Introduction

The present study aims to explore in-depth the factors that influence students’ occupational aspirations at the end of compulsory secondary education, in an attempt to respond to the need to create a more consolidated link between theoretical and empirical studies and to enable the most appropriate educative intervention to be developed. Throughout a person’s life, few decisions have as decisive an influence as choosing a career path or profession (Hackett, 1999), which is considered as one of the main areas of development during adolescence (Super, 1990). Adolescents’ ambitions are associated with educational and occupational attainment and provide an important window from which to watch young people contemplating their choices and also how they see themselves (Johnson & Reynolds, 2013). This process, in which people either choose, or avoid, certain activities, involves cognitive mechanisms in which the individual assesses his/her own skills and compares them with those required to perform these actions. This choice is, therefore, strongly influenced by the person’s self-concept or feeling of self-efficacy (Bandura, 1991).

Vocational expectations and aspirations are conditioned by several individual, cognitive and non-cognitive, contextual and social factors, which largely determine an individual’s decisions and consequently their personal and professional future. Authors such as Andres et al. (1999) consider that these occupational aspirations and expectations influence the academic decisions the students make and their willingness to initiate new adult roles. According to Bandura et al. (2001), the choices an individual makes during his/her educational training will shape the trajectory of their lives and will determine which of their potentials they will cultivate, and which will be left undeveloped. This explains why occupational aspirations during adolescence have an important predictive value on future occupational achievements (Schoon & Parsons, 2002). At this point, it is important to clarify the difference between occupational “aspirations”, occupational “expectations” and professional “interest”. Although there is no clear and unanimous distinction among these terms in the literature, vocational “aspirations” reflect an attraction towards, or a preference for, an occupation and can be considered as an expression of an individual’s ideal professional goals at any given time. Occupational “expectations” refer to realistic or probable expressions of an individual’s professional objectives (ideal aspirations versus real expectations) (Rojewski, 2005). Finally, professional “interest” reflects a person’s inclination towards activities related to a specific profession (Rodríguez-Mateo et al., 2017) and is, therefore, qualified as the driving force behind an individual’s vocational behaviour, owing to its motivating and reinforcing capacity (Rivas, 2007).

Numerous research projects have focused on aiming to better understand the role of an individual’s choices in a vocational setting. Several studies have corroborated the existence of a association between academic results in secondary education (middle and higher) and academic, and subsequent occupational expectations (Bandura et al., 2001; Chow et al., 2012; Gao & Eccles, 2020; Guo et al., 2015; Lauermann et al., 2017; Mau, 1995). There is also clear evidence for the influence on educational expectations of some demographic factors. Among the influencing dimensions examined, we can highlight gender, as men tend to show a greater interest in scientific or technological professions (Mozahem et al., 2019; Sainz & Meneses, 2018). In turn, in middle and high school, women have greater educational and occupational aspirations than men (Howard et al., 2011; Mau, 1995; Mau & Bikos, 2000). Together with gender, a higher economic and social status (Bandura et al., 2001), positive identification and parents´ values (Jodl et al., 2001), and the family’s educational style based on maternal responsiveness, complementary congruence and self-efficacy (Ambiel et al., 2019) are also linked to occupational aspirations. Hence, students with parents with higher educational levels tend to present higher academic and occupational aspirations (Andres et al., 1999; Cochran et al., 2011; Howard et al., 2011; Mau & Bikos, 2000). In addition to the social setting, personal factors such as indecisiveness, dysfunctional beliefs and a lack of motivation, affecting the process of adolescents’ decision-making, must also be taken into account (Bacanli, 2016; Mau & Li, 2018; Tolentino et al., 2019; Vergauwe et al., 2017).

