Preliminary Analysis
The final sample consisted of 612 adolescents (43.46% male and 56.54% female), with an average age of 14.92 years (SD = 1.50). In terms of their sociodemographic characteristics, 564 were Chilean (92.16%) and 48 were foreign (7.84%). In terms of ethnicity, 498 adolescents (81.37%) did not identify with any specific group, while 114 reported having ethnic ancestry (18.63%). Regarding family structure, 354 lived with both parents (57.84%), 236 lived with only one parent (38.56%), and 22 did not live with either parent (3.60%). In terms of parental education, 363 adolescents (59.31%) had parents with higher education, 236 (38.56%) with secondary education, 12 (1.96%) with basic education, and 1 (0.16%) without formal education. The other variables assessed showed non-normal distributions, with medium to medium-high scores for family affluence, 5Cs of PYD, and PMH, and low scores for NMH (see Table 1). The internal consistency reliability was satisfactory for all the measurements, with Cronbach’s alpha (α) and McDonald’s omega (ω) coefficients ranging from 0.748 to 0.950.
-------------------------------------------
INSERT TABLE 1 APPROX HERE
------------------------------------------
Measurement Models of Positive Youth Development and Mental Health
The five-factor measurement model of PYD reported by Marín-Gutiérrez et al. (2024) was evaluated, which includes the correlated errors of three item pairs: items 3 and 19 (physical competence); items 14 and 31 (school connection); and items 17 and 34 (peers connection). This model showed a satisfactory fit to the data (χ2(514) = 2035.427; χ2/df = 3.959; RMSEA = 0.070 [90% CI = 0.066 – 0.073]; CFI = 0.925; TLI = 0.918; SRMR = 0.067), with significant loadings between all indicators and their corresponding latent variables (coefficients ranging from 0.389 and 0.901, p<0.001). Correlations between the latent variables indicated significant relationships between all dimensions of PYD, with coefficients ranging from 0.126 to 0.782 (p<0.001).
On the other hand, the dual-factor mental health measurement model was tested using the indicators from the PWI-SC and DASS-21, which form the dimensions of Positive Mental Health (PMH) and Negative Mental Health (NMH) respectively, as reported by Marín-Gutiérrez et al. (43). This model includes the correlated errors of two item pairs: items 4 and 7 from the PWI-SC, and items 10 and 21 (both from the depression subscale) from the DASS-21. This model showed an acceptable fit (χ2(402) = 1340.536; χ2/df = 3.334; RMSEA = 0.062 [90% CI = 0.058 – 0.065]; CFI = 0,951; TLI = 0.947; SRMR = 0.047), with significant factor loadings between all indicators and their latent variables (coefficients ranging from 0.439 and 0.890, p<0.001). Additionally, the correlation between PMH and NMH was negative and significant (r = -0,541; p<0.001).
Effects of Social Determinants on the 5Cs of Positive Youth Development and Mental Health
Two structural equation models were tested to evaluate the impact of social determinants (parental educational level, family affluence, gender, migration status, and ethnicity), with the first model focusing on the 5Cs of PYD (Figure 1A) and the second on mental health (Figure 1B). The standardized effects of both structural models are detailed in their respective panels in Figure 1.
The goodness of fit indices for the first model were adequate (χ2(659) = 2384.937; χ2/df = 3.619; RMSEA = 0.065 [90% CI = 0.063 – 0.068]; CFI = 0.914; TLI = 0.904; SRMR = 0.065. The model explained 16% of the variability in Competence (p<0.001), 9.6% in Confidence (p<0.001), 8.9% in Connection (p<0.001) and 4.2% in Caring (p=0.012). Only 2.5% of the variability in Character was explained by the model, although the relationship with social determinants did not reach statistical significance (p=0.066). In terms of effects, parental educational level had positive effects on Character (β = 0.119, SE= 0.049, p=0.015) and Caring (β = 0.136, SE= 0.046, p=0.003). Family affluence showed positive effects on Competence (β = 0.209, SE= 0.044, p<0.001), Confidence (β = 0.126, SE= 0.040, p=0.002) and Connection (β = 0.223, SE= 0.043, p<0.001). Conversely, gender had differentiated effects on the evaluated variables, being associated with a decrease in levels of Competence (β = -0.327, SE= 0.042, p<0.001), Confidence(β = -0.273, SE= 0.039, p<0.001), and Connection (β = -0.193, SE= 0.042, p<0.001), and an increase in levels of Caring (β = 0,126, SE= 0.042, p=0.003). Regarding ethnicity, it was positively associated with Connection only (β = 0.091, SE= 0.042, p=0.031).
