Introduction

In 2022, 12.3 million people, which accounts for 26% of the Spanish population, were at risk of poverty and/or social exclusion, according to the European Anti-Poverty Network (EAPN, 2023). Social exclusion is a structural and multidimensional process that affects various spheres, including the economic, labor, educational, socio-health, residential, relational, and participatory (Contreras-Montero, 2020; Hernández, 2010). One of the most significant factors related to social exclusion is the lack of access to formal education. Individuals with limited formal education have reduced access to employment and participation, increased health problems, and are more likely to fall into poverty (Hernández, 2010; Moreno, 2015).

Dropping out of the educational system is a significant issue in Spain. Early school leaving (ESL) is defined as the percentage of students between 18 and 24 years of age who have not completed and are not currently studying any type of post-compulsory education (Eurostat, 2022). In 2022, Spain had a percentage of 13.9%, which is one of the highest rates among European countries (X̅ = 9.6%) (Eurostat, 2022). Students can also drop out without finishing compulsory studies. In the scientific literature, school dropout is attributed to a process of progressive disengagement of students from the educational system (Mena et al., 2010).

Students who drop out generally refer to discourses of deficit, to a perceived lack of ability to complete their studies, and place a low value to formal education (Romero & Hernández, 2019). These students often encounter multiple risk factors that increase the likelihood of dropping out of formal schooling. Most of these factors are associated with social and economic inequalities, such as low levels of education and poverty in the family (Bayón-Calvo et al., 2021; EAPN, 2023; Huisman & Smits, 2015). Conflicting relationships with teachers and/or family members (Romero & Hernández, 2019) are also related.

In addition, gender is considered a significant risk factor in the literature. Male students tend to drop out of school more frequently than their female counterparts (Eurostat, 2022). This trend may be attributed to gender roles, as men often face greater pressure to enter the labor market (Bayón-Calvo et al., 2021; Ruiz et al., 2013). Additionally, male gender socialization can make it challenging for adolescents to integrate into school, as it is sometimes incompatible with maintaining status within their group of friends (Reay, 2018). Males exhibit more behavioral problems than females, which can impede their integration into school and their relationship with teachers (Fortin et al., 2010).

In institutional and organizational contexts, factors such as school segregation both between and within schools (Save the Children, 2019) and the higher rate of teacher turnover in areas at risk of social exclusion (Allen et al., 2018; Llorent-Bedmar et al., 2021) hinder students’ school attainment. The presence of more risk factors increases the likelihood of students dropping out of school (Rouse et al., 2020). Therefore, areas at risk of social exclusion are the most affected. We can consider that social class and inequalities are at the root of students’ school attainment and success (Save the Children, 2019).

However, even in areas of higher risk, some students complete their compulsory studies and continue with further formal schooling. The concept of resilience helps define these situations for study. Resilience refers to the process of successfully coping or recovering in the face of adverse situations (Gartland et al., 2019; Toland & Carrigan, 2011). It comprises two elements: significant exposure to severe risk and evidence of positive adaptation despite such risk (Masten, 2007). In the educational context, positive adaptation refers to a student’s ability to perform adequately in school, achieving similar or better results than the normative population that does not face risk (Masten et al., 2006; Lessard et al., 2014; Longás et al., 2019).

It is essential to emphasize that resilience should be understood as an interactive and dynamic process that involves individuals, families, schools, and communities (Gartland et al., 2019; Toland & Carrigan, 2011). It is not an ability or a characteristic of individuals, and does not make people invulnerable, even if achieved in a specific context under certain conditions. If the conditions are altered, individuals who exhibit resilience may no longer demonstrate it. In this sense, the concept of resilience constitutes a definition of situations of positive adaptation in the face of adversity, and not an explanation of such adaptation (Matías-García, 2022).