Studies that analyses the sociocognitive factors associated with vocational choice trajectories in certain professional fields, such as mathematics and science (Mau & Li, 2018; Moore & Burrus, 2019; Vázquez & Blanco-Blanco, 2019), emphasized the importance of self-efficacy, beliefs, attitudes, results expectations, interests and occupational aspirations, and any perceived social support or barriers. Other studies have explored in-depth the difficulties associated with deciding which career to follow in relation to an individual’s vocational interests, intelligence (Albion & Fogarty, 2002; Kirdök & Harman, 2018) and also to variables linked to personality (Kowalski et al., 2017; Volodina et al., 2015). The relationship between this latter type of variable and vocational behaviour has been widely studied as reflected by the literature review by Hirschfeld and Van Scotter (2019). Thus, researchers from a broad spectrum of disciplines have studied individual and contextual determinants of individuals’ choices in a vocational setting and have included a wide groups of factors related to the processes of career choice and construction. In a task such as this, it is essential to consider the structures of complex relationships between the elements that influence these processes. In this regard, a socio-cognitive approach is appropriate for exploring the effects of factors associated with the different spheres of individual lives. More specifically, the Social Cognitive Career Theory (SCCT) (Lent et al., 1994, 2000) provides a theoretical framework that allows us to explore in-depth the factors that influence students’ occupational aspirations. This consists of interconnected models of career and academic interest, choice, and performance controlled through the interaction of cognitive processes (self-efficacy, outcome expectations, interests, choice and performance goals) with personal and environmental influences. Some of its applications, such as the career self-management model (Lent & Brown, 2013), have been used to study career exploration and decision-making behaviour (Fort & Murariu, 2018; Lent et al., 2016).

Bearing this in mind, in today’s society vocational guidance is acquiring a key role owing to its importance in improving the educational choices and the individuals' choices in labour market, and fomenting social equity (OECD, 2004). The students, however, do not always perceive this importance, and tend to consider the career guidance received in school as having little influence on their vocational decisions, regarding it as less important than the opinion of their parents, friends, siblings or personal preferences (Cepero, 2009; Fernández et al., 2007). This implies that the orientation services provided are in some way deficient, and the scientific community specialized in educational and vocational guidance advocates for the career development process and the learning of competences and skills to take place during the entire period of schooling (Gore et al., 2017) and throughout an individual’s lifetime (Pablo-Lerchundi et al., 2015). This approach is also aligned with the results of the study of Azpilicueta et al. (2019) which emphasizes that, although career guidance should be provided for life, the final years of secondary school constitute a especially important stage, as this is when the process of exploration of educational-occupational objectives is completed, leading to crucial decisions such as seeking employment or going on to university studies.

Nonetheless, in spite of the strong implications of adolescents’ occupational aspirations for education and professional attainment in adulthood, the specificity and way in which these aspirations are developed have rarely been studied (Gao & Eccles, 2020). The learning of competences related to vocational guidance, such as planning for the academic and professional project and making decisions about employment and training, is strategic for providing quality education. The educative system must not shirk its responsibility in this respect and should look beyond those characteristics related to the student for which intervention is not possible, to focus its efforts on providing orientation systems that can help to guarantee that the best decisions are made during this critical stage of adolescence. In most countries, adolescents of around 15 to 16 years old are required to make decisions about their educational and professional future, since this is when different training itineraries can be clearly distinguished. In Spain, this decision-making process is particularly important because this also coincides with the end of compulsory education and students can legally enter the labour market (Articles 6 and 7 of Royal Decree 2/2015, 2015).

This research has a dual objective. On the one hand, it aims to prioritize the cognitive and contextual variables in relation to the importance they play in explaining the occupational aspirations of 15-year-old students. On the other hand, it endeavours to identify the type of variables that can best discriminate between subjects with higher or lower occupational aspirations, thus establishing different profiles. In this way, by following an exploratory and holistic quantitative approach, we have searched for information regarding which actions would most contribute to an educational and vocational guidance practice that best satisfies the students’ different needs.

Methods

This work carries out a secondary analysis of the database derived from the PISA2018 test. The PISA program (Programme of International Student Assessment) is an international study by the Organisation for Economic Co-operation and Development (OECD) that assesses the extent to which 15-year-old students approaching the end of their compulsory education have acquired the knowledge and competences required to participate fully in modern society (OECD, 2019a).

Population and sample

The study population is comprised 15-year-old Spanish students aged between 15 years and 3 months and 16 years and 2 months old. The sample was selected by stratified two-stage sampling in which the schools were selected in the first stage, followed by selection of the students (for more information see the OECD, 2019b). In this way, the study sample is comprised 35,943 Spanish students who had participated in the PISA test of 2018. These students were enrolled in 1089 schools of which 63.9% were state schools, 28.9% state-assisted and 7.1% were private.

