The influence of emotional intelligence on life satisfaction in adolescence: partial mediation of resilience, positive and negative affect.

Original Article The Influence of Emotional Intelligence on Life Satisfaction in Adolescence Partial Mediation of Resilience, Positive and Negative Affect Lorea Azpiazu 1 , Arantzazu Rodríguez-Fernández 2 , Eider Goñi 2 , and Arantza Fernández-Zabala 2 1 Faculty of Education, Philosophy and Anthropology, Evolutionary and Educational Psychology, University of the Basque Country (UPV/EHU), Donostia-San Sebastian, Spain 2 Faculty of Education and Sport, Evolutionary and Educational Psychology, University of the Basque Country (UPV/EHU), Vitoria-Gasteiz, Spain Abstract: Introduction: Adolescence is a developmental stage during which adolescents often report lower levels of well-being. Understanding the factors that enhance subjective well-being is essential to its improvement. However, the specific relationships between emotional intelligence, resilience, and subjective well-being remain unclear. Aim: This study compares six theoretical models based on preliminary research exploring the relationship dynamics among emotional intelligence, resilience, and subjective well-being. Methods: Participants included 1,397 schoolchildren aged 12 to 16 years (M= 13.88; SD = 1.27). Results: The findings support a sequential model in which emotional intelligence directly predicts both positive (emotional clarity and repair) and negative (emotional attention and clarity) effects. Emotional intelligence also indirectly influences these variables through resilience, while resilience directly and indirectly predicts life satisfaction via the affective domain. Discussion: The results reveal a stepwise relationship dynamic, highlighting the mediational role of resilience and affective balance. Additionally, the affective components of subjective well-being play distinct roles compared to life satisfaction when analyzed alongside other wellbeing-related variables in a multivariate framework. This study opens up new avenues for enhancing adolescent life satisfaction. Keywords: emotional intelligence, resilience, positive affect, negative affect, life satisfaction Traditionally, much psychological research and practice has focused on mental illness and human deficits, ignoring aspects of well-being and the promotion of positive functioning (Chakhssi et al., 2018). This changed with the advent of Positive Psychology, because a shift occurred toward exploring the factors that promote well-being and optimal functioning. According to this positive approach, one key indicator of mental health is subjective well-being (Magyar & Keyes, 2019;Yıldırım et al., 2022), which refers to the hedonic nature of well-being, defined as an individual’s overall evaluation of their life and emotional experiences (Diener et al., 2017). In conceptual terms, it is a three-part model with three associated elements (Diener, 1984; Diener, 2009; Pavot & Diener, 2013): life satisfaction (global cognitive assessments of how satisfied the respondent is with their life), positive affect (frequent positive feelings), and negative affect (infrequent negative feelings). There is a general consensus regarding the need to determine which factors facilitate subjective well-being in adolescence (Magyar & Keyes, 2019), since the neurobiological and psychosocial changes experienced at the onset of puberty appear to lead to a significant decline in this area (González-Carrasco et al., 2017). Among these factors, emotional intelligence and resilience stand out. Resilience is defined as the ability to cope with and effectively adapt to stressors (Southwick et al., 2014; Windle et al., 2011). Both factors are particularly important because of their impact on subjective well-being (Noble & McGrath, 2012; Ramos-Díaz et al., 2019; Sánchez-Álvarez et al., 2016), althoughthistopichasbeenlessstudiedinadolescentsthan in adults (Tian et al., 2015) and requires further scientific research (Jebb et al., 2020; Sánchez-Álvarez et al., 2016). Perceived emotional intelligence (PEI) is a person’s perception of their own emotional skills and is linked to emotional attention, emotional clarity, and emotional repair (or emotion regulation) (Mayer et al., 2000; Salovey et al., 2002). Consequently, in this paper, we conceptualize PEI in accordance with the correlated three-factor model (emotional attention, emotional clarity, and emotional repair), Ó2025 The Author(s). Distributed as a Hogrefe OpenMind article under the European Journal of Psychology Open (2025) under the license CC BY-NC-ND 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0) https://doi.org/10.1024/2673-8627/a000084 https://econtent.hogrefe.com/doi/pdf/10.1024/2673-8627/a000084 - Wednesday, October 22, 2025 1:28:30 AM - IP Address:95.18.210.110 which is still accepted and continues to arouse a great deal of scientific interest (Blasco-Belled et al., 2022; CobosSánchez et al., 2020; Martínez-Monteagudo et al., 2021). Althoughpreviousresearch(Dietal.,2020;Prado-Gascó et al., 2018; Sánchez-Álvarez et al., 2016) generally supported the relationship between PEI and the three components of subjective well-being –suggesting that lower emotional attention and higher emotional clarity and regulationenhancelifesatisfactionandaffectivebalance–some studies qualified these findings. For example, one metaanalysis (Sánchez-Álvarez et al., 2016) mentions the need for further research to clarify the underlying