Precisely, for resilience to occur, risk factors must be controlled by protective factors, defined as such because they allow or facilitate positive adaptation in individuals (Gartland et al., 2019). Protective factors against ESL in situations of social exclusion may include greater public spending on scholarships and grants (Alegre & Benito, 2010), opportunities for family participation in school (Borman & Overman, 2004), and the formation of supportive relationships with family, teachers, and/or other students (Lessard et al., 2014; Longás et al., 2019), among others (Matías-García, 2022; Matias-Garcia et al., 2024). Protective factors enable students to establish positive academic expectations for the future, engage more with school, and develop an internal locus of control (Lessard et al., 2014; Longás et al., 2019), which helps them better adapt to the school environment.

Interwoven between the processes of academic risk and resilience, the development of students’ learner identity plays a vital role. According to Lawson (2014), learner identity is defined as “…how an individual feels about himself/herself as a learner and the extent to which he/she describes himself/herself as a ‘learner’” (p. 344). These identity constructs are known to mediate achievement, particularly in students with low socioeconomic status (Li et al., 2020; Hansen & Henderson, 2019).

From a cultural psychology theoretical background, our approach to identity involves the study of self or self-construal, which refers to an individual’s sense of self in relation to others (self-description or self-view) (Sedikides & Brewer, 2001). From our perspective, cognition is situated in activity settings (Werstch, 1991). Thus, we understand the self as a dynamic, distributed, and dialogical entity, and not as stable and homogeneous. For instance, Bruner (1996) proposed the existence of a distributed self, which he described as a “swarm” of participations that arise from the situations in which an individual participates. While constructing their identity as an individual distinct from others in each sociocultural setting, the person maintains a close relationship with them.

In this sense, the literature shows that students internalize the positive or negative academic discourses from their interpersonal relationships with teachers, peers, and family members, impacting their identity as learners (Matías-García, 2022; Matias-Garcia et al., 2024; McFarland et al., 2016). In the educational contexts, the influence of academic activities and discourses has a particularly high effect on its development. Characteristics such as the ability to learn or to be attentive during classes are constantly evaluated throughout formal schooling, explicitly and individually. These tasks and discourses shape students’ identity construction towards teachers’ demands, in a positive or negative way.

Also, there is extensive evidence from cross-cultural research about the relationship between formal schooling and changes in cognitive processes (Cole, 1996; Rogoff, 1990; Scribner & Cole, 1981), such as the development of abstract, decontextualized verbal thinking. Due to the type of discourse employed in formal instructional contexts, the literature relates formal schooling to the development of more individual, autonomous, reflective, and agentive selves (De la Mata et al., 2015; 2019). Authors such as Kagitcibasi (2007), Greenfield (2009), and Keller (2007) claim that formal schooling is an important factor that fosters a cultural model of the self that is characterized by autonomy and agency.

Through educational institutions and interactions with others, between risk and protective factors, students develop a unique understanding of themselves, others, and their environment. This understanding allows them to make the identitarian decision to continue with their studies. Thus, the aim of the present work is to describe and analyze the development of the learner identity of students who present a trajectory of resilience in contexts at risk of social exclusion. This will be done so through variables such as the degree of general risk to which they are exposed to, the educational level attained (academic year), or their gender.

In line with Agenda 2030's Sustainable Development Goals current concerns on No Poverty, Education, Inclusion, and Peace and Justice (SDG 1, 4, 11, and 16; United Nations, n.d.), this study will help us gain the necessary knowledge for the development of psychoeducational interventions that favor educational continuity in students facing social exclusion. Therefore, the present research presents the following exploratory objectives:

  1. 1.

    To analyze the characteristics of the learner identity of students with resilience trajectories in contexts at risk of social exclusion.

  2. 2.

    To compare the learner identity of students from two high schools that present different degrees of risk for dropping out.

  3. 3.

    To study the role of academic experience (in terms of participants’ academic year) in the development of learner identity.

  4. 4.

    To investigate the impact of gender on the development of learner identity.

Methods

Participants

An a priori power analysis to estimate the sample size required for Mann-Whitney U tests, based on data from Santamaría et al. (2010), was conducted. This study employed the same instrument and similar analytical categories as the one used in the present work and reported moderate to large effect sizes, so a moderate effect size according to Cohen’s (1988) criteria was established for the estimation (d = 0.5). With an alpha of 0.05 (two-tailed) and a statistical power of 0.80, the sample size required for detecting this effect size is approximately N = 134.

