**2. Method**

#### **2.1. Participants**

Participants were chosen from among secondary school students attending schools in the Autonomous Region of the Basque Country (ARBC). The sample group comprised 945 adolescent students (425 boys and 520 girls; Mage = 14.50, SD = 1.82; range 12–17) from a mid-level socio-cultural context. The students were distributed throughout the different school years as follows: Year 1 of Compulsory Secondary Education (CSE) (25.2%); Year 2 of CSE (18.7%); Year 3 of CSE (18.7%); Year 4 of CSE (16.1%); and the 2-year Spanish Baccalaureate (21.9%).

#### **2.2. Variables and measurement instruments**

*Resilience* was evaluated using the *CD-RISC 10 Resilience Scale* [67]. The 10 items of this abbreviated version of the Connor-Davidson Resilience Scale [12] are scored on a 5-point Likerttype scale (higher scores reflecting greater resilience until 40). The reliability and validity of the CD-RISC 10 to be adequate in one large sample of adolescents were found [69]. In this study, the internal consistency coefficient obtained was *α* = 0.75.

The Spanish version of the *Satisfaction With Life Scale* (SWLS) by Diener, Emmons, Larsen and Griffin [28, 68] was used to evaluate *satisfaction with life.* This scale measures global cognitive judgments of satisfaction with one's life on a 7-point Likert-type scale. The internal consistency coefficient obtained for the sample used in the present study was *α* = 0.82. The minimum score is set at 5, while the maximum score is 35 points. The authors have also established the following rating ranges for a better interpretation of their results: from 31 to 25 = extremely satisfied; from 26 to 30 = satisfied; from 21 to 25 = slightly satisfied; 20 = neither satisfied nor dissatisfied; from 15 to 19 = slightly dissatisfied; from 10 to 14 = dissatisfied; from 4 to 9 = extremely dissatisfied. This questionnaire has been implemented successfully in various studies with populations of adolescents [66, 69].

Affect balance was measured using Bradburn's *Affect Balance Scale* [29]. The scale comprises 18 items to which responses are given on a 4-point Likert-type scale. The scale has shown adequate reliability and validity in a population of adolescents [24]. The Cronbach's alpha reliability coefficients obtained with our sample were positive affect (0.78) and negative affect (0.78). The score obtained by a subject is within a theoretical range between 9 and 36 points for each positive or negative scale.

*School engagement* was evaluated using the *School Engagement Measure* (SEM) by Fredericks, Blumenfeld, Friedel and Paris [70, 71]. The measure consists of 19 items to which participants respond on a 5-point Likert-type scale. Based on these 19 items, the authors obtain 3 factors which measure behavioral (with 4 items), emotional (5 items) and cognitive engagement (7 items), which are the 3 measures used for this study. The reliability assessment resulted in adequate internal consistency indexes for all three factors. With the sample group used in this study, the internal consistency for the scale was *α* = 0.72 for behavioral engagement, *α* = 0.70 for emotional engagement and *α* = 0.78 for cognitive engagement. The possible score in the behavioral dimension ranges from 4 to 20, in the emotional dimension from 5 to 25 and in the cognitive one from 7 to 35 [71].

*Perceived academic performance* was evaluated using the *Brief School Adjustment Scale* (EBAE-10) by Moral de la Rubia et al. [72]. This multidimensional questionnaire comprises 10 items with 6 response options, grouped into 3 indicators of school adjustment: problems with school integration, academic performance and academic expectations. For the purposes of this study, only the *academic performance* subscale was used, referring to participants' perceptions of their own performance as students. The subscale comprises three items, including "I get good grades" and "I think I'm a good student." The internal consistency of the subscale was *α* = 0.77. The total score ranges from 3 to 18 and a higher score indicates a higher level of academic performance.

