Adolescents’ Emotions in Spanish Education: Development and Validation of the Social and Emotional Learning Scale

: The aim of this study was to develop and validate a scale for measuring the different areas of competence of the social and emotional learning model in students in compulsory secondary education. For this purpose, following the recommendations and quality standards established by the specialised literature, the Social and Emotional Learning Scale was constructed and administered to a convenience sample of 1385 students in compulsory secondary education. The results reveal adequate reliability of the instrument and fairly satisfactory fit indices for the structural model proposed. In addition, external evidence is provided for the validity of the instrument with the variables life satisfaction and school performance. The Social and Emotional Learning Scale is of satisfactory metric quality and provides a detailed and consolidated picture of adolescents’ level of self-awareness, social awareness, self-control, interpersonal skills and responsible decision making, which provides students with an instrument that can help to understand their interpersonal and intrapersonal skills and to establish mechanisms for their improvement.


Introduction
The need to urgently address the increasingly worrying incidence of social, emotional and behavioural problems among pupils in compulsory education, as well as to stimulate protective factors that increase their productivity and well-being, has led educational professionals in recent years to considerably increase their interest in social and emotional learning [1][2][3][4]. This component of students' personal and socioemotional development involves the ability to coordinate cognition, affect and behaviour [5], and is conceived as the process through which a person acquires and effectively uses the knowledge, skills and attitudes necessary to develop healthy identities, manage emotions, formulate and achieve positive personal and collective goals, feel and show empathy for others, initiate and maintain positive and supportive interpersonal relationships, manage interpersonal situations constructively and make responsible and caring decisions [6][7][8][9][10].

Literature Review
Social and emotional learning equips learners with the necessary tools to deal effectively and efficiently with the various tasks and challenges of their daily lives, leading to increased performance and satisfaction [2,3,10]. In fact, the accumulated evidence from basic and applied research on this construct has led to a rapid and wide diversification and incorporation in schools of programmes and practices aimed at promoting the following socioemotional competences among their students [1,5,6,11]: (a) self-awareness, i.e., the skills to accurately recognise one's own emotions, thoughts, values and how they influence behaviour, including the skills to assess one's own strengths and limitations with confidence, optimism and a growth mindset; (b) social awareness, seen as the skills to empathise with and understand the perspective of others, from diverse contexts and cultures, including the skills to defend one's own ideas without belittling others; (c) selfcontrol, or the ability to successfully regulate one's emotions, thoughts and behaviour in different situations, including the skills to set school goals and work towards their achievement, with self-discipline and self-motivation, using planning and organisational strategies; (d) interpersonal skills, conceived as the skills to establish and maintain healthy and nurturing relationships with diverse individuals and groups, including the ability to communicate clearly and effectively, listen actively and cooperate with others; and (e) responsible decision making, i.e., the skills to make constructive decisions about personal behaviour and social interactions based on ethical standards, safety and social norms, including the realistic assessment of the consequences of various actions and consideration of personal well-being and that of others, as well as the identification of problems, proposal of solutions and implementation of actions that contribute to the improvement of the immediate environment.
Social and emotional learning is a strong predictor of school achievement and is strongly related to psychological well-being and high levels of curiosity, motivation, optimism, creativity and prosocial behaviour among children and adolescents from diverse racial, ethnic and socioeconomic backgrounds, with and without emotional and behavioural problems, from different educational levels and environments [11]. This is reflected in the many systematic reviews that have been carried out to determine the impact of intervention measures based on this socioemotional component [8,9,[12][13][14][15][16][17][18][19][20][21].
For their part, numerous studies have established causal links between programmes and practices based on the social and emotional learning model and certain improvements in the socioemotional, behavioural and school level of their participants, even generating effects on the rest of the members of the educational community (e.g., in the case of teachers, higher rates of effectiveness and achievements in their teaching planning have been identified).
Evidently, for the evaluation of the effectiveness of such programmes and practices, various measurement instruments have been used for the different competence areas of the social and emotional learning model (i.e., self-awareness, social awareness, self-control, relationship skills and responsible decision making). Among these, the batteries, surveys and guides developed by CASEL [22], Panorama Education [23] or the Organisation for Economic Co-operation and Development [3] should be highlighted at an international level. However, the evaluative research on social and emotional learning carried out in the Spanish context has relied more on the use of other instruments [15], such as the Emotional Intelligence Inventory [24], the Socialization Battery 3 [25] or the Trait Emotional Intelligence Questionnaire [26], tests that unfortunately do not assess all the areas of competence of this model. This has stimulated the frequent use of the Social Emotional Competence Questionnaire [27] in recent years, despite the fact that its adaptation and validation have proven to be inconclusive in the Spanish population [28,29].
Therefore, in response to the growing interest of Spanish professionals and educational centres in the implementation of programmes, practices and actions based on the social and emotional learning model, especially as a preventive measure against the problems of educational exclusion that their students usually present [3,4], it seems essential to develop self-report measures to assess their effectiveness on socioemotional competences. Specifically, the purposes of this research were, firstly, to develop a scale to measure the different areas of competence of the social and emotional learning model (i.e., selfawareness, social awareness, self-control, relationship skills and responsible decision making) of students in compulsory secondary education; and secondly, to validate the theoretical framework and examine the psychometric properties of the scale in a sample of students in compulsory secondary education.