Rates of social exclusion are higher in the southern half of Spain (Bayón-Calvo et al., 2021; EAPN, 2023), which constituted the focus of this research. The high schools from areas at risk of social exclusion in Seville (Andalusia) were sampled. These areas are labelled as Zones in Need of Social Transformation (ZNST) by local administrators. We used purposeful sampling and selected two high schools from different neighborhoods, which, although both were in ZNST, differed in the degree of risk they presented. According to their principals and teachers, High School A presented a higher proportion of migrant students and ethnic minorities (i.e. gypsy population), a higher rate of conflict, and greater deterioration of the school facilities and the neighborhood. Likewise, all the students in the High School A come from the same neighborhood in which the school is located, which according to City Hall of Seville (2017) has the highest rate of absenteeism in the city. According to the literature, these variables are associated to increased risk for dropping out (Cordero et al., 2014; Mena et al., 2010; Romero & Hernández, 2019). However, the students from High School B come both from the high school’s neighborhood and from other surrounding neighborhoods of lower risk, providing a different profile of school. More differences between both high schools can be seen in the “Results” section, where sociodemographic data is presented.

All students from post-compulsory secondary education (Bachillerato, ISCED Level 3) from the aforementioned high schools were studied. In Spain, this academic course is comprised of 2 years and traditionally prepares the students for accessing university studies. Following Masten (2007)’s definition, we considered these students presented a resilience trajectory. They completed compulsory education and continue in the education system beyond that level, showing proper academic performance in the face of risk (Lessard et al., 2014; Longás et al., 2019). Students from vocational training (Grado Medio, ISCED Level 3) could also fall within this definition but were not the focus of the present research. The final sample was comprised of 132 Bachillerato students (statistical power = 0.795).

The variables considered for the study were high school, academic year, and gender (Table 1). The mean age of the participants was 17.43. Concerning the students’ educational gap, 21.4% of them were older than the age expected for their grade level, with a gap from 1 to 7 years. In the most extreme cases, these students had dropped out of school and returned several years later.

Table 1  Sociodemographic data
Full size table

Many of the students came from single-parent (21.9%), extended (15.1%), or reunited (3.8%) families. In general, they belonged to large families, with an average of 3.82 persons in the household. A mean of 1.37 people in the family had an income, which essentially came from low-skilled or temporary jobs. 44.7% were low-skilled or unskilled, unstable jobs; 25.2% were in medium-skilled jobs or which involved merit selection or passing an exam, while only 3.5% were in high-skilled jobs. Also, 24.8% were unemployed or housewives and 1.7% were retired or pensioned. Finally, parents’ formal education was predominantly low, with 11.8% of the parents having no formal education, 41.4% having basic education (primary or secondary school), 33.8% having intermediate education (post-compulsory secondary school or vocational training), and 12.9% having higher education (technical college or university).

Instruments

A sociodemographic questionnaire and, as a self-report measure of learner identity, a modified version of the Twenty Statement Test (TST) (Kuhn & McPartland, 1954; Santamaría et al., 2010; Santamaría & Cubero, 2019, among others) was applied. The original version consisted of 20 sentences beginning with “I am.” The participants were asked to complete them with adjectives, nouns, or short phrases describing what they thought of themselves, or what others thought of them, whether positive or negative. The TST does not provide a priori information like other questionnaires or Likert-type scales, and therefore has the advantage of a more qualitative and flexible approach to the dimensions of meaning individuals construct about themselves. Based on previous research from the literature, some modifications to the instrument were made in order to adapt it to the goals and the context of assessment:

  1. a)

    Reducing the 20 items to 10. Previous studies have shown participants have difficulty in completing all 20 sentences (Santamaría et al., 2010).

  2. b)

    To specifically ask their characteristics as a student to study their learner identity.

