficiently representative and were removed [28]. Table 3 presents the goodness-of-fit statistics for the tools used to assess psychosocial factors and stress in adolescents.

2. Structural model: Following the measurement model, the structural model was analysed to examine the directionality and significance of the relationships. The goal was to evaluate the serial mediation model using path analysis. This serial mediation path analysis was carried out in two steps: hypothetical model and model confirmation.

Hypothetical model. A hypothetical model was constructed based on a comprehensive literature review, with social networks, psychosocial support availability, perceived physical environment, and promotive factors considered as independent variables. Family health was proposed as the first mediator, directly associated with emotional efficacy, protective factors, and emotional instability, which together act as the second mediator leading to the dependent variable, stress. The hypothetical model is depicted in Figure 1.

Figure 1:

Hypothetical model for depicting serial mediation path model between psychosocial factors and stress in adolescents.

Model confirmation. The hypothetical model was tested for confirmation by examining the direction and significance of the pathways. Necessary modifications were made to arrive at the final structural model, which is presented in Figure 2.

Figure 2:

Serial mediation path model.

In the final model, social networks, psychosocial support availability, perceived physical environment, and promotive factors were the independent variables that contributed to family health, which acted as a mediator for stress. Family health was directly related to emotional efficacy (mediator 2) and frustrative non-reward responsiveness (mediator 3). Protective factors also contributed to frustrative non-reward responsiveness, which further influenced emotional instability (mediator 4). Both mediators 3 and 4 had direct relationships with the dependent variable, stress.

The standardized estimates for all pathways are presented in Table 4.

The serial mediation model revealed significant pathways linking psychosocial factors to adolescent stress. Social networks positively influenced family health (β=0.13, p<0.001), as did psychosocial support availability (β=0.17, p<0.001). Conversely, perceived physical environment negatively affected family health (β=−0.29, p<0.001), indicating that adverse environmental perceptions reduced family health. Promotive factors also contributed positively to family health (β=0.23, p<0.001). Protective factors decreased frustrative non-reward responsiveness (β=−0.10, p<0.001), which was further reduced by better family health (β=−0.14, p<0.001). Family health positively influenced emotional efficacy (β=0.08, p<0.01), which directly reduced stress (β=−0.08, p<0.01), explaining 8 % of the variance. Frustrative non-reward responsiveness increased emotional instability (β=0.26, p<0.001), which in turn increased stress (β=0.13, p<0.001), explaining 13 % of the variance. The model’s fit indices supported its adequacy: χ²=162.813 (df=25, p=0.00), χ²/df=6.513, AGFI=0.939, CFI=0.923, RMSEA=0.071, and PCLOSE=0.000. While the ideal χ²/df is 5 or below, a value of 6.5 is acceptable given the large sample size [29]. Values of AGFI and CFI above 0.90, and RMSEA below 0.08, indicate good model fit. Thus, the model was considered acceptable.

The indirect effects of psychosocial factors on adolescent stress, as presented in Table 5, were also analyzed through the serial mediation model. All indirect effects were statistically significant (p<0.001). Specifically, social networks (β=−0.001), psychosocial support availability (β=−0.002), promotive factors (β=−0.002), protective factors (β=−0.003), and family health (β=−0.011) exhibited negative indirect effects, indicating that increases in these factors were associated with reduced stress levels. Conversely, perceived physical environment (β=0.003) and frustrative non-reward responsiveness (β=0.034) showed positive indirect effects, suggesting that higher levels of these factors were linked to increased stress. Although these indirect effects accounted for a small proportion of the variance, their significance was supported by 95 % confidence intervals that excluded zero [27]. These findings confirm partial mediation within the model, with both direct and indirect pathways contributing significantly to adolescent stress.

Hierarchical multiple regression

The final regression model revealed that ill health experiences, conscientiousness, emotional instability, and psychosocial support significantly predicted stress. Among these, emotional instability had the strongest association. However, these factors collectively explained only 6 % of the variance, suggesting that other important contributors to adolescent stress were not captured in this study.