Participants
The sample consisted of 1385 students in compulsory secondary education, of whom 362 were in the first year (26.14%), 358 in the second year (25.84%), 339 in the third year (24.48%) and 326 in the fourth year (23.54%) in five secondary schools in the autonomous city of Ceuta (Spain). The mean age of the participants was 15.77 years (SD = 1.66), ranging from 12 to 18 years, while their gender distribution was 797 males (57.54%) and 588 females (42.46%).
The sample selection procedure was non-probabilistic, by convenience [30], while the sample size was determined according to the objective of the work, being above the minimum established in different simulation studies for structural models similar to those proposed in this work [31]. However, for greater certainty, the minimum sample size required for the structural complexity of the specified model was calculated (N = 200), considering the number of observed (30) and latent (5) variables, the expected effect size (0.10), the associated probability (0.05) and the desired levels of statistical power (0.80) [32].

Measures
The Social and Emotional Learning Scale is a Likert-type scale consisting of 30 items of estimation scored with 1 to 4 points (i.e., 1 = never; 2 = occasionally; 3 = often; 4 = always), grouped into 5 areas of socioemotional competence, such as self-awareness, social awareness, self-control, relationship skills and responsible decision making. This instrument, whose psychometric properties are presented in this research work, has been constructed with the purpose of measuring the different areas of competence of the social and emotional learning model in students in compulsory secondary education.
The Spanish adaptation of the Life Satisfaction Scale is a Likert-type scale composed of 5 items of estimation scored with 1 to 7 points (i.e., 1 = do not agree at all; 2 = disagree; 3 = slightly disagree; 4 = neither agree nor disagree; 5 = slightly agree; 6 = agree; 7 = strongly agree), which yields a total score for each participant considering the average of its items [33]. This scale was chosen because it is one of the few that measures subjective well-being or level of life satisfaction in adolescents and young adults, as well as for the adequacy of its psychometric properties, i.e., a Cronbach's alpha (α) of 0.85 and tests showing excellent goodness-of-fit indices that confirm its internal structure [33]. In the sample of this study, it yielded an α of 0.82 and a McDonald's omega (ω) of 0.83, while the Confirmatory Factor Analysis (CFA) showed adequate goodness-of-fit indices and statistics: chi-square (χ 2 ) (5; N = 1385) = 6.13; p < 0.29; χ 2 /degrees of freedom (df) = 1.24; Comparative Fit Index (CFI) = 0.99; Goodness of Fit Index (GFI) = 0.98; Standardised Root Mean Squared Residual (SRMR) = 0.03; Root Mean Squared Error of Approximation (RMSEA) = 0.02 (90% confidence interval (CI) = 0.00-0.08).
The Sociodemographic and School Data Questionnaire is an ad hoc self-report composed of five items with different response alternatives, aimed at collecting relevant sociodemographic (i.e., age and gender) and school (i.e., secondary school, grade and performance in the last grade) information from the participants.