  3. c)

    Changing the “I am” (“Yo soy”) item structure for only “I” (“Yo”). The English “I am” may be translated by two different verbs in Spanish, either “ser” or “estar,” which offers a richer range of possible answers than its direct translation in Spanish, “Yo soy.” This change enabled a similar range of answers for the Spanish population.

  4. d)

    The instructions specified that answers could also include membership to groups or personal relationships with others. The reason was to avoid the individualistic bias shown by previous studies (Santamaría et al., 2010), since these elements tend to be underrepresented.

G*power ver. 3.1 and SPSS ver. 25 were used for calculating the a priori estimation of the sample and analyzing the data, respectively.

Procedure

Approval was obtained from the local ethics committee (Granting Body: Portal de Ética de la Investigación Biomédica de Andalucía; Protocol Number: 0191-N-18). For the selection of participants, we obtained research approval and the support of the local education administration. Then, the high schools were contacted. Informed consent was obtained from parents and students over 18 years old. The dates for the data gathering sessions were agreed with the teachers from each high school.

A pilot study was carried out in the same high schools and classrooms to ensure that the instructions were understood and that there were no problems with the TST and sociodemographic questionnaire. Some modifications were done to the latter to facilitate its completion and avoid missing data. The definitive data were collected by researchers from the research group over 2 years, applying the TST to all post-compulsory students at both high schools. The instruments were completed in writing and during class time, after a brief explanation of the research and the procedure to be followed. The students took approximately 15 to 20 min to complete them. The students did not receive financial compensation for their participation.

Data analysis

A category system was applied to analyze the self-descriptions that students provided in the TST. Each statement was categorized under four different dimensions. These were organization of the self, emotional valence, plane of action, and thematic reference (Table 2). For the reliability analysis of each dimension, 20% of participants were randomly selected from the total sample and their self-descriptions as students were independently coded by two researchers, authors of the present work. Subsequently, the kappa index was calculated as a measure of inter-rater reliability (Cohen, 1960).

The first three dimensions were conformed a priori and have already been used in the literature. The organization of the self dimension is based on the distinction of Triandis (1989) between the private, public, and collective self (Santamaría et al., 2010; Yuste et al., 2021). A second dimension focuses on the emotional valence of the participants’ description of the self, distinguishing between positive, negative, and ambivalent (Santamaría & Cubero, 2019). The third dimension analyzes the plane of actions of self-descriptions. It moves in relation to the trichotomy descriptions, actions, and evaluations (Yuste et al., 2021). These dimensions showed good reliability (Table 2) according to Cohen (1960).

Table 2 Organization of the self, emotional valence, and plane of action dimensions category system
Full size table
Table 3  Thematic reference dimension category system
Full size table

Finally, an inductive thematic analysis of the self-descriptions was developed, based on the participants’ answers (Clarke et al., 2015). Seventy-six different categories were iteratively elaborated a posteriori, which were very similar to the descriptions that the participants themselves filled out in the TST. Subsequently, they were recategorized into hierarchically higher categories, until five final general categories were created: academic world, social relations, ways of being, work, and finally, leisure and other activities (Matías-García et al., 2023). Given the interest in academics, data are also presented for the five subcategories that form Academic World at its previous level (Table 3). For reasons of space, microlevel categories will be presented only if they show statistical differences. The kappa index was calculated from the micro level, from 76 different categories, giving a value of 0.866, considered a very good value (Cohen, 1960).

Chi-squared (X2), Fishers’ exact, and Mann-Whitney U tests were used depending on data type. Percentage of appearance of TST categories was used for analysis. Cramer’s V and Rosenthal’s r were calculated as effect size measures and interpreted according to Cohen (1988).

Results

First, the descriptive data obtained from the TST will be shown. Subsequently, data referring to the comparison between the different explanatory variables, school, academic year, and gender will be presented.