Consistent with prior research, frustrative non-reward responsiveness and emotional instability are highly associated with negative affect. Frustration arises when adolescents do not receive expected rewards, leading to dissatisfaction and increased distress [30], 31]. Chronic stress can also negatively impact physical health, causing distress that exacerbates stress levels. Although conscientiousness is often considered protective against stress [32], high levels of conscientiousness may lead to overthinking and heightened stress due to self-imposed pressure for perfection. Emotional instability emerged as the strongest predictor reflecting a small-to-moderate effect size consistent with prior studies [33], [34], [35]; who reported similar effect sizes for emotional instability in adolescent stress. Ill health experiences and conscientiousness also showed small effect sizes, aligning with Bartley and Roesch's study [32], who noted conscientiousness effects ranging from β=0.10 to 0.14. Psychosocial support had a comparable effect, though its positive association with stress was unexpected, as discussed below.

Interestingly, the study found that psychosocial support positively predicted stress, which contradicts some previous findings. For instance Cohen and Wills [36] suggested that social support typically buffers stress, but they also noted that it can increase stress when perceived as burdensome or imposing expectations of reciprocity. The dynamics of social support have changed in recent years, with increased nuclear families and virtual social networks. High availability of virtual support may not translate to actual support, leading to elevated stress perception. For adolescents striving for autonomy, unsolicited support can be perceived as intrusive. Rui et al. [37] found that unsolicited or intrusive support, such as excessive parental monitoring, can heighten adolescent stress by undermining autonomy. Barrera [38] further observed that a discrepancy between perceived and actual support availability may lead to increased stress when adolescents’ expectations are unmet. These findings suggest that, in the context of this study, psychosocial support may have been perceived as intrusive or inadequate, contributing to elevated stress levels among adolescents striving for independence.

Serial mediation model

The serial mediation analysis revealed partial mediation through family health, frustrative non-reward responsiveness, emotional efficacy, and emotional instability, each exerting significant direct and indirect effects on adolescent stress. The path model showed that social networks, psychosocial support availability, perceived physical environment, and promotive factors influenced family health, which in turn shaped frustrative non-reward responsiveness and emotional efficacy. Protective factors mitigated frustrative non-reward responsiveness, which, when heightened, increased emotional instability, thereby elevating stress levels. In contrast, emotional efficacy reduced stress exhibiting a small but significant effect consistent with Muris [17], while emotional instability amplified stress reflecting a small-to-moderate effect size comparable to previous findings [10] (β=0.16).

Two primary pathways emerged from the model. The first pathway linked social networks, psychosocial support availability, perceived physical environment, and promotive factors to family health, which subsequently influenced emotional efficacy and consequently, stress. Family health served as a critical mediator, with secure relationships and resource availability fostering emotional awareness among adolescents [39]. Supportive parent-child relationships, indicative of positive family functioning, can buffer the impact of stressful events, as supported by Masten and Narayan [40] and Prime et al. [41]. The second pathway demonstrated that protective factors diminished frustrative non-reward responsiveness, which in turn reduced emotional instability, thereby lowering stress. Among indirect effects, frustrative non-reward responsiveness showed the strongest association with stress underscoring its pivotal role in stress pathways [42].

Despite the model’s low explained variance (6 %), the identified psychosocial factors contribute meaningfully to understanding adolescent stress. Human studies often exhibit greater unexplained variation due to the complexity of human behavior, yet significant predictors remain valuable.

Limitations

This study has limitations. Its cross-sectional design constraints causal inferences about the relationships between psychosocial factors and stress. The sample, drawn from India may restrict generalizability to other cultural contexts. Reliance on self-report measures introduces potential bias, and the modest explained variance suggests that unexamined factors, such as genetic predispositions or broader environmental influences, likely play substantial roles. Adolescent stress is a multifaceted and dynamic phenomenon, shaped by diverse life aspects that drive individual differences in perception, reactivity, and coping. Despite the low explained variance, the significant role of psychosocial factors in this study remains crucial and warrants further exploration.