mechanisms relating PEI to subjective well-being, given that few studies to date include the affective components of subjective wellbeing (Jebb et al., 2020; Zhao et al., 2020). Moreover, the different components of PEI appear to have no direct impact on life satisfaction when studied inconjunction with other variables (Estévez et al., 2020; Vergara et al., 2015) fundamental to subjective well-being, such as resilience (Bajaj & Pande, 2016; Zhao et al., 2020). Emotion regulation has been shown to directly influence life satisfaction through resilience (Azpiazu et al., 2021; Ramos-Díaz et al., 2019). However, the reported relationship coefficients are typically low and often diminish when theaffectivecomponentsofsubjectivewell-beingareincorporated into the analysis (Zhao et al., 2020). In this context, Zeidner et al. (2012) propose a Model of Affective Mediation, which posits that positive and negative affect mediate the relationship between perceived emotional intelligence (PEI) and life satisfaction. According to this model, individuals with high emotional intelligence experience greater positive affect and lower negative affect, leading to enhanced life satisfaction (Eid & Larsen, 2008; Kong et al., 2019; Xiang et al., 2021). This perspective contrasts with Diener’s well-being model (Diener, 1984; Pavot & Diener, 2013; RodríguezFernández & Goñi, 2011), in which life satisfaction is treatedasindependentfromaffectivecomponents,emphasizing the autonomy of each well-being dimension. Regarding the specific components of PEI, Vergara et al., 2015 demonstrated that emotional attention positively influencesnegativeaffect,whileemotionalrepairenhances positive affect. Additionally, emotional clarity affects both affective components, with a positive impact on positive affect and a negative impact on negative affect. However, it remains unclear whether these relationships persist when resilience is factored into the model. Evidence suggests that resilience promotes positive affect and reduces negative affect (Bajaj & Pande, 2016; Yang et al., 2022). Zhao et al. (2020) identified full mediation by resilience in the link between PEI and both types of affect. In contrast, other studies have reportedonly partial mediation between PEI and life satisfaction (Bajaj & Pande, 2016; Ramos-Díaz et al., 2019). Notably, these studies focus solely on the cognitive component of life satisfaction, with some relying on constructs that approximate rather than directly measure PEI. These discrepancies cast doubt on whether resilience fully or partially mediates this dynamic. Giventhisevidence,astaggeredrelationshipseemsplausible (Armstrong et al., 2011; Liu et al., 2013), where resilience mediates the association between PEI and subjective well-being.However,furtherclarificationisneeded.Specifically, does resilience impact all three components of wellbeing, consistent with Diener’s independent component framework (Pavot & Diener, 2013)? Or do the affective components mediate the relationship between resilience andlifesatisfaction,asstudiesincorporatingadditionalpsychological variables suggested (Kong et al., 2019; Xiang et al., 2021)? If the latter is true, it remains to be determined whether this mediation is partial (Luque-Reca et al., 2022; Yang et al., 2022) or full (Liu et al., 2013). It is also necessary to explore the relationship dynamic itself, taking into account all components of PEI –such as emotionalattention,emotionalclarity,andemotionalregulation (and well-being –positive affect, negative affect, and life satisfaction) since the few studies that analyze these variables in a multivariate way include only global constructs measures or analyze only a single component (Luque-Reca et al., 2022; Ramos-Díaz et al., 2019) rather thanalltheirspecific components(Liuetal.,2013).Alternatively, they fail to analyze relationships between variables that previous research showed to be relevant (Bajaj & Pande, 2016). This study addresses these gaps by providing a more comprehensive analysis that integrates not only cognitive but also affective components of well-being, thereby offering a more nuanced understanding of the dynamics of the relationships between PEI, resilience, and subjective well-being. Thepresentstudyhadatwofoldaim:(1)toexplorethefull or partial influence of resilience on the association between PEI and well-being (negative affect, positive affect, and life satisfaction);(2)toanalyze thepossiblemediatingrole(partial or full) played by the affective components ofsubjective well-being (negative affect and positive affect) in the relationship between PEI and life satisfaction. To this end, and based on previous research,we proposed to test the following hypotheses through the six structural models shown in Figure 1. -Hypothesis 1: Resilience fully mediates the relationship between PEI and the three components of subjective well-being. M 1a. Constrained model of resilience. -Hypothesis 2: Resilience partially mediates the relationship between PEI and the three components of subjective well-being. M 1b. Partial model of resilience. European Journal of Psychology Open (2025) Ó2025 The Author(s). Distributed as a Hogrefe OpenMind article under the under the license CC BY-NC-ND 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0) 2 L. Azpiazu et al., Facilitators of Adolescent Well-Being https://econtent.hogrefe.com/doi/pdf/10.1024/2673-8627/a000084 - Wednesday, October 22, 2025 1:28:30 AM - IP Address:95.18.210.110 -Hypothesis 3: Resilience fully mediates the relationship between PEI and the two affective components of subjective well-being, and, at the same time, these predict life satisfaction. M2 . Constrained model. -Hypothesis 4: Resilience fully mediates the relationship between PEI and the three components of subjective well-being, and, at the same time, negative affect and positive affect predict life satisfaction. M3 a. Partially restricted model 1. -Hypothesis 5: Resilience partially mediates the association between PEI and the two affective components of subjective well-being, and, at the same time, negative affect and positive affect predict life satisfaction M3 b. Partially restricted model 2. -Hypothesis 6: Resilience partially mediates the association between PEI and the three components of subjective well-being, and, at the same time, negative affect and positive affect predict life satisfaction. M4 . Unrestricted model. Method Participants A total of 1397 schoolchildren from 10 different schools in compulsory secondary education, between 12 and 16 years, participated in the study (M=13.88;SD =1.27). Of these, Figure 1. Conceptual diagrams of proposed theoretical models. EA = emotional attention, EC = emotional clarity, ER = emotional repair, RS = resilience, PA = positive affect, NA = negative affect, LS = life satisfaction. Ó2025 The Author(s). Distributed as a Hogrefe OpenMind article under the European Journal of Psychology Open (2025) under the license CC BY-NC-ND 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0) L. Azpiazu et al., Facilitators of Adolescent Well-Being 3 https://econtent.hogrefe.com/doi/pdf/10.1024/2673-8627/a000084 - Wednesday, October 22, 2025 1:28:30 AM - IP Address:95.18.210.110 670 (48%) were male and 727 (52.0%) female (w 2 =2.33, p=.127). All came from public (832 students) and semiprivate schools (i.e., private schools that receive some state funding) (565 students) in the Autonomous Community of the Basque Country. The families of the participating students predominantly belonged to a middle socioeconomic and cultural stratum (73.2%), with smaller proportions classified as low (13.4%) and high (13.3%) socioeconomic status. We employed a convenience sampling method. Procedure This research was approved by the Ethics Committee for Research on Human Subjects of the University of the Basque Country (EHU/UPV) (M10_2018_261). We contacted the headteachers and deputy heads of different schoolsbytelephone.Afteragreeingonadateforameeting, they were informed of the aims of the study and were asked to take part and to give their authorization. A considerable numberofschoolsagreedtoparticipate,whilefourdeclined because of administrative constraints, scheduling conflicts, or institutional priorities. Ultimately, a total of 10 schools participated in the study. Since all participating students were minors, we requested informed consent from their parents or legal guardians. Once the school management had agreed to participate, we sent a written notification to thefamilies,specifyingthattheirchildrenwouldparticipate in the study and requesting their informed consent through the signature of a document specifying the objectives of the research, the questionnaires to be applied, the voluntary nature of participation, and the possibility of accessing the data in the future to remove the data of their son or daughter. Only students who had signed the informed consent formcouldvoluntarilyparticipateintheresearch.Thequestionnaireswerecompletedonpaperintheclassroomduring schoolhours,inasessionlastingapproximately40 minutes. To ensure the rigor of the data collection process, the researchers administering the questionnaires were highly experienced and knowledgeable in this task, with solid training in ethical guidelines, data confidentiality, and the standardized administration of the questionnaires. They were also responsible for answering any questions from the participants during data completion in a consistent and unbiased manner, ensuring the reliability and uniformity of the data collection process across all participants. This study applied “single blinding,”meaning that participants were unaware of the specific objectives of the research. The purpose of this blinding was to ensure that participants’responses were not influenced by prior knowledge of the study’s aims, preventing them from consciously or unconsciously attempting to confirm or reject the study hypotheses. This approach was crucial to maintaining the integrity of the data collection process. Response anonymity and voluntary participation were also assured. Instruments We assessed perceived emotional intelligence (PEI) using the Trait Meta Mood Scale–23 (TMMS; Salovey et al., 1995), adapted to Spanish and abridged by Salguero et al. (2009). This 12-item scale consists of three correlated dimensions: Emotional Attention, Emotional Clarity, and Emotional Repair. Items are rated on a Likert-type scale with five response options ranging from 1(strongly disagree) to 5(strongly agree). In this study, we found acceptable fit indexes for the factor structure of the questionnaire: CFI = .935, IFI = .935, TLI = .914, RMSEA [CI] =.064 [.058–.071] , SRMR = .067 and χ 2 /df =6.74; and the following reliability indexes: emotional attention: α=.812, H coefficient = .806; emotional clarity: α=.777, H coefficient = .795; and emotional repair: α=.696, H coefficient = .800. To assess resilience, we used the Connor-Davidson Resilience Scale-10 (CD-RISC10; Campbell-Sills & Stein, 2007), adapted to Spanish by Notario-Pacheco et al. (2014). The scale comprises 10 items rated on a Likert-type scalewithfive response options ranging from1(strongly disagree)to5(strongly agree). The confirmatory factor analysis revealed the following fit indexes for this unidimensional scale: CFI = .959, IFI = .959, TLI = .939, RMSEA [CI] = .052 [.039–.064] , SRMR = .033 and w 2 /df = 4.7; and the following reliability indexes: α=.734, and H coefficient = .750. We measured life satisfaction using the Satisfaction with Life Scale (SWLS; Diener et al., 1985), validated in Spanish by Atienza et al. (2000). This five-item unidimensional scale has a Likert-type response format with seven optionsrangingfrom1=stronglydisagreeto7=stronglyagree. In this study, the results confirmed the unidimensional structure of the scale: CFI = .983, IFI = .983, TLI = .966, RMSEA [CI] =.059 [.040–.081] , SRMR = .023,w 2 /df = 5.8; and revealed acceptable reliability indexes: a=.826 and H coefficient = .858. We assessed positive and negative effects using the Positive and Negative Affect-10 scale (PNA-10; Bradburn, 1969),revisedbyWarretal.(1983)andvalidatedinSpanish by Yárnoz-Yaben et al. (2014). This 10-item two-dimensional questionnaire directly measures the experience of both positive and negative effects. It has a Likert-type format with four response options ranging from 1=never to 4 =all the time. Acceptable fit indexes resulted in the confirmatory factor analysis of this two-factor correlated scale: TLI = .935, CFI = .951, IFI = .930, RMSEA [CI] =.059 [.040– .081] , SRMR = .056,w 2 /df = 5.8. The reliability values obtained for each subscale were: positive affect: α=.797, H coefficient = .848; and negative affect: α=.781, H coefficient = .792. European Journal of Psychology Open (2025) Ó2025 The Author(s). Distributed as a Hogrefe OpenMind article under the under the license CC BY-NC-ND 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0) 4 L. Azpiazu et al., Facilitators of Adolescent Well-Being https://econtent.hogrefe.com/doi/pdf/10.1024/2673-8627/a000084 - Wednesday, October 22, 2025 1:28:30 AM - IP Address:95.18.210.110 Data Analysis We calculated the missing values (2.1%) using the expectationmaximization(EM)algorithmandMarkovchainMonte Carlo (MCMC), both offered by the LISREL 8.8program. We eliminated a total of 273 outliers using the SAS program, so the final base was 1,397 participants without missing values. Participants had to be between 12 and 16 years of age; we established this inclusion criterion prior to analysis. The K-S test revealed a significant result (p<.001), indicating that the data did not follow a univariate normal distribution. Similarly, Mardia’s test for multivariate normality showed a significant value (Mardia = 179.32,Z= 67.75), with a critical threshold for significance being Z> 5.0(Mardia, 1970). These results confirmed that the assumption of normality was violated, leading us to reject this assumption. Given the violation of normality, we applied robust estimators to ensure accurate model estimation despite nonnormality. Specifically, we employed robust goodness-of-fit indices in EQS 6.3, which are statistical measures used to evaluate how well a model fits the data, even when assumptions of normality are violated (Byrne, 2006; Hu & Bentler, 1999). These robust estimators help to provide reliable parameter estimates and model fit indices, even in the presence of skewed distributions or outliers (Byrne, 2006; Hu & Bentler, 1999). This approach ensures that our conclusions are valid despite the nonnormality observed in the data. Parametric procedures were applied using the SPSS 22 program, since they remain robust in the event of noncompliance with the normality assumption (Montilla & Kromrey, 2010). The results from Harman’s exploratory factor analysis indicated that the unrotated solution explained 16.6%of the total variance. In line with the criteria commonly applied in this type of analysis, a variance explanation of less than 20% is generally considered low (Harman, 1976), suggesting that a single underlying factor does not dominate the data. This low value implies that common method bias is unlikely to be a significant issue in this study. In addition, we applied the unmeasured latent construct technique, which revealed a common variance of 7.3%. This value is also considered relatively low (Podsakoff et al., 2012), further supporting the conclusion that common method bias has a minimal impact on the study’s results. The low common variance indicates that the variance shared between the variables is not substantial enough to distort the findings. We used the SEM methodology to estimate the measurement and structural model, following the criteria proposed by Byrne (2006). We examined the residual covariance matrix and the most common measures of fit (McDonald & Ho, 2002). To determine whether a model adequately fit the data, we considered specific cut-off values: We deemed CFI, TLI, and IFI values of .90 acceptable, while we considered values of .95 optimal. Similarly, we classified RMSEA and SRMR values .06 as good and values .08 as acceptable, while we regarded a w 2 /df ratio <3as indicative of a reasonable fit (Hu & Bentler, 1999; Marsh & Hocevar, 1985). Once we had identified models meeting acceptable fit thresholds, we conducted model comparisons to determine whether they represented statistically different structures. We assessed this using the