The second model, which examined the effects of social determinants on mental health, yielded satisfactory goodness of fit indices (χ2(542)=1448.681; χ2/df = 2.672; RMSEA = 0.052 [90% CI = 0.049 – 0.056]; CFI = 0.951; TLI = 0.947; SRMR = 0.075). This model explained 14.5% of the variability in PMH (p<0.001) and 13.4% in NMH (p<0.001). Neither parental educational level nor ethnicity significantly affected adolescent mental health. However, family affluence had positive effects on PMH (β = 0,244, SE= 0.039, p<0.001) as well as a negative effect on NMH (β = -0.135, SE= 0.041, p<0.001). On the other hand, gender showed consistently unfavorable effects for adolescent girls, being associated with a reduction in their levels of PMH (β = -0.278, SE= 0.039, p<0.001), and an increase in NMH (β = 0.340, SE= 0.037, p<0.001). Surprisingly, migration status was positively related to an increase in PMH (β = 0.085, SE= 0.038, p=0.027) and a decrease in NMH (β = -0.104, SE= 0.037, p=0.005).
-------------------------------------------
Figure 1. Structural Equation Models of Social Determinants on the 5Cs of Positive Youth Development (A) and on Mental Health (B)
INSERT FIGURE 1 APPROX HERE
Note. *p<0.05; **p<0.01; ***p<0.001. The observed indicators of the latent variables have been omitted for clarity. Dashed gray arrows indicate non-significant results, so regression coefficients for these paths are not shown.
------------------------------------------
Effects of the 5Cs of Positive Youth Development on Mental Health
A structural equation model was used to assess how the 5Cs of Positive Youth Development (PYD) influence mental health. The goodness of fit indices indicated a satisfactory fit to the data (χ21926 = 4228.652; χ2/df = 2.196; RMSEA = 0.044 [90% CI = 0.042 – 0.046]; CFI = 0.932; TLI = 0.929; SRMR = 0.058), supporting its capacity to describe the relationships between the 5Cs and the dimensions of mental health. This model accounted for a significant portion of the variance in adolescent mental health, explaining 67.7% for PMH (p<0.001) and 37.5% for NMH (p<0.001).
The standardized effects, detailed in Figure 2, show that only Confidence and Connection from PYD had an influence on mental health measures, underscoring their relevance as key predictors of mental health in Chilean adolescents. Specifically, Confidence had a direct positive effect on PMH (β = 0.394, p<0.000) and a negative effect on NMH (β = -0.480, p<0.001). Similarly, Connection had a direct positive effect on PMH (β = 0.454, p< 0.001) and a negative effect on NMH (β = -0.235, p=0.006).
-------------------------------------------
Figure 2. Structural Equation Model of the 5Cs of Positive Youth Development on Mental Health
INSERT FIGURE 2 APPROX HERE
Note. ***p<0.001. The observed indicators of the latent variables have been omitted for clarity. Dashed gray arrows indicate non-significant results, so regression coefficients for these paths are not shown.
-------------------------------------------
Mediating Role of Positive Youth Development Between Social Determinants and Mental Health
To assess the mediating role of Positive Youth Development (PYD), a structural equation model (SEM) was estimated, incorporating only those variables that showed significant effects on mental health in previous analyses. Specifically, gender, migration status, and family affluence were included as predictors, while the PYD dimensions of Confidence and Connection were analyzed as mediators, with PMH and NMH as dependent variables. A bootstrap procedure with 5000 samples was used to strengthen the confidence interval estimates. The goodness of fit indices demonstrated the adequacy of the model (χ2(1012) = 2475.606, χ2/df = 2.446, RMSEA = 0.049 [90% CI= 0.046 – 0.051], CFI = 0.942, TLI = 0.938, SRMR = 0.058). This last one model explained 69.4% of the variance in PMH (p<0.001) and 38.1% in NMH (p<0.001). Detailed results on the direct and indirect effects from this model are presented in Figure 3 and Table 2, respectively.