Descriptive data

The participants’ self-descriptions were mainly personal, positive, evaluative, and referred to the academic world (Table 4). The organization of the self shows a small percentage of relational categories and minimal group categories, while in the emotional valence, negative valence is found in one-third of the responses and ambivalent in one-tenth. In the plane of action, very few actions and even fewer descriptions were categorized. Finally, in the thematic reference dimension, in addition to the high percentage referring to the academic world, we found some categories related to social relations and ways of being, and very few referring to work or leisure.

Table 4 Descriptive data (mean of percentages)
Full size table

Within the academic world, the highest percentage of self-descriptions are those referring to the abilities and activities of the students. The next most numerous category would be academic emotions. Finally, with less than 5%, we find the categories global conception, academic relationships, and academic goals.

Differences in terms of the explanatory variables high school, academic year, and gender

High school

First, further differences will be presented between the two high schools to help contextualize and interpret data, based on the analysis of the sociodemographic questionnaire. Subsequently, the differences referred to the TST will be shown.

High School A had a lower number of students who reached the second year of post-compulsory education (34.1%) than High School B (52.3%), X2 (1, N = 132) = 3.9, p = .048, V = 0.17). Likewise, High School A’s students are significantly older (M = 18.3, DT = 1.71), U = 985, p < .001, r = .41. In terms of family configuration, High School A also showed a higher percentage of extended family (27.3%) than High School B (9.1%), X2 (1, N = 132) = 7.54, p = .006, V = 0.24). Father’s (M = 1.93, SD = 0.73) and mother’s (M = 1.97, SD = 0.67) educational levels were also lower relative to High School B (M = 2.82, SD = 0.95 and M = 2.61, SD = 0.71, respectively), U = 500, p < .001, r = .43 y U = 740, p < .001, r = .38, respectively. The parents’ job analysis also revealed significant socioeconomic differences. Students at High School A came from families whose parents had mostly low-level jobs (81.5%). Few had medium-level jobs (11.1%) and there were no parents with high-level jobs. Finally, 7.4% were unemployed. On the other hand, at High School B, their families had mostly medium level jobs (53.6%). A lower percentage had low-level jobs (32.1%), and few had high level jobs (1.8%). Finally, 1.8% were retired and/or pensioned and 10.7% were unemployed. These differences were significant (p < .001, Fisher’s exact test, V = 0.47). All these data confirmed that High School A presents a higher degree of risk for dropout than High School B.

Data referring to the analysis of self-descriptions as a function of the high schools will be next presented (Table 5). High School A has a significantly higher percentage of unspecific relational and positive self-descriptions. There also were more self-descriptions related to academic relationships and more social (non-academic) relationships.

Table 5 Percentage data referring to the categorization of the TST according to the high school
Full size table

On the other hand, High School B presents a higher percentage of personal self-descriptions and of ambivalent emotional valence, and referred more to the academic world. Within the academic world, we also find more self-descriptions related to abilities and activities and to academic emotions in High School B. No significant differences were found in the plane of action.

In addition, there were significant differences related to the micro-level categories that make up the thematic reference analysis (Table 6). For reasons of space, only statistically significant categories are included.

Table 6 Percentage data referring to the micro-level categorization of the TST according to the high school
Full size table

Thus, we find that students from High School A use more self-descriptions categorized as referring to supportive and helpful relationships with classmates, high non-academic relational abilities, and high attendance and punctuality. However, they also present more self-descriptions of overall poor student conception and other people’s (perceived as) negative academic motivations, in which participants feel pressured or obligated to perform their studies. On the other hand, students from High School B use more self-descriptions that allude to their capacity for effort and organization as a student, either low or high, and to high intelligence. However, they present more descriptions referring to low study and learning and to negative academic emotions, such as anxiety or felling overwhelmed.

All significant differences related to their high school have a small-moderate effect size.