chi-square difference test (Bryant & Satorra, 2012): A significant result indicated that the models were distinct, necessitating further evaluation based on fit indices and model parsimony. We also examined the comparative AIC (Akaike, 1987) andCAIC(Bozdogan,1987)indices,giventhatlowervalues indicateabetterandmore parsimonious model (West etal., 2014).Throughthisprocess,weselectedthemostappropriate model by considering both the absolute fit of the model and its relative superiority over alternative specifications, Finally, to test the type of mediation, we followed the steps outlined by VanderWeele and Vansteelandt (2014) and Vansteelandt and Daniel (2017) for mediation analysis involving multiple mediators in series with parallel characteristics. We chose This approach because of its ability to properly account for the complexity of mediation models that include both parallel and serial mediators. It is particularly advantageous because it allows for the estimation of both direct and indirect effects while controlling for potential confounders, making it more robust than simpler mediation models. This method is especially well-suited for studies involving multiple mediators, where understanding the relationships between them is crucial to explaining the overall process. We adhered to the key assumptions of Holmbeck’s(2002) framework for mediational analysis, which include (1) the direct noninterventional model: The direct relationship between the predictor and the outcome is evaluated without considering the mediator; (2) the multiple indirect effects model: The mediator is introduced to assess how the predictor influences the outcomethrough it; (3) the multiple mediator effects model: Both direct and indirect effects are jointly analyzed to fully understand the relationships between the variables. Results Descriptive Statistics Table 1shows the means, standard deviations, and correlations between the study variables. We observed significant positive associations between all variables except in the case of emotional attention and life satisfaction (r=.003, p=< .001), for which the association was negative. We also found negative associations between negative affect and emotional clarity (r=.107,p=< .001), Ó2025 The Author(s). Distributed as a Hogrefe OpenMind article under the European Journal of Psychology Open (2025) under the license CC BY-NC-ND 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0) L. Azpiazu et al., Facilitators of Adolescent Well-Being 5 https://econtent.hogrefe.com/doi/pdf/10.1024/2673-8627/a000084 - Wednesday, October 22, 2025 1:28:30 AM - IP Address:95.18.210.110 emotionalrepair(r=.150,p=<.001),resilience(r=.229, p=< .001), positive affect (r=.105,p=< .001), and life satisfaction (r=.414,p=< .001). Preliminary Analysis Before testing the structural models shown in Figure 1, we developed a measurement model. The results revealed acceptable fit indexes: w 2 [df] =1631.11 [505] ,w 2 /df = 3.23, TLI=.903,CFI=.912,IFI=.913,SRMR=.055,RMSEA [CI] = .040 [.038–.024] . We also examined the conditions proposed by Holmbeck (2002). First, we tested the estimation and fit of the direct noninterventional model of PEI as a predictor of life satisfaction, positive affect, and negative affect as a prerequisite for testing the M 1a and M 1b models. The results revealed that this model did not return satisfactory values in all the fit indexes: w 2 [df] =1431.43 [311] ,w 2 /df =4.6, TLI = .884, CFI = .897, IFI = .897, SRMR = .073, RMSEA [CI] =.051 [.048–.053] , indicating that M 1a and M 1b are probably not plausible, and that the pathways found in this model should be interpreted with caution in the presence of acceptable fit indexes. We also tested thedirect noninterventional model of PEI as a predictor of life satisfaction as a prerequisite for analyzing the M 2 M 3a, M 3b , and M 4 structural models (Holmbeck, 2002). Indeed, the model hypothesizing the direct influence of emotional attention (β d =.09,z= 2.362,p<.01), emotional clarity (β d =.23,z=5.395,p< .01) and emotional repair(β d =.36,z=9.046,p<.01) on life satisfaction had significant coefficients and an acceptable fit to the data: w 2 [df] =594.04 [112] ,w 2 /df =5.30, TLI = .911, CFI = .927, IFI = .927, SRMR = .063, RMSEA [CI] = .055 [.051–.060] . Next, following Vansteelandt and Daniel (2017)–and since models M 2 ,M 3a ,M 3b , and M 4 were multiple mediations –we tested the fit and estimation of the indirect coefficients of those models in which: (1) PEI predicted life satisfaction through resilience (w 2 [df] = 934.41 [244] ,w 2 /df =3.83, TLI = .907, CFI = .917, IFI = .918, SRMR = .058, RMSEA [CI] =.045 [.042–.048] ); and (2) PEI predicted life satisfaction through the affective components of subjective well-being (w 2 [df] =1407.71 [312] , w 2 /df =4.51, TLI = .907, CFI = .917, IFI = .917, SRMR = .053, RMSEA [CI] =.064 [.047–.053] ). This also served to verify Holmbeck (2002) second step (multiple indirect effects model) and third step (multiple mediator effects model). Both models had acceptable fit indexes, and the indirect effects were significant at .01 (Table 2), which made it possible to test the constrained model (M 2 ) as well as its respective partially constrained variants (M 3a and M 3b ) and the unconstrained model (M 4 ), all of which are shown in Figure 1, and to interpret the significance of the pathways found with relative confidence. Structural Models Theanalysisofthegoodness-of-fitindexesshowninTable3 revealedthatonlymodelsM 