Design and Procedure
In this work, as part of a research project aimed at evaluating the impact of a social and emotional learning programme among students in compulsory secondary education, an instrumental methodological design was adopted [34]. This research was approved by the Ethics Committee of the University of Granada (1736/CEIH/2020).
The construction and validation process of the Social and Emotional Learning Scale was developed following the recommendations and quality standards established by the specialised literature [35]. In this sense, in a first phase, as a result of a systematic review, the operational definition of the construct was established, identifying and defining its facets or domains.
In the second phase, once the requirements for the application of the instrument had been determined, the items for the different areas of competence or domains of social and emotional learning were developed. To this end, 23 items from the instruments specified below were initially adapted linguistically and culturally: (a) Social Emotional Competence Questionnaire [27]; (b) Brief Self-Control Scale [36]; (c) Brief Version of the Revised Social Problem Solving Inventory [37]; (d) Youth Teamwork Scale [38]; and (e) Civic Engagement Scale [39]. The procedure that we followed for the adaptation of these items was back-translation [40], which involved the following actions: (a) two professional translators independently generated their own Spanish version of the items; (b) the linguistic and cultural equivalence of these translations was reviewed and analysed by three specialists, who, after making some adjustments, proposed a consensus translation, on which a third professional translator performed a back translation into English; and (c) the degree of comprehensibility of the items in Spanish was assessed by a committee of three specialists. Subsequently, 7 new items were constructed to reflect the specific characteristics of the competence areas of the social and emotional learning model for students in compulsory secondary education.
In the third phase, four international experts examined the content validity of the Social and Emotional Learning Scale by assessing the representativeness, relevance, diversity, clarity, simplicity and comprehensibility of the items, as well as the need to eliminate or include items for any of the competence areas [35]. The Social and Emotional Learning Scale was modified several times until it was in line with the experts' comments, finally confirming its approval. A pilot study was also carried out with a small sample of students in compulsory secondary education (N = 46). This pilot test allowed us to confirm the direction and meaning of the psychometric properties of the Social and Emotional Learning Scale, the approximate time needed to complete it and the appropriate content and format of its items [35].
Finally, in the fourth phase, once the relevant institutional permissions had been requested and granted, the management teams of the city's secondary schools were contacted in order to request their voluntary participation in the research. For this, the purpose and procedure to be carried out in the study were explained to them, and then, once they had accepted the proposal to participate, they signed the agreements and informed consent forms. The families or guardians of the compulsory secondary education pupils were also informed and signed the appropriate informed consent forms. The instruments were administered collectively, during class time, in each of the participants' reference class groups using the computer-assisted web interview technique [41]. This process was carried out by a member of the research team, who first informed the compulsory secondary school students of the purpose of the study, the voluntary nature of their participation and the confidentiality of their responses. The written instructions were then supplemented with oral explanations by the researcher and the questions raised by the participants were answered. They completed the Social and Emotional Learning Scale in 20-25 min. Data were collected in February 2021. The psychometric properties of the Social and Emotional Learning Scale were assessed, and then we used the scale to study the different areas of socioemotional competence in the participants.

Data Analysis
First, descriptive analysis of the collected data was performed, and the data were checked for distribution, linearity, outliers, missing and influential values [42]. Second, in order to examine the theoretical structure of the Social and Emotional Learning Scale, a CFA of five first-order correlated factors was carried out. The CFA was estimated by the Weighted Least Squares (WLS) method, using different indices to assess its goodness of fit [43,44]: (a) χ 2 , which should not be statistically significant, although due to its sensitivity to sample size, the χ 2 /df ratio can also be considered, which should be less than 5 points; (b) CFI and GFI, which should yield values above 0.90 points; and (c) SRMR and RMSEA, which should yield values below 0.08 points.
Third, to assess the reliability of the Social and Emotional Learning Scale, internal consistency and composite reliability were calculated. Specifically, to assess internal consistency, we calculated the α and ω, which should yield coefficients greater than 0.70, while to assess composite reliability, we used the Composite Reliability Index (CRI) and the Average Variance Extracted (AVE), which should have values greater than 0.70 and 0.50, respectively [45].
Fourthly, in order to obtain external evidence of validity of the Social and Emotional Learning Scale, the different competence areas of the social and emotional learning model (i.e., self-awareness, social awareness, self-control, interpersonal skills and responsible decision making) were correlated (Pearson's r) with others theoretically linked to this factor in the literature, such as life satisfaction and school performance. Likewise, after checking the absence of normal distribution in the scores by means of the Kolmogorov-Smirnov test, the Mann-Whitney U-test for 2 independent samples and the Kruskal-Wallis H-test for independent samples were performed in order to identify differences in the areas of competence of the model according to the variables sex, age and school performance. However, given the impossibility of performing multivariate contrasts, the error rate per family, resulting from the multiple comparisons problem, was controlled with the Bonferroni correction.