Academic year

Relating to organization of the self, second-year students presented a significantly higher percentage of personal self-descriptions (M = 86.2, SD = 12.3) than first-year students (M = 75.6, SD = 20.4), U = 1548, p = .004, with a small effect size (r = .25). No significant differences were found for emotional valence. In the plane of action, there were less descriptive self-descriptions in second-year students (M = 0, SD = 0) than in first-year students (M = 1.65, SD = 6.81), U = 1983, p = .021, with a small effect size (r = .2). In the thematic reference, second-year participants presented a higher percentage of academic world categories (M = 88.6, SD = 14.2) than the first-year students (M = 75, SD = 25.7), U = 1477, p = .001, with a small effect size (r = .28).

Likewise, within the academic world, second-year students used more self-descriptions referring to abilities and activities (M = 64.4, SD = 19.2) than first-year students (M = 52.2, SD = 23.2), U = 1526, p = .003, with a small effect size (r = .25). On the other hand, second-year students presented less self-descriptions related to social relations (M = 3.55, SD = 6.66) than the first-year students (M = 13.9, SD = 17.1), U = 1402, p < .001, with a small effect size (r = .34).

In addition, there also were significant differences related to micro-level categories (Table 7). The second-year students presented more self-descriptions related to their capacity for effort and organization, whether low or high, high participation in class, and positive academic emotions, such as satisfaction with their studies.

Table 7 Percentage data referring to TST micro-level categorization according to academic year
Full size table

However, they presented a higher proportion of categories related to low attention in class. As for the first-year students, they presented a greater number of categories related to their behavior in class, either behaving calmly or nervously in class, non-academic relational skills, and high creativity. All these significant differences have small to moderate effect sizes.

Gender

No significant gender differences were found in organization of the self, emotional valence, or plane of action. In thematic reference, female students presented more academic world categories (M = 84.9, SD = 21.6) than males (M = 78.4, SD = 22.4), U = 1612, p = .011, with a small effect size (r = .22). However, males presented more ways of being categories (M = 9.7, SD = 14.5) than females (M = 6.03, SD = 12.2), U = 1740, p = .032, with a small effect size (r = .19).

In the micro-level analysis, there also were statistical differences (Table 8). Female students presented a greater number of self-descriptions referring to a high capacity for effort and organization. However, they presented more self-descriptions referring to a low ability to pass or get a good grade, as well as more self-descriptions referring to both academic optimism and pessimism. On the other hand, male students had more self-descriptions referring to support and help among classmates, calm classroom behavior, and high creativity. All these differences showed small effect sizes.

Table 8 Percentage data referring to micro-level categorization of TST according to gender
Full size table

Discussion

The present study analyzed the learner identity of students presenting a trajectory of educational resilience in contexts at risk of social exclusion (Lessard et al., 2014, Longás, 2019; Toland & Carrigan, 2011). For this purpose, the participants’ self-descriptions as students were studied through the TST. Understanding how these students elaborate their learner identity allows us to better understand educational resilience in these contexts.

Students who drop out generally allude to a lack of ability, motivation, and social value of school (Romero & Hernández, 2019). Leaving the education system is viewed as a success rather than a failure by these students, as they believe they can stop wasting time and engage in more productive activities (Mena et al., 2010). However, when asked about their learner identity, the study’s participants employed mainly personal, positive, evaluative self-descriptions, linked to the academic world. These descriptions are associated with higher academic performance and mastery-focused academic goals (Hansen & Henderson, 2019; Li et al., 2020).

In addition, the most common micro-level category was effort in academic tasks, whether high or low (see Tables 6 or 7). For the participants, making an effort is one of their primary characteristics as learners. In this sense, incremental beliefs of one’s own abilities, that is, beliefs that one’s ability and intelligence can be improved and developed through effort, have been related to greater motivation, academic performance, task mastery goals, persistence in the face of failure, and greater academic resilience, especially in students at risk of social exclusion (Burnette et al., 2013; Matías-García & Cubero-Pérez, 2021; Sarrasin et al., 2018). Likewise, interventions on these beliefs have been shown to be especially beneficial to at-risk students (Sarrasin et al., 2018). Explaining successes and failures in terms of effort has the advantage of making academic difficulties controllable, which gives them agency over their own life trajectory (Burnette et al., 2013; Skinner & Zimmer-Gembeck, 2011). This alternative to deficit discourses in at-risk students who show a trajectory of resilience is consistent with previous research in the literature (Cubero-Pérez et al., 2023; Matias-Garcia et al., 2024).