-------------------------------------------
Figure 3. Direct Effects of Social Determinants and Positive Youth Development on Mental Health
INSERT FIGURE 3 APPROX HERE
Note. *p< 0.05; **p< 0.01; ***p< 0.001. The observed indicators of the latent variables have been omitted for clarity. Dashed gray arrows indicate non-significant results, so regression coefficients for these paths are not shown.
------------------------------------------
In terms of direct effects, family affluence showed a significant impact on Confidence (β=0,143, SE=0.043 [95% CI: 0.057 – 0,230], p<0.001), Connection (β=0,218, SE=0.045 [95% CI: 0,128 – 0,307], p<0.001), and PMH (β=0,112, SE=0.028 [95% CI: 0.056 – 0,169], p<0.001). Additionally, partial mediation was observed in its effect on PMH through Confidence (β=0.063, SE=0.020 [95% CI: 0.027 – 0,108], p=0.002), and Connection (β=0.086, SE=0.021 [95% CI: 0.049 – 0,133], p<0.001), resulting in a total effect on PMH of 0.261 (SE=0.040 [95% CI: 0,180 – 0,340], p<0.001).
In contrast to previous analyses (Figure 1B), family affluence lost its direct statistical significance on NMH (β=-0.043, SE=0.038 [95% CI: -0.121 – 0.034], p=0.262). In this context, Confidence emerged as a full mediator in this relationship (β=-0.063, SE=0.021 [95% CI: -0.111 – -0.028], p=0.002), while Connection did not show a significant indirect effect on NMH (β=-0.023, SE=0.017 [95% CI: -0.064 – 0.005], p=0.183). Overall, the total effect of family affluence on NMH was -0.129 (SE=0.042 [95% CI: -0.214 – -0.046], p=0.002).
On the other hand, being female had a direct negative impact on Confidence (β=-0.278, SE=0.038 [95% CI: -0.352 – -0.201], p<0.001), Connection (β=-0.186, SE=0.043 [95% CI: -0.268 – -0.101], p<0.001), and PMH (β=-0.083, SE=0.029 [95% CI: -0.139 – -0.026], p=0.004), as well as a positive effect on NMH (β=0,199, SE=0,34 [95% CI: 0,132 – 0,266], p <0.001).
Partial mediation was identified in the effect of gender on PMH through reductions in Confidence (β=-0,123, SE=0.023 [95% CI: -0,172 – -0.082], p<0.001) and Connection (β=-0.073, SE=0.020 [95% CI: -0,117 – -0.038], p<0.001). Overall, the total effect of gender on PMH was -0.279 (SE=0.038, [95% CI: -0,347 – -0,200], p<0.001). Regarding the indirect effect of gender on NMH through the dimensions of PYD, Confidence was identified as a partial mediator, (β=0,122, SE=0.025 [95% CI: 0.077 – 0,177], p<0.001), indicating that a decrease in Confidence among females is associated with an increase in mental health problems. In contrast, Connection did not play a significant mediating role (β=0.020, SE=0.014 [95% CI: -0.005 – 0.055], p=0,174). Consequently, the total effect of gender on NMH reached a β coefficient of 0.341 (SE= 0.035 [IC = 0,273 – 0,407], p<0.001).
Regarding migration status, a direct negative effect was observed on NMH (β=-0.072, SE=0.034 [95% CI: -0,139 – -0.005], p=0.036), indicating a statistically significant difference in NMH levels, with migrant adolescents showing lower levels compared to their Chilean peers. No significant indirect effects of migration status on mental health were detected, mediated by the dimensions of PYD considered in the model.
-------------------------------------------
INSERT TABLE 2 APPROX HERE
-------------------------------------------