The mediating role of self-efficacy in the relationship between resilience and academic performance in adolescence

A B


Introduction
The investigation of psychological variables in academic environments can help not only to better understand students' cognitive processes during their school years but also to stimulate certain behavioural traits that greatly contribute to shape their adult personality (García & Meira, 2019).
While most students go through this stage without major issues in their personal and academic development, others can undergo sensations and perceptions that undermine their motivation and commitment towards school tasks, leading to school dropout. This is the result of personal and contextual variables and the lack of skills and/or strategies to cope with academic demands (Barreno, Haro & Flores, 2019). involved in academic performance, and to develop strategies to improve the students' personal development and academic satisfaction, ultimately contributing to reduce early school dropout (Vizoso & Arias, 2018).
In this context, and given the absence of studies that directly relate the variables under consideration, the main aim of this study is to analyse the relationship between self-efficacy, resilience and academic performance in a sample of adolescent secondary school students.
The study's two main hypotheses are: (a) Self-efficacy is related to resilience and academic performance, stimulating adaptive behaviours (b) Self-efficacy plays a mediating role in the relationship between resilience and academic performance.

Sample
The sample comprises 2652 students with ages ranging from 12 to 19 (M=14.55; SD=1.70), both male (N = 1368; 51.58%) and female (N = 1284; 48.41%) from 14 secondary schools in Zaragoza, Aragón, Spain. Inclusion criteria were the ability to read and communicate in perfect Spanish (a necessary condition to understand the questionnaires). Incomplete questionnaires (34) and students with cognitive disorders that hampered the full understanding of the questionnaires were excluded. The schools were chosen by random sampling and all students in the chosen schools were furnished with a questionnaire; 98.72% of questionnaires were returned and counted. The study was undertaken between January and February 2020.

Instruments
The following questionnaires were used. In order to measure self-efficacy, the Academic Self-efficacy Scale (ASES), validated for adolescents by Garcíaet al., (2010), was used. The scale comprises 10 items to measure self-efficacy in an academic setting (e.g. "I am convinced that I can carry out outstanding exams"). The responses measure the degree of agreement and disagreement of the subject in a 5-point Likert scale ranging from "Strongly disagree" (1) to "Strongly agree" (5). Previous studies have shown the reliability of the questionnaire in academic settings, yielding an overall Cronbach-α value of.91, and of.89 in our study.
Concerning resilience, the Brief Resilience Scale (BRS), translated and adapted to Spanish adolescents by Rodríguez, Alonso, and Hernansaiz (2016), was used. This mono-factorial scale comprises 6 items (e.g. "I tend to bounce back quickly after hard times"). Answers are expressed in a 5-point Likert scale ranging from "Strongly disagree" (1) to "Strongly agree" (5). The original instrument yields a Cronbach-α value of. 88, and of.90 in our study.
Finally, academic performance was measured on the basis of the average marks in the first trimester, ranging from 0 (minimum) to 10 (maximum). This variable is widely used, and is regarded as an effective predictor of the students' academic performance (Barca et al. 2011;Córdoba et al. 2012;Risso, Peralbo, & Barca, 2010). For our study, this variable yielded a Cronbach-α of.85.

Protocol
The questionnaires were handed out to the students in the classrooms, all students in each school receiving the questionnaire in the same day, set out in advance in coordination with the school's management. The schools and parents/tutors signed informed consent forms. All students and parents/tutors were informed about the nature of the study, and participation was voluntary, in line with the ethical directives set out in the Declaration of Helsinki (AMM, 2000). The protocol was endorsed by the Ethics Committee of the Psychology and Sociology Department, University of Zaragoza. Questionnaires were anonymous and confidential, and students could opt out at any point in the process.

Data analysis
Descriptive statistics were undertaken to establish the socio-demographic profile of the sample, including such variables as gender, age, course, type of school and course repeats, as well as the variables analysed in the study. Following this, correlations between selfefficacy, resilience and academic performance were investigated using statistical software IBM SPSS v26.0. In order to establish the predictive value over academic performance, a stepwise multiple regression model was adopted. Finally, SPSS v26.0 ′ s MACRO tool was used to carry out mediation analyses by bootstrapping (10,000 runs). For all the operations, a p ≤ 0.05 alpha level was adopted, with a 95% confidence level.

Descriptive variables
As illustrated in Table 2, scores for self-efficacy, resilience and academic performance were relatively even. Males scored slightly higher in terms of self-efficacy, while females scored slightly higher in terms of resilience and academic performance. Table 3 illustrates correlations between the variables under consideration. They are all significant correlations, but in different ways.

Correlational analysis between self-efficacy, resilience and academic performance
Self-efficacy is positively and significantly correlated with resilience (r = .390) and with academic performance (r = .353), while resilience showed a much weaker correlation with academic performance (r = .125).

Mediation effects of self-efficacy in the relationship between resilience and academic performance
In order to establish whether the relationship between resilience and academic performance is mediated by self-efficacy, Hayes's (2018) MACRO tool in Process 3.0 de SPSS (v 26.0) was used, following the methodology put forth by Tal-Or, Cohen, Tsarfati, and Gunther (2010) in a OLS regression. The three variables (self-efficacy, resilience and academic performance) was normally distributed.
The mediation model has been tested through the Sobel test, yielding data with a coefficient of 6.61, p < 0.001.
As shown in Fig. 1, self-efficacy was found to play a mediating role between resilience and academic performance. The results indicate a mediating effect of resilience (VI) on self-efficacy of.31, and a mediating effect on academic performance (VD) of.35; in both cases p > .001. Zero was not included in the bootstrap interval, B= .11, SE= .02, 95% [CI.06,.15], so it can be argued that self-efficacy mediates in the relationship between resilience and academic performance.
These results suggest that, in and by itself, resilience has no direct significant effect on academic performance (.07, p < .10), but its combination with self-efficacy yields a result of.18, p < .001 (direct effect + indirect effect), the proportion of variance being explained by model R 2 = .31***. This suggests that self-efficacy plays a mediating role in the relationship between resilience and academic performance. This has important practical implications.

Discussion
The aim of this study was to analyse the relationship between self-efficacy, resilience and academic performance in adolescent secondary school students. The first hypothesis, that self-efficacy, is related to resilience and academic performance, was accepted; the results show that selfefficacy is positively correlated with resilience and academic performance.
This conclusion agrees with the existing literature. While few studies have explored the direct relationship between self-efficacy and resilience in secondary schools, some studies take both these variables into consideration, among others. Morales and González (2014) argued that intellectual capacity is positively correlated with resilience and self-esteem, promoting self-determined behaviour among secondary school students; Gaxiola, González, & Contreras (2012) argue that self-efficacy and resilience are positively correlated to goal orientation among secondary school students; Fínez & Morán (2017) draw a link between intrapersonal self-evaluation factors, such as self-concept, self-esteem, self-efficacy and resilience in adolescent students.
Other studies emphasise the close relationship between self-esteem and resilience in university students, also taking into consideration other variables, such as social skills, coping strategies and anxiety and stress control (Alfian, Adam, & Ibrahim, 2017;Blanco, Aguirre, Barrón, & Blanco, 2016;. On the other hand, the relationship between self-efficacy and academic performance has been paid greater scholarly attention. Some studies establish a direct relationship between these variables; Avalos, Oropeza, Ramírez, and Palos (2018) and Galleguillos & Olmedo (2017) find statistically significant correlations between self-efficacy and academic performance in secondary school students; Castro (2020) draws a similar link in both primary and secondary school contexts. Other studies examine the relationship between self-efficacy and academic performance among other significant variables, such as study habits, the development of creativity, commitment and personal satisfaction, and motivation towards school tasks (Honicke & Broadbent, 2016;Martínez & Medina, 2019;Stajkovic, Bandura, Locke, Lee, & Sergent, 2018). Our second hypothesis, that self-efficacy plays a mediating role in the relationship between resilience and academic performance, was also fully accepted, as self-efficacy was found to have an effect on the other two variables.
On the one hand, mediation analysis suggests that resilience is a poor predictor of academic performance; that is, the direct effect of the former over the latter is not statistically significant. However, the mediating role of self-efficacy in the relationship between resilience and academic performance was found to be significant. This confirms the importance of self-efficacy for adolescent students, which has important practical implications.
To our knowledge, no previous study has directly addressed the mediating role of self-efficacy in that relationship, as all existing studies have limited their analysis to the bidirectional correlations between the constructs.
However, many previous studies have examined the mediating role of self-efficacy in adolescent students; Avalos et al. (2018) establish the mediating role of self-efficacy in the relationship between academic skills and achievements ;Cerezo et. al., (2019) establish a direct correlation between training in self-regulated learning strategies and an increased understanding of said strategies, a relationship in which self-efficacy plays a mediating role; Soler, Fernández, and Meseguer (2017) establish the effect of self-efficacy on the relationship between personal roles and emotional wellbeing; Udavar, Fiori, and Bausseron (2020) underscore the modulating role  Table 3 Correlational analysis between self-efficacy, resilience and academic performance. of self-efficacy in the relationship between emotional intelligence and academic performance in secondary school students. All of this emphasises the important role played by the constructs under consideration which, alongside personal and contextual circumstances, have a direct effect on school performance. It is, therefore, imperative to address these issues in order to create the conditions for the adequate personal and academic development of the students (Bisquerra, Pérez, & García, 2015).

Limitations of the study
The limitations of this study are chiefly related to its lateral nature being a cross-sectional data-collection, which was a one-off event, and as a result the data have not temporal depth, while scores can easily change significantly from year to year, and even within the same school year. Hence, we must to be cautious when interpreting the established mediation model. In a similar fashion, the schools were selected randomly, and are not a reliable section of the city in which the study was undertaken, in terms of types of school, students and teachers, socio-economic conditions and social/cultural settings.

Future prospects
Future studies should examine the relationship between academic self-efficacy and other psychological variables, besides resilience and academic performance. It is also necessary to undertake longitudinal studies that allow us to examine the evolution of these constructs over a longer time spam. In addition, it would be interesting to take into consideration other academic tiers, such as primary school (6-11 years) and university (18 years and over). It would also be of interest to take into consideration other socio-demographic variables, such as gender, age, type of school.

Practical implications
Our results have practical implications for educational strategies, namely the promotion of teaching strategies that encourage selfdetermined and motivated behaviours from an early age, such as the development of self-efficacy, dedication and motivation towards academic tasks, contributing to students feeling more self-confident in coping with their daily work. In this way, students will be immersed in adaptive contexts, which will increase their chances of academic and, possibly, personal success. Similar implications apply in the emotional sphere, especially during adolescence, in which great physical and behavioural changes ensue, and in which a good training in emotional regulation can contribute to increase self-efficacy and resilience and other psychological variables.
Related to this, programmes directed by psychology and educational professionals and government policies can also help to improve students' overall experience, decreasing the risk of early school dropout. At the same time, parents must go hand in hand with their children's education at school, being the engine for promoting adaptive behaviours that enhance those reflected at school to ensure an optimal climate for adequate personal and school development.
Finally, it is worth emphasising that our results are but one step which encourages us to continue our research in the field of educational psychology and, specifically, in the characterisation of the different variables that affect the personal and academic development of adolescent students.