Instruments and variables

For this study, information was compiled from four of the five questionnaires used with students in the PISA 2018 test: the student questionnaire, the well-being questionnaire, the educational career questionnaire and the ICT familiarity questionnaire. The student questionnaire is completed by students after finishing the skills’ assessments and takes approximately 35 min to do (OECD, 2019a). This instrument assesses information about the student and his/her family environment, their view on life, their experience at school, school timetable and learning time. Moreover, since the main domain of the 2018 test was reading comprehension, in the present research we have also incorporated some questions specifically related to reading (OECD, 2019a). The educational career questionnaire compiles information about schooling, the preparation of students for their future professional career, employment and educational aspirations and learning support. This was optional and the participating countries decided whether or not they wanted to apply it. The well-being questionnaire is also optional and was used for the first time in the PISA test of 2018. This looks for information about students’ perception about their health, life satisfaction, social connections and activities they do in and outside of school. Finally, the ICT familiarity questionnaire is also optional and aims to determine access to and use of the ICT at home and school, and the students’ attitudes to ICT.

The students’ expected occupational status (BSMJ) has been introduced as a dependent variable as a measure of “occupational aspirations”. This variable is calculated from the students’ response to the question “which job would you like to have when you are 30 years old”. The job descriptions provided by the students are coded according to the International Standard Classification of Occupations (ISCO) and then transformed to the International Socio-Economic Index of Occupational Status (ISEI). After recoding the ISCO codes in the ISEI index, the result corresponds to the scores of the students’ expected employment status (BSMJ) (OECD, 2017), where higher ISEI scores indicate higher expected levels of occupational status (high skilled professions); the distribution of these scores shows a mean of 68.3, a standard deviation equal to 17.56, a maximum value of 88.96 and a minimum value of 11.01.

The 81 items from the educational career questionnaire and 4 indicators constructed from students’ response to this questionnaire have been introduced as predictors, enabling their individual or group response to be determined. Moreover, 54 indicators have also been included that were calculated from responses to the students’ questionnaire, 2 indicators of well-being and 11 indicators from the ICT familiarity questionnaire. The indicators introduced, together with items from the educational career questionnaire and gender, are compiled in Annex 1. It is worth noting that since we use a data mining approach that involves the examination of large data bases, the use of regression trees has the advantage of being based on the discovery of the most important variables rather than corresponding to a predetermined model (Hernández Orallo et al., 2004).

Analyses

Data were analysed by the decision tree approach. Of the algorithms available we opted for the CART algorithm (Classification and Regression Trees), implemented by Breiman et al. (1984), as this offers binary segmentation and an estimation of the relative importance of each independent variable. As we are working with a dependent variable measured at the scale level, the approach followed corresponds to regression trees. In a complementary manner, we have used the CHAID algorithm (Chi Automatic Interaction Detection) to identify independent variables that interact significantly with the dependent variable (Kass, 1980).

This nonparametric technique of regression trees enables us to adequately manage the complexity of data in a study like ours which has a large variety of measuring levels (dependent variable with a scale level of measurement and nominal, ordinal or scale independent variables). This makes it possible to operate in a single analysis without having to transform the original data scale (Asensio et al., 2018). They are also robust in the presence of outliers and missing values (Streifer & Schumann, 2005).

The procedure followed is the one proposed by Asensio et al. (2018), which consists of the following four stages. First of all, an estimation of the initial model with the CART procedure, introducing “vocational expectations” as the dependent variable and the items and indicators from the questionnaires described in the previous section as predictors. Next, using CART the importance of each independent variable is calculated as the sum of the reduced impurity resulting from a better division of this variable into each of the nodes (Breiman et al., 1984, p. 147). Subsequently, an estimation of the initial models is carried out, using in this case the CHAID algorithm, which provides a selection of predictors. This can determine the variables that interact with the dependent variable. Finally, the variables are selected, using CART, with a normalised importance of at least 10%, indicating in each case if they have been included as the influencing variable (significant interaction with the dependent variable) by the complementary test with the CHAID procedure.