#### **2.3. Procedure**

**2. Method**

**2.1. Participants**

166 Health and Academic Achievement

**2.2. Variables and measurement instruments**

**Figure 1.** Proposed theoretical structural model.

studies with populations of adolescents [66, 69].

for each positive or negative scale.

study, the internal consistency coefficient obtained was *α* = 0.75.

Participants were chosen from among secondary school students attending schools in the Autonomous Region of the Basque Country (ARBC). The sample group comprised 945 adolescent students (425 boys and 520 girls; Mage = 14.50, SD = 1.82; range 12–17) from a mid-level socio-cultural context. The students were distributed throughout the different school years as follows: Year 1 of Compulsory Secondary Education (CSE) (25.2%); Year 2 of CSE (18.7%); Year 3 of CSE (18.7%); Year 4 of CSE (16.1%); and the 2-year Spanish Baccalaureate (21.9%).

*Resilience* was evaluated using the *CD-RISC 10 Resilience Scale* [67]. The 10 items of this abbreviated version of the Connor-Davidson Resilience Scale [12] are scored on a 5-point Likerttype scale (higher scores reflecting greater resilience until 40). The reliability and validity of the CD-RISC 10 to be adequate in one large sample of adolescents were found [69]. In this

The Spanish version of the *Satisfaction With Life Scale* (SWLS) by Diener, Emmons, Larsen and Griffin [28, 68] was used to evaluate *satisfaction with life.* This scale measures global cognitive judgments of satisfaction with one's life on a 7-point Likert-type scale. The internal consistency coefficient obtained for the sample used in the present study was *α* = 0.82. The minimum score is set at 5, while the maximum score is 35 points. The authors have also established the following rating ranges for a better interpretation of their results: from 31 to 25 = extremely satisfied; from 26 to 30 = satisfied; from 21 to 25 = slightly satisfied; 20 = neither satisfied nor dissatisfied; from 15 to 19 = slightly dissatisfied; from 10 to 14 = dissatisfied; from 4 to 9 = extremely dissatisfied. This questionnaire has been implemented successfully in various

Affect balance was measured using Bradburn's *Affect Balance Scale* [29]. The scale comprises 18 items to which responses are given on a 4-point Likert-type scale. The scale has shown adequate reliability and validity in a population of adolescents [24]. The Cronbach's alpha reliability coefficients obtained with our sample were positive affect (0.78) and negative affect (0.78). The score obtained by a subject is within a theoretical range between 9 and 36 points A number of schools were randomly chosen from a list of all schools in the Autonomous Region of the Basque Country (ARBC), and different year groups within each school were selected in accordance with the interests of the study. The battery of questionnaires was administered to participants class by class during school hours. Throughout the process, care was taken to ensure that all participation was strictly voluntary, and the anonymity of the responses given was protected in order to reduce the social desirability bias. The single blind criterion was used, with students being unaware of the purpose of the study. The study complied with the ethical values established for psychological research and assessment and respected the basic principles laid out in the American Psychology Association's (APA's) ethics code and in current regulations (informed consent and the right to information, protection of personal data and confidentiality guarantees, non-discrimination, non-remuneration and the right to withdraw from the study at any time).

#### **2.4. Data analysis**

Missing values (2.1%) were inferred using the expectation maximization (EM) algorithm and the Markov chain Monte Carlo (MCMC), offered by the LISREL 8.8 program. Extreme values (1.3%) were eliminated using the SAS program. To ensure normality, the bootstrap method was applied, as offered by the AMOS 24 program.

A Pearson correlation analysis was conducted between the study variables with the aim of determining any possible connections between them and to verify the non-existence of multicollinearity. A descriptive analysis was also carried out of the means and standard deviations of all the study variables. Both the descriptive statistics and the correlation coefficients were calculated using the SPSS 22 program. To test the structural regression model we used the structural equation modeling (SEM) technique, provided by the AMOS 24 program. In the first step, the measurement model is analyzed to check that each of the latent variables is represented by its indicators. In the second step, the analyses for testing the structural model are carried out using the maximum likelihood (ML) procedure.