Thus, identity serves as an attributional resource that explains the students’ successes and failures. Despite being in ZNST, these students identify themselves as students who make an effort and are capable and participative in academic settings. This is identitarily coherent with pursuing further studies, allowing them to make such a decision. The importance of building positive learner identities has also been highlighted by other authors (Rosales & Cubero, 2016) and their development depends on both the context and the students’ interactions with significant others, such as families, teachers, and classmates (Matías-García, 2022; Matias-Garcia et al., 2024; McFarland et al., 2016). The next section will discuss the impact of risk, academic experience, and gender on the development of learner identity.

Although both high schools are located in contexts at risk of social exclusion, they differed in the degree of risk they presented. High School A had older students, resulting in a greater educational gap. Its students only came from ZNST and had families with a significantly lower socioeconomic status. Additionally, High School A had a higher number of migrant students, a higher rate of conflict, and greater facility deterioration.

In High School A, there was a higher frequency of self-descriptions related to supporting and helping classmates as well as attending class. In contrast, High School B had more self-descriptions related to abilities and activities. This suggests that, even though both high schools are in ZNST, the learning relationships and interactions between teachers and students that occur within them are different (McFarland et al., 2016). The higher rate of conflict and absenteeism in High School A indicates that one of the main tasks of its teachers is to ensure that students attend class and maintain good student behavior, in terms of them staying quiet during lessons and not provoke fights with others. Thus, those High School A students who do not fail and who remain in the educational system highlight to a greater extent their good behavior and attendance as something important in their role as students. They elaborate their learner identity in relation to others (Sedikides & Brewer, 2001), as a way of differentiating themselves from more troubling students. In contrast, in High School B, the students focused more on their own abilities as students (study, intelligence, effort, organization, etc.), characteristics that education usually values highly in the absence of other difficulties.

According to the data provided by the TST, the students from High School A had the most interdependent selves, based on supportive relationships among peers and friends. On the other hand, the High School B students’ selves were more personal and associated with education. They also presented a more ambivalent and less positive emotional valence, showing a certain greater degree of reflection on their own self-descriptions (De la Mata et al., 2015).

These differences are coherent with cross-cultural research. The literature on self-construals advocates a broad distinction between independent and interdependent selves (De la Mata et al., 2015; 2019; Greenfield, 2009; Kitayama et al., 2007). People with an independent self-construal view the self as unique, private, and autonomous, whereas people with an interdependent self-construal view the self as related to others, incorporating and referencing the views of other people and groups in their identity. Interdependent selves are related to rural societies and low levels of schooling, whereas independent selves are associated with urban societies which has a greater role for school (De la Mata et al., 2015; 2019; Greenfield, 2009; Kitayama et al., 2007). In our sample, the parents from High School A had a lower level of formal education than those from High School B families.

The idea that educational experience fosters the development of independent selves is supported by the significant differences observed between first- and second-year students (Matías-García et al. 2023). Second-year students, who are more acclimated to the school environment (Mena et al., 2010) and have more educational experience than first-year students, also exhibited more personal self-descriptions and greater association with the academic world (effort, organization, participation, emotions) than first-year students. In contrast, first-year students referenced their social relationships and behavior in class more frequently than second-year students. The educational discourse in class interaction influenced the development of students’ identities, leading to increasingly independent selves associated with school characteristics (De la Mata et al., 2015; 2019). Matías-García et al. (2023) show how this process takes place in the development of future selves as well.

However, it is important to note that both profiles of learner identities in the sample, independent or interdependent, demonstrate a trajectory of resilience in their respective high schools and are strongly associated with academic selves. Therefore, it is important to emphasize the relevance of interdependent academic identities in the especially disadvantaged schools and neighborhoods, such as High School A. Cultivating such identities, as well as building good relationships among students and with teachers, could be effective ways to achieve good school integration of students at risk of social exclusion (Longás et al., 2019).

In fact, the results of High School A show that fostering an autonomous self does not necessarily imply the undermining of relational selves (Santamaría et al., 2010). In contrast to the consideration of two major models of the self, the independent and the interdependent self (Markus & Kitayama, 1991), there are authors such as Kagitcibasi (2007) who reject the traditional consideration of autonomy as opposed to relatedness, which has been predominantly assumed in psychology and in Western thought in general. It is precisely in the specific field of education that this idea has been reinforced in recent years, pointing out that formal education should not only promote aspects of rationality, but also relationality.

Regarding gender, research shows that dropout rates are higher in males than in females (Eurostat, 2022). In the present study’s data, female students presented more school-associated identities than male students, indicating greater adaptation and integration in the educational system. Qualitative research suggests that working-class male students may face challenges in maintaining their status within their peer group while integrating into school due to male gender roles (Reay, 2018). As a result, they may be more reticent than female students to identify with academic selves. Although female students have more academic success (Eurostat, 2022) and display positive self-descriptions regarding their school performance (such as high capacity for effort and organization), they also exhibited more categories of academic pessimism and about obtaining poor grades. In this sense, the literature demonstrates that female students experience greater pressure in school, particularly in certain areas (Kim et al., 2018).

In turn, male students presented fewer academic categories and more categories of calm behavior in class, while male students traditionally manifest more behavioral problems (Moreno et al., 2012). As mentioned, students who exhibit resilience and do not experience such issues, or at least experience them to a lesser extent, tend to identify these characteristics of themselves as relevant to their identity as a student.

Limitations

Regarding limitations, this research understood resilience as the completion of secondary education and its maintenance in the educational system as a goal in itself. However, the students from this research are still at risk of ESL. Economic needs and other factors may still force them out of the educational system (Mena et al., 2010, Romero & Hernández, 2019). Therefore, measuring other academic adaptation indices, such as the number of passed subjects or the completion of post-compulsory studies, could complement the results. In addition to academic achievement, mental health indicators could be further analyzed in this population. Some students might progress with high levels of stress, discomfort, or “suffering great pressure,” which could also be a problem to address.

Also, it would be worthwhile to analyze the relationship between the students’ learner identity and other non-academic characteristics, such as emotional and affective development, or the social relationships they have. These students face a multitude of difficulties, both in the school context and in the family or peer context, which are interrelated (Toland & Carrigan, 2011).

Finally, our results have explanatory value only for students from the post-compulsory bachillerato, a course that traditionally prepares them for university studies. Our sample did not include students from different academically resilient trajectories, such as students from vocational training Grado Medio. These students might develop different types of learner identities, which should be further explored in future research.

Conclusions

This study examined the characteristics of the learner identity of students who exhibit a resilient trajectory. These characteristics align with the healthy learner identity construct found in the literature (Rosales & Cubero, 2016). Through their participation in class and interaction with teachers and other students, they elaborated their learner identity in a positive way, while being connected to others. They rejected deficit discourses about the self, common in areas at risk of social exclusion, and built an agentic learner identity mostly based on effort.

The identities constructed by the participants are influenced by the degree of risk present in their high school (and neighborhood), gender, and educational experience. These identities may be more independent or interdependent, more focused on behavior and attendance or on capabilities as learners. In one way or another, all of these students remain resiliently in the educational system beyond the compulsory.

The identification of these characteristics adds necessary information for intervening in ZNST, aligned with national, European, and international goals (Agenda 2030's Sustainable Development Goals; United Nations, n.d.). The school, in conjunction with families, should promote the construction of positive learner identities attending to the emotional needs that students face when navigating the various challenges presented by their school and context. Identifying and reinforcing characteristics such as effort, attendance, behaving calmly during lessons, being supportive to other students, or the overall conception of being a good student can help build identities consistent with remaining in the educational system. This is especially important for ZNST students, as it can reduce feelings of disengagement and promote adaptive behaviors and persistence.