Results

In relation to model's performance, the global values show a variance of 308.527 and a value of risk of 253.68. The predictors included in the model explain 17.7% of the differences observed in the variable "students' expected occupational status". In turn, the overall predictive value of the model was 42.2%.

With regards to the first of the proposed objectives, below we present a synthesis of the most important variables in the model. As described in detail in the procedures section, to obtain this selection of variables the results of applying the algorithms CHAID and CART are combined. Table 1 shows the selection of variables with a normalised importance of at least 10% and the order of importance in the model (considering this percentage of variance), provided by the CART algorithm. The interaction of this variable with the dependent variable, provided by the CHAID algorithm, is also included.

Table 1 Selected independent variables and importance values (total and normalised)
Full size table

Grade repetition is presented as the first influencing variable. The mere fact of repeating a grade appears to significantly hamper the students’ occupational expectations at the young age of 15. Next, we can see that mastery goal orientation is an important variable to explain occupational aspirations. Other variables such as enjoying or liking reading and self-concept of reading competence present normalised importance of 36.9% and 34.5%, respectively.

In an attempt to identify the type of variables that emerge as the most discriminatory between subjects with greater or lesser occupational aspirations (Objective 2), a detailed analysis of the resulting tree is presented below. Owing to the vast amount of data provided, the diagram has been divided into different branches depending on their growth. In other words, Figure 1 shows the main trunk with the predictors responsible for the first divisions, indicating in which figure the subsequent growth of the tree is represented (Figures 2, 3, 4, 5).

Figure 1
figure 1

Main trunk of the regression tree with student´s expected occupational status (SEI) as the dependent variable

Full size image

As can be observed in Figure 1, the first influencing variable that emerges is grade repetition, which has a pronounced effect on the students’ occupational aspirations. Students who have not repeated any grade reach a mean score of 71.5 points, versus 58.3 mean points for students who have repeated at least one grade. For both groups of students (repeaters and non-repeaters alike), the second influencing variable is gender, with women having the highest aspirations in both cases. This difference is even clearer in the group of non-repeaters (68.7 points for the males versus 74.0 for the females).

In Figure 2 the first influencing variable for boys who have not repeated grade is the score obtained in mastery goal orientation. Male students with higher levels of mastery goal orientation (Node 8), have higher levels of occupational aspiration (69.6 versus 63.0). It is striking that for both groups (greater and lesser mastery goal orientation), the next influencing variables are associated with reading. For students with lower mastery goal orientation, a good self-concept of reading improves their aspirations (65.4 versus 58.5), and for those with greater mastery goal orientation the variable joy/liking is the most differentiating, in favour of students who most enjoy reading (Nodes 17 and 18). In Node 16, we can see how an additional distinction can be discerned for students with the highest level of self-concept of reading. This is in favour of boys with greater ability to assess credibility, who reach a mean of occupational expectations of 67.2 points (Node 32). Finally, from Node 17 a division appears in favour of students with a higher economic, social and cultural index who reach higher occupational expectations (70.4, Node 34) than male students from the same group with an ESCS below this value (66.4, Node 33).

Figure 2
figure 2

Branch A showing boy students who have not repeated grade with students’ expected occupational status (SEI) as the dependent variable

Full size image

With an in-depth analysis of this branch, we can identify the profile of students who have not repeated grade and who have the highest occupational aspirations (Node 18). These are characterized by high mastery goal orientation and greater enjoyment of reading, with mean occupational aspirations of 74.1 points. At the other extreme, we can find male students in this group with the poorest occupational expectations (identified in Node 15, mean 58.5) who, in spite of not having repeated a grade, have lower scores in mastery goal orientation and a lower reading self-concept.

Analysing Branch B (Figure 3) we can observe how the first influencing variable in the expected occupational status of students girls who have repeated grade is mastery goal orientation, which is also the first influencing variable in the group of non-repeaters. For students with lower mastery goal orientation scores (Node 9), the next influencing variable is the Index of Economic, Social and Cultural Status (ESCS). Students with low ESCS have a mean of occupational aspirations of 70.2 (Node 19), versus a mean of 73.8 for those with a higher ESCS (Node 20). The influencing variable for students with a lower ESCS Index is attitude towards school learning activities and for those with higher levels of ESCS it is the importance attached to their qualifications in decisions they make about their occupational future. Girls who consider their qualifications to be very important have the highest occupational aspirations (75.5) (Node 37 versus Node 38).

Figure 3
figure 3

Branch B showing girl students who have not repeated grade with students’ expected occupational status (SEI) as the dependent variable

Full size image

If we study the growth of the tree for non-repeater students with the highest mastery goal orientation scores (Node 10), we find that the first influencing variable is self-concept of reading, with students with a higher self-concept of reading having the greatest occupational aspirations (76.9 versus 73.7) (Nodes 21 and 22). Within this group, the next influencing variable is the time spent during the week studying sciences, with students studying more minutes per week having the greatest occupational aspirations (77.4 versus 74.4) (Nodes 41 and 42). For the group of students with the lowest self-concept of reading, the influencing variable would correspond to the ESCS Index, with the students with the highest status having the greatest occupational aspirations (75.4 versus 71.3) (Nodes 39, 40).

On analysing this branch, we can classify the group of female students who have not repeated grade into those with higher or lower occupational aspirations. In the first case (Node 42), these are students with a higher score in mastery goal orientation, a higher self-concept of reading and who spend more minutes per week studying sciences. These students reach a mean of occupational aspirations of 77.4 points. On the other hand, the group of female students who have lower occupational expectations in spite of not having repeated grade (Node 35) are distinguished by presenting an lower ESCS and bottom scores in attitudes towards school learning activities, reaching a mean score in occupational expectations of 65.4 points. This is more than 10 points lower than the subgroup of women who have not repeated grade and who have higher occupational expectations.

Branch C (Figure 4) represents the in-depth analysis of the influencing variables of male students who have repeated grade. In this case, the first distinguishing variable is self-concept of reading. We can see that male students with higher scores in self-concept of reading present higher occupational expectations (57.4 versus 52.1) (Nodes 11 and 12). Once again, the influencing variable for students with a lower self-concept of reading is this same variable. In this case, male students with a lower self-concept of reading have lower occupational aspirations (Node 23), compared with those with a higher self-concept of reading (Node 24). In the branch of male students with a higher self-concept of reading the next influencing variable (Nodes 25 and 26) is another variable associated with reading (joy/like of reading). We find that the male students who most enjoy reading have higher occupational aspirations (62.0) than those who do not enjoy reading as much, who also have lower occupational aspirations (55.7). In this group, the next influencing variable corresponds to attitudes towards learning activities at school, with students with the poorest attitudes (Node 45) having the lowest occupational aspirations.

Figure 4
figure 4

Branch C showing male students who have repeated grade. Dependent variable Students’ expected occupational status (SEI)

Full size image

It is noteworthy that the most important influencing variables in this branch are analogous to those observed in Figure 2 (nonrepeating students), after mastery goal orientation. This demonstrates in the case of male students, both for grade repeaters and non-repeaters, that self-concept of reading and enjoyment of reading have an important effect on the occupational expectations of 15-year-old students, and are positioned as explanatory variables among the first positions in the analysis. Within the group of male students who have repeated grade are those with a lower self-concept of reading (Node 23). At the opposite extreme, male students with higher occupational aspirations (Node 47) coincide with those who enjoy reading more and have a poorer body image.

Finally, we analyse the occupational aspirations of female students who have repeated grade (Figure 5). Here, we find that the variable that most affects the occupational aspirations of this group corresponds to attitude towards learning activities at school, with female students with the best attitudes having the greatest occupational aspirations (64.0), and those with lower attitudes only reaching a mean score in occupational aspirations of 55.8 points (Node 13).

Figure 5
figure 5

Branch D, female students who have repeated grade. Dependent variable Students’ expected occupational status (SEI)

Full size image

Within this subgroup of worse attitudes towards school learning activities, the first segmentation variable is ESCS, showing that girls with less access to resources have the lowest aspirations (52.4) (Node 27) compared to female students in this same group who have a greater access to resources whose aspirations are higher (59.3) (Node 28). The variable overall fear of failure affects the occupational aspirations of the students in this group with a lower ESCS, with girls shower a higher overall fear of failure coinciding with those with the highest aspirations (Nodes 49 and 50). On the other hand, in the group of female students with a higher ESCS (Node 28), the following segmentation variable is self-concept of reading, with girls with a higher value showing a higher mean value of occupational aspirations (Node 52) versus those with lower values of self-concept of reading (Node 51).

In the group of girls with the best attitudes towards school learning activities (Node 14), the first segmentation variable was the index highest parental occupational status, also related to family background. Although this variable is not included in Table 1 because its normalized importance is 9%, it becomes relevant in explaining the expected occupational status of female grade repeater students with a good attitude towards learning. More specifically, girls whose parents have a higher occupational status (higher than 36.1) also have the highest occupational aspirations (Node 30), compared to girls whose parents have an occupational status lower than this threshold, who reach a lower mean score of occupational aspirations (Node 29). Self-concept of reading becomes part of the segmentation of this subgroup (Nodes 53 and 54) having a positive effect on occupational aspirations. In the subgroup of female students whose parents have an occupational status higher than 36.1, the segmentation variable was mastery goal orientation, with a positive effect in favour of girls with a higher level in this type of goal orientation (Node 56).

At node 49, we can classify the group of female students who have repeated grade and have lower occupational aspirations (50.2): students with a lower level in attitudes towards school learning activities, with a lower ESCS, who present lower levels of overall fear of failure. At the other extreme (Node 56, mean 69.0), we find girls who have repeated grade and have higher occupational aspirations: girls with higher attitudes towards school learning activities, whose parents have a higher occupational status and who have a higher score in the mastery goal orientation dimension.

From the overall analysis of the results, some relevant data emerge. The group with the lowest occupational aspirations at 15 years old, coinciding with node 23 (Figure 4), corresponds to male students who have repeated grade and whose self-concept of reading is lower. At the other extreme, we have node 42 (Figure 3) which groups together the students with the highest occupational aspirations. These correspond to female students who have not repeated grade, with higher mastery goal orientation, a higher self-concept of reading and who spend more minutes per week studying sciences. These girls reach mean occupational aspirations of 77.4, much higher than the students with the lowest aspirations, who only attain a mean score of 47.2 (a difference of more than 30 points).

If we compare the situation of students who repeat and those who do not repeat grade, we can see that the mere fact of repeating grade, hampers their occupational aspirations at the early age of 15 years old. If the other influencing variables are favourable, male students who have repeated grade reach a maximum mean score in occupational aspirations of 68.5 points, and female students in this situation attain a similar mean score of 69.0 points.

Discussion

Grade repetition has been a controversial practice ever since it was first practiced. Research into this approach has found more drawbacks than benefits, more negative than positive effects (Carabaña, 2011; Rathmann et al., 2020). The large proportion of students repeating grade in Spain is one of the greatest concerns of the educative system, as the condition of being a repeater in itself is clearly linked to failure at school and poorer academic performance (López-Martín et al., 2018). In this study, this was the first explanatory variable analysed and was found to strongly influence students’ occupational aspirations. This educative measure, which should only be applied in exceptional circumstances, greatly harms the students’ occupational aspirations. On the basic premise of the Expectancy-Value (E-V) theory, the individuals choose to participate in activities and tasks that they consider to have value, and in which they expect to be successful (Wigfield & Eccles, 2000). The fact of repeating grade could be behind this loss of “hope” the students experience, which makes them regard their chances of success as greatly diminished and, in turn, significantly reduces their future occupational aspirations.

Regarding mastery goal orientation, it is worth noting that in the original proposal by Dweck (1986) this refers to interindividual differences in the response to different learning situations. Subjects with mastery goal orientation feel impelled to learn and master a set of tasks (Johnson et al., 2018). From the moment when they first approach the learning task, subjects with mastery goal orientation become more effective in learning contexts. It is important to point out that mastery goal orientation is relevant in students who have not repeated grade, and within this group differences can be found between students with lower and higher levels.

For gender, the differences found agree with previous studies which show that in middle and high school women have greater educational and occupational aspirations than men (Howard et al., 2011; Mau, 1995; Mau & Bikos, 2000). However, this increase in occupational aspirations contrasts with the reality observed in the labour market. Most women tend to go for posts traditionally considered as women’s jobs, possibly owing to socialization processes, a difficulty to integrate their professional life and family life, the perception of a possible discrimination and anticipation of negative experiences in jobs mainly carried out by men, among other factors (Eccles, 2011). If, as we observed in this study, the occupational aspirations of girls exceed those of boys at 15 years old, it would seem that the barrier to women entering certain professions appears in later stages, in which social roles could significantly condition their choices. Consequently, policies should be promoted that favour more flexibility for men and women to make it easier for them to achieve a work-life balance, which in turn could increase the range of employment available to both sexes and help them to perform all kinds of adult work and family roles (Eccles, 2011).

Regarding students’ self-efficacy in reading, this finding corroborates the results of other studies (Ahmavaara & Houston, 2007), which report that more confidence in an academic skill is associated with greater educational and occupational aspirations. Students’ confidence in their own intelligence and self-esteem were identified as influencing variables in this increase in aspirations. Enjoyment of reading was also a significant predictor in the models studied. Consequently, educative intervention in this area is one of the actions that can most contribute to improving students’ occupational aspirations.

In a similar vein to findings reported in previous works, ESCS has considerable importance in the model, with a normalized impact of 24% and a presence in all the branches of knowledge. Students with higher levels of ESCS present higher occupational aspirations, with the family’s financial and educative resources being particularly important in the occupational aspirations that the adolescents develop during the transition period from school to work (Noack et al., 2010).

When studying the effect of parents’ educational level, many researchers found this to be a powerful predictor of adolescents’ educational aspirations, with students with parents with higher educational levels tending to present higher academic and occupational aspirations (Andres et al., 1999; Cochran et al., 2011; Howard et al., 2011; Mau & Bikos, 2000). This study shows how the educational level of parents, considered individually, does not emerge as a determinant of students' educational aspirations in any of the resulting models. However, educational level, when is combined with parental occupation and home possessions in the ESCS index (OECD, 2017), emerge as an explanatory variable of students' educational aspirations, with a normalized significance of 24.2%. Together with the above, a variable related to the educational level of the parents, such as the index of highest parental occupational status, has explanatory importance in women who had repeated grade, leading to an improvement in expectations, although its normalized importance was lower than 10%.

The main results have shown that repeating grade, mastery goal orientation, enjoyment of reading and self-concept of reading are the most important variables at explaining students’ occupational aspirations. In accordance with the Social Cognitive Career Theory (SCCT) (Lent et al., 1994, 2000), the factors that influence students’ vocational behaviour are controlled by the interaction of cognitive processes with personal and environmental influences. Clearly, the fact that the first four influencing variables refer to factors that are susceptible to intervention throughout the schooling lifetime is promising and reflects the importance of these results from a perspective of educational and vocational guidance as lifelong learning (ELGPN, 2010). These results corroborate that, although the process of learning skills related to vocational maturity is especially important in the final years of secondary education, factors that influence decisions made by the students in relation to their professional careers transcend this educational stage and tend to extend throughout an individual’s lifetime. Therefore, vocational guidance should be considered as a continuous process throughout life (UNESCO et al., 2019), empowering individuals to accomplish their own educational and career paths.

The application of data mining techniques may be beneficial in the preliminary stages of a research to pre-select the predictors to be included in the definition of explanatory models. However, these models must be confirmed using other methodological approaches such as structural equation modelling (SEM). In the words of Asensio et al. (2018), they can metaphorically "separate the gold from the sand", by providing robust criteria for the pre-selection of predictors. Together with the above, this technique makes it possible to assess the individual effect of certain variables that are frequently embedded in indicators, and also to answer the following questions: is it always useful to use indicators? Is the predictive value of some items more powerful than that of the indices of which they are part?

In conclusion, it should be noted that the predictors introduced in the model explained 17% of the differences in students' academic aspirations. Although this is not a negligible percentage, it is lower than the values observed when analyzing the effect of other types of variables on students' future aspirations, such as their psychological characteristics (Verdugo & Sánchez-Sandoval, 2020) or their parents' expectations (Patton & Creed, 2007). Because this research is a secondary study using data from PISA 2018, it was not possible to consider these other factors that would have allowed us to reach a more comprehensive view of the phenomenon under study.