3b andM 4 obtainedsatisfactory indexes. Although the TLI of M 3b was below the cut-off point, the rest of the indices were considered acceptable. We therefore decided to reject the theoretical models M 1a, M 1b, M 2 , and M 3a . Consequently, we rejected Hypotheses 1,2, and 3and accepted Hypothesis 4. Table 1.Correlations and descriptive statistics Variables 1234567 1. Emotional attention ––––––– 2. Emotional clarity .275 *** ––– –– 3. Emotional repair .187 *** .438 *** ––––– 4. Resilience .074 ** .441 *** .502 *** –––– 5. Positive affect .113 *** .406 *** .425 *** .496 *** ––– 6. Negative affect .303 *** .107 *** .150 *** .229 *** .105 *** –– 7. Life satisfaction .003 *** .328 *** .383 *** .481 *** .504 *** .414 *** – M(SD) 24.45 (7.11) 25.10 (6.66) 27.60 (6.72) 30.54 (4.33) 26.12 (4.72) 18.77 (4.72) 26.20 (5.63) Note. *p< .05, **p< .01, ***p< .001. Table 2.Indirect effects Model IV !(MV) !DV β i z i Resilience as a mediating variable EA!(RS) !LS .098 ** 3.960 EC!(RS) !LS .215 ** 6.014 ER!(RS) !LS .244 ** 7.147 Affective balance as a mediating variable EA!(NA, PA) !LS .230 ** 5.889 EC!(NA, PA) !LS .234 ** 6.037 ER!(NA, PA) !LS .285 ** 7.794 Note. *p < .05, **p< .01. EA = emotional attention, EC = emotional clarity, ER = emotional repair, RS = resilience, PA = positive affect, NA = negative affect, LS = life satisfaction. European Journal of Psychology Open (2025) Ó2025 The Author(s). Distributed as a Hogrefe OpenMind article under the under the license CC BY-NC-ND 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0) 6 L. Azpiazu et al., Facilitators of Adolescent Well-Being https://econtent.hogrefe.com/doi/pdf/10.1024/2673-8627/a000084 - Wednesday, October 22, 2025 1:28:30 AM - IP Address:95.18.210.110 The analysis of the goodness-of-fit indexes for the remaining models (M 3b and M 4 ) revealed overall good levels of fit, although M 4 (w 2 [df] =1669.68 [511] ,w 2 /df = 3.26, TLI = .907, CFI = .910, IFI = .910, SRMR = .056, RMSEA [90% CI] =040 [.038–.042] ) returned better indexes than M 3b , which reported an unacceptable TLI (w 2 [df] = 1724.03 [512] ,w 2 /df =3.36, TLI = .897, CFI = .906, IFI = .906, SRMR = .060, RMSEA [90% CI] =041 [.039–.043] ). The chi-square test on the discrepancy between the two models (Δw 2 M3b-M4 =68.78,p<.001) was statistically significant. Therefore,ofthesetwomodels,weselectedtheonereflecting a partial mediation by the affective components of subjective well-being (M 4 ) as the most plausible and the one that best represents the data. The comparative indexes support this choice, since M 4 was the model with the lowest coefficients (AIC = 647.68, CAIC = 2542.38). These data, together with the preliminary analyses, allowed us to reject Hypothesis 5and accept Hypothesis 6. In short, taken together, the results indicate that resilience partially mediates the relationship between PEI and the affective components of subjective well-being, and in turn, positive and negative affect partially mediate the influence of resilience on life satisfaction. Moreover, the pathways specified in the model were statistically significant, and the LM test and Wald test modification indexes do not indicate structural links that would suggest the addition or subtraction of parameters. Figure 2shows the final structural model, with its corresponding standardized regression coefficients and significance levels. The amount of variance explained by resilience in the model is 42.4%, negative affect is explained by 30.8%, positive affect by 31.8%, and life satisfaction by 52.1%. Direct, Indirect, and Total Effects of the Variables Studied The regression coefficients of the most plausible model (M 4 ) (Table 4) revealed that all the proposed direct pathways were significant at p<.01. Theresultsrevealtherelevanceofallthreedimensionsof PEI on resilience, with the effect of emotional repair (β d = .47,p<.01) having the strongest direct influence of all the direct and indirect relationships analyzed. Likewise, all three dimensions of PEI were found to have a direct and significant influence on the affective components of subjective well-being, with effect sizes of between .14 and .21, with the exception of emotional attention on negative affect, in which the effect size was almost double (β d =.41,p<.01). Moreover, although it did not exert a direct influence, emotional attention indirectly Table 3.Comparison of nested models w 2[df] w 2 /df TLI CFI IFI SRMR RMSEA [90% CI] AIC CAIC M 1a 2006.01 [517] 3.88 .874 .884 .885 .072 .045 [.043–.047] 972.01 2255.51 M 1b 1838.35 [512] 3.59 .887 .896 .896 .072 .043 [.041–.045] 814.35 2381.96 M 2 1889.44 [516] 3.66 .884 .893 .894 .071 .044 [.042–.046] 857.44 2363.83 M 3a 1836.00 [515] 3.57 .888 .897 .898 .067 .043 [.041–.045] 806.00 2409.03 M 3b 1724.03 [512] 3.36 .897 .906 .906 .060 .041 [.039–.043] 700.04 2496.27 M 4 1669.68 [511] 3.26 .901 .910 .910 .056 .040 [.038–.042] 647.68 2542.38 Δw 2M3b–M4 68.78, p< .001 Note. CFI, TLI, IFI .90 (acceptable), .95 (optimal); RMSEA, SRMR .06 (good), .08 (acceptable); w 2 /df < 3 (reasonable fit) (Hu & Bentler, 1999; Marsh & Hocevar, 1985). Figure 2. Standardized solution. *p< .05, **p< .01. Ó2025 The Author(s). Distributed as a Hogrefe OpenMind article under the European Journal of Psychology Open (2025) under the license CC BY-NC-ND 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0) L. Azpiazu et al., Facilitators of Adolescent Well-Being 7 https://econtent.hogrefe.com/doi/pdf/10.1024/2673-8627/a000084 - Wednesday, October 22, 2025 1:28:30 AM - IP Address:95.18.210.110 andnegativelyinfluencedpositiveaffect(β i =.04,p<.01), andemotionalrepairinfluencednegative affect (β i =.18,p <.01),althoughinboth casesthecoefficientswereverylow. Thedatarevealedthatresilienceindirectlyinfluenceslife satisfaction (β i =.22,p<.01) through the affective components of well-being, indicating a partial mediation of both positive and negative affect in this relationship dynamic. Furthermore, the total effect of resilience on life satisfaction (β t =.51,p<.01) had the highest regression coefficient of all the relationships studied. Also noteworthy was the indirect effect of emotional repair on life satisfaction (β i= .31,p<.01). Discussion There is a widespread consensus among researchers about the need to explore the factors that foster subjective wellbeing in adolescence (Magyar & Keyes, 2019), given that the biological and psychosocial changes experienced at the onset of puberty appear to result in a significant decline in this variable (González-Carrasco et al., 2017). Studies to date have found that PEI and resilience are two important variables associated with subjective wellbeing (Prado-Gascó et al., 2018; Ramos-Díaz et al., 2019). However, the specific means whereby all these variables interact with each other is as yet unclear, as some studies included only a global measure of PEI or some of its facets (Bajaj & Pande, 2016; Liu et al., 2013; Luque-Reca et al., 2022), while others failed to analyze the affective components of well-being (Ramos-Díaz et al., 2019) or focused only on affective balance (Liu et al., 2013). Consequently, this study, framed within the Positive Psychology approach, sought to clarify how these variables interact with each other at a developmental stage that has been relatively neglected by research compared to adulthood (Tian et al., 2015). Based on previous research (Bajaj & Pande, 2016; Liu et al., 2013; Ramos-Díaz et al., 2019), we tested six theoretical models: first, to verify whether affectivecomponents,togetherwithresilience,playamediating role in the relationship between PEI and life satisfaction; and second, to ascertain the type of mediating role (partial or full) played by resilience in the associations between the different variables studied. Ingeneral,weobservedastaggeredrelationshipdynamic (this model reflects a process consisting of four psychological components that seek to explain how emotional intelligence, resilience and affective components affect life satisfaction), as indeed some previous studies (Armstrong et al., 2011; Liu et al., 2013) predicted. Specifically, the results revealed that the most plausible model, with both theoretical and empirical support, is one in which resilience partially mediates between PEI and the affective components of subjective well-being, and the affective components of well-being partially mediate the influence of resilience on life satisfaction. This finding corroborates theidea thatadolescentswhounderstandandmanagetheir emotions better and have a greater ability to overcome difficult situations also have a better perception of themselves in affective terms and report greater life satisfaction (Bajaj & Pande, 2016; Di et al., 2020; Noble & McGrath, 2012; Table 4.Direct, indirect, and total effects Hypothesis β d z d β i z i β t z t EA!RS .148 ** 3.759 ––.148 ** – EC!RS .320 ** 6.243 ––.320 ** – ER!RS .472 ** 9.753 ––.472 ** – EA !(RS) !NA .413 ** 8.891 .056 ** 3.298 .469 ** 9.565 EC !(RS)!NA .143 ** 2.990 .120 ** 4.658 .263 ** 4.420 ER !(RS) !NA ––.178 ** 6.091 .178 ** 6.091 EA !(RS) !PA ––.043 ** 3.108 .043 ** 3.108 EC !(RS) !PA .177 ** 4.119 .093 ** 4.152 .270 ** 5.209 ER !(RS) !PA .209 ** 4.220 .138 ** 4.713 .347 ** 5.082 EA !(RS, NA, PA) !LS ––.218 ** 6.720 .218 ** 6.720 EC !(RS, NA, PA) !LS ––.270 ** 6.648 .270 ** 6.648 ER !(RS, NA, PA) !LS ––.308 ** 7.719 .308 ** 7.719 RS!NA .376 ** 8.149 ––.376 ** 8.149 RS!PA .291 ** 5.726 ––.291 ** 5.726 RS !(PA, NA) !LS .285 ** 7.388 .224 ** 7.652 .509 ** 12.232 NA !LS .346 ** 9.378 ––.346 ** 9.378 PA !LS .323 ** 8.573 ––.323 ** 8.573 Note.*p< .05, **p< .01. EA = emotional attention, EC = emotional clarity, ER = emotional repair, RS = resilience, PA = positive affect, NA = negative affect, LS = life satisfaction. European Journal of Psychology Open (2025) Ó2025 The Author(s). Distributed as a Hogrefe OpenMind article under the under the license CC BY-NC-ND 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0) 8 L. Azpiazu et al., Facilitators of Adolescent Well-Being https://econtent.hogrefe.com/doi/pdf/10.1024/2673-8627/a000084 - Wednesday, October 22, 2025 1:28:30 AM - IP Address:95.18.210.110 Ramos-Díaz et al., 2019; Sánchez-Álvarez et al., 2016; Vergara et al., 2015;Yıldırım et al., 2022). According to these results, the specific components of emotional intelligence (PEI) influence resilience and subjective well-being in a differentiated manner. Specifically, theabilitytointerpretandunderstand emotions (emotional clarity) and the ability to regulate them (emotional repair) seemtoplayakeyrole in theability to copewithadversesituations, favoring adaptive responses to them and, consequently, promoting greater emotional well-being. On the other hand, emotional attention, while facilitating the perception of one’s own emotional states, could be related to greater negative affect when not accompanied by appropriate regulation strategies. This finding is consistent with previous literature (Vergara et al., 2015; Ramos-Díaz et al., 2019) and reinforces the idea that PEI contributes to well-being not only directly but also through its impact on resilience and affective components. Another relevant finding of this study is that, although Diener’s Subjective Well-Being Model (Diener, 1984; Diener, 2009; Pavot & Diener, 2013) is conceptualized as a three-part model comprising three independent yet related variables (positive affect, negative affect, and life satisfaction) (Rodríguez-Fernández & Goñi, 2011), when its components are analyzed in a multivariate manner together with other psychological variables, affects play a mediating and predictive role in life satisfaction, thereby supporting the affective mediation model posited by Zeidner et al. (2012). This finding (Eid & Larsen, 2008; Kong et al., 2019; Xiang et al., 2021) is consistent with the fact that affective experience is of greatimportance in judging life satisfaction, since positive emotions (and lack of negative ones) stimulate individuals to think more freely, creatively and thoughtfully, which in turn is conducive to finding positive meaning in one’s life (Fredrickson, 2001). Diener et al. (2017) argue that the inclusion of all the components of subjective well-being provides more accurate information than that offered by any one of them alone. Therefore, including and analyzing affective components together with other psychological variables offers greater insight and expands on that reported by previous studies, revealing that affective components may make an independent contribution to the association between PEI and life satisfaction among adolescents. However, the strength of this mediation may vary depending on several contextual factors. For instance, cultural influences can shape how individuals experience and regulateemotions(Maetal.,2018),whichinturnmayaffect the extent to which affective components mediate the impactofresilienceonlifesatisfaction.Consistentwithprevious research (Yang et al., 2022), this study reveals that affective components partially and significantly mediate the influence of resilience on life satisfaction, a finding that sheds some light on the debate as to whether they mediate fully or partially (Liu et al., 2013; Yang et al., 2022). It is also clear that both resilience (Liu et al., 2013) and affective experience (Eid & Larsen, 2008; Kong et al., 2019; Xiang et al., 2021) are of particular importance in the assessments people make of their lives: Those who cope better with difficulties and who feel pleasure frequently andrarely experience negative emotions are generally more satisfied with theirlives(Yangetal.,2022).Whilesomeresearchsuggests that socioeconomic status may play a role –by providing access to resources that facilitate positive affective experiences and emotional coping strategies (Rusu et al., 2018)– the evidence regarding its specific influence on affective mediators remains inconclusive (Steptoe et al., 2008). Further studies are needed to clarify this relationship. Further,familydynamics,includingparentalsupportand communication patterns, may significantly influence the resilience of individuals (Steptoe et al., 2008), further moderating this relationship. Affective experiences may not be uniform across different demographic groups: Adolescents at various developmental stages may process and express emotions in distinct ways (Abbruzzese et al., 2019), potentially altering the mediation effects. Similarly, gender differences in emotional expression and regulation could lead to variations in how affective components influence the relationship between resilience and life satisfaction (Abbruzzese et al., 2019; Zimmermann & Iwanski, 2014). A deeper exploration of these factors may provide a more nuanced understanding of the mechanisms underlying this mediation process. The present study also confirms that resilience partially mediates the association between PEI and the affective components of subjective well-being (Bajaj & Pande, 2016; Ramos-Díaz et al., 2019). Although this indirect influence is relatively weak, the data indicate that it is a significant mediation, thereby providing empirical evidence that helps clarify previous inconclusive results (Ramos-Díaz et al., 2019; Zhao et al., 2020) and shows that PEI promotes the use of emotional resources that lead to adaptive responses and good psychological outcomes (Salovey et al., 2000; Sánchez-Álvarez et al., 2016; Wang et al., 2017). In other words, emotional and resilient skills help people cope more successfully with stressors and in turn promote affective benefits that generate greater positive affect and less negative affect. Moreover,resilienceis particularly relevantinthe association between PEI and affective balance when the specific components of PEI are analyzed. Interestingly, emotional attention has a fairly strong influence on negative affect, compared to that exerted by emotional clarity and emotional repair on positive affect, whereas the latter two emotional skills have a more substantial impact on resilience. Consequently, rather than contributing directly to Ó2025 The Author(s). Distributed as a Hogrefe OpenMind article under the European Journal of Psychology Open (2025) under the license CC BY-NC-ND 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0) L. Azpiazu et al., Facilitators of Adolescent Well-Being 9 https://econtent.hogrefe.com/doi/pdf/10.1024/2673-8627/a000084 - Wednesday, October 22, 2025 1:28:30 AM - IP Address:95.18.210.110