Results
Descriptive statistics for the different items of the Social and Emotional Learning Scale are shown in Table 1. The skewness and kurtosis values, as well as the Kolmogorov-Smirnov statistic, reveal a non-normal univariate distribution for all items, while the multivariate Mardia skewness (47.47, χ 2 = 10983.17; p < 0.001) and kurtosis (1083.88; χ 2 = 2767.60; p < 0.001) coefficients indicate that the joint distribution of the items is not normal; (χ 2 = 10983.17; p < 0.001) and kurtosis (1083.88; χ 2 = 2767.60; p < 0.001) indicate that the joint distribution of the items does not conform to the multivariate normal distribution, justifying the choice of the WLS estimator for the CFA. No missing values were detected, and outliers identified through the Mahalanobis distance were processed using robust estimators. Descriptive analysis of the items reveals mean scores ranging from 2.36 (SD = 1.01) for item 28 to 3.49 (SD = 0.75) for item 4. In turn, the results of the CFA model of five firstorder correlated factors yield quite satisfactory goodness-of-fit indices and statistics: χ 2 (395; N = 1385) = 1087.29; p > 0.05; χ 2 /df = 2.75; CFI = 0.97; GFI = 0.98; SRMR = 0.05; RMSEA = 0.04 (90% CI = 0.03-0.04). All items showed factor loadings above 0.4 ( Table 1). The correlations between the different competence areas of the Social and Emotional Learning Scale ranged from 0.38 (self-awareness and social awareness) to 0.70 (self-awareness and self-control).  Regarding the internal consistency of the Social and Emotional Learning Scale, the results reveal an α and ω of 0.90 for the total scale, with scores ranging from 0.70 in the area of self-awareness competence to 0.84 in the area of self-control competence, although it is necessary to mention that the elimination of any item would generate a reduction in these coefficients.
On the other hand, the results related to the external evidence of validity of the instrument reveal a consistent pattern of positive relationships between the different areas of competence of the Social and Emotional Learning Scale, including between them and the rest of the variables included in the study (i.e., level of life satisfaction and school performance). The LRC and AVE are also shown ( Table 2). Finally, the significance level for each of the multiple comparison tests was adjusted with the Bonferroni correction, being 0.01 for the different contrasts (0.05/5). In this line, the results of the comparisons according to the gender variable reveal statistically significant differences in favour of females in the area of social awareness competence (U = 204,839.00, p < 0.01), as well as in favour of males in the area of self-control competence (U = 213,923.50, p < 0.01), while according to the age variable, no statistically significant differences were observed. Similarly, statistically significant differences were found in the different competency areas of the Social and Emotional Learning Scale as a function of school performance, with participants with higher school performance scoring higher in the competency areas of the model (Table 3).

Discussion and Conclusions
The purpose of this research was to develop and validate a scale aimed at measuring the different areas of competence of the social and emotional learning model (i.e., selfawareness, social awareness, self-control, relationship skills and responsible decision making) in students in compulsory secondary education. Therefore, based on the results obtained, the main conclusion of this study is that the Social and Emotional Learning Scale has an adequate metric quality, with satisfactory levels of validity and reliability.
Indeed, the factor structure of the Social and Emotional Learning Scale model, with a high level of similarity to that of the Social Emotional Competence Questionnaire [27], yielding an excellent fit of the proposed structural model, provides evidence in favour of the instrument and the underlying theoretical construct measure [43], while reliability shows adequate levels of internal consistency and composite reliability of the instrument as a whole and in each of the different competence areas of the social and emotional learning model [45]. Therefore, these results support its use to assess socioemotional competencies in adolescents, i.e., how they are aware of themselves and others, as well as to measure how they respond to the contexts of family, school and community, in personal, social and ethical terms [1,[5][6][7][8][9][10].
In terms of external evidence of the validity of the Social and Emotional Learning Scale, the different competency areas of the social and emotional learning model were shown to be positively linked to the level of life satisfaction and school achievement, as reported in the literature [2,3]. These results clearly show that social and emotional learning stimulates students to develop the tools they need to successfully deal with different situations in their school and personal lives [2,10]. Hence, over the past few years, a huge number of programmes and practices based on this model have proliferated as intervention measures to address the problems of educational exclusion that students in compulsory education often present [3,4].
Nevertheless, in conclusion, it can be stated that the empirical evidence provided in this work supports the use of the Social and Emotional Learning Scale to obtain a detailed and consolidated picture of adolescents' level of self-awareness, social awareness, selfcontrol, interpersonal skills and responsible decision making. Likewise, this research represents an advance in evaluative research on this construct in the Spanish context [15], especially considering the absence of adequate instruments to evaluate the effectiveness of social and emotional learning programmes and practices. This scale is a tool that, in addition to measuring the different areas of competence of the social and emotional learning model, can help to predict certain improvements at the socioemotional, behavioural and school levels in students in compulsory secondary education and, therefore, increase performance and success in different domains or environments (e.g., school, work, personal, etc.).

Limitations and Future Studies
When using the Social and Emotional Learning Scale, and also with a view of future research, it is necessary to take into account certain aspects. Firstly, in relation to the linguistic and cultural adaptation of the items, the results of other research carried out in the Spanish context could not be considered, as recommended by López-Walle et al. [46]. Secondly, with regard to the characteristics of the participants, their homogeneity may have affected the results obtained with respect to the psychometric properties of the instrument, which may generate limitations when used with other populations. In this sense, in future research, it would be appropriate to extend the analyses of the psychometric properties of the Social and Emotional Learning Scale (e.g., temporal stability or test/retest reliability, predictive validity on other psychological and educational variables), as well as to replicate and refine the results obtained in this study with larger and more representative samples from a probability sampling. Data Availability Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest.