The effect of teacher agency support, students’ personal perseverance and work experience on student agency in secondary schools with Estonian and Russian instructional language

Abstract

This article explores how teacher agency support, students’ perseverance and work experience outside of school are related to student agency in the context of Estonia, where Estonian and Russian-speaking students are often educated separately in schools with either Estonian or Russian instructional language. The student-level data (n = 9060) were collected after piloting the survey instrument in 2022. To ascertain the factors and common causes that correlate with student agency, which is measured in a scale constructed by factor analysis, a structural equation modelling analysis was conducted. Besides the main positive effects of teacher agency support, students’ perseverance and work experience on student agency, other statistically positive covariates were found to be Estonian instructional language, male gender, higher school stage and students’ socio-economic background as measured in number of books at home. Among different types of work experience, working on holidays, in student work camps, doing other paid work (i.e., working in their own business), and voluntary unpaid work were significantly positively correlated with the agency. Therefore, students learned many specific and generic skills while working. Overall, our findings highlight the importance of teacher support and work experience in fostering student agency. The type of agency measured in this scale, which combines agentic engagement in the classroom with capacity for resistance to perceived injustice, is more articulated by male students in higher school stages.

Keywords:

• Student agency
• learning at work
• generic competences
• ethnic minority

REVIEWING EDITOR:

• Hengtao Tang, University of South Carolina, Columbia, South Carolina, USA

SUBJECTS:

• Theories of Learning
• Educational Research
• International & Comparative Education
• Middle School Education
• Secondary Education
• Sociology of Education
• Education Policy & Politics

1. Introduction

Many research papers and policy documents in the recent decades have been globally dedicated to the need to develop the so-called generic or transversal skills, sometimes also called key or future competences in students in order to adapt to the demands of life-long learning and rapidly changing social and work environments (Council Recommendations on Key Competences for Lifelong Learning, 2018; OECD, 2019). The realisation that academic knowledge is not enough to successfully cope with the challenges of life in the contemporary world, particularly in terms of employability, has led to a change in the concept of teaching and learning worldwide (Shek et al., 2017). Powerful supranational organisations such as OECD have been promoting the change by emphasising the need for application of knowledge and independent decision-making as well as having suitable attitudes and values (OECD, 2005). These suggestions have an empirically proven basis: for example, personal non-academic traits such as motivation, concentration, perseverance, socioemotional wellbeing, and good behaviour are proven to be beneficial for positive life outcomes such as educational achievement, socio-economic status (SES) and health while contributing to human capital formation (Heckman, 2007). They are discussed in the research literature interchangeably as non-cognitive or soft skills which refer to certain thoughts, feelings and patterns of behaviour (Heckman & Kautz, 2012; Gil-Hernández, 2021).

The list of soft skills or non-cognitive skills varies internationally, but recently, the notion of student agency appears in the OECD Learning Compass for 2030 (OECD, 2019) as a meta-level term which emphasises along foundational competences also students’ ‘ability and the will to positively influence their own lives and the world around them.’ However, the OECD definition of agency tends to overemphasise agency as an individual capacity. According to the socio-cultural view, student agency is a concept that enables to understand learning as a complex process in a particular socio-cultural environment (Crick et al., 2015). Ratner (2000, 413) postulates that agency is ‘the active element of culture’ and as such depends ‘upon cultural processes for its realisation, forms culture and has a cultural form.’ It is also a social phenomenon which depends on social relationships within a particular culture and a particular time frame (Emirbayer & Mische, 1998; Ratner, 2000).

We combine in our definition of agency the socio-cultural and the ecological view of student agency. Agency is therefore a capacity to set goals, make choices and deliberative decisions, solve problems and act depending on the affordances and constraints of the changing social environments and culture at any given time (Emirbayer & Mische, 1998; Jääskelä et al., 2020). This suggests that agency is a capacity that combines both cognitive and non-cognitive skills while being conditioned by the environment. Despite this limitation, agency is one of the most important life skills that is useful in various contexts: at school, work and in personal life.

Development of agency requires the presence of empowering, democratic relationships within society (Ratner, 2000) and, in this case, educational institutions. The latter may vary in their democracy orientation in different cultural contexts. For example, a comparative qualitative study of control and agency in student-teacher relationships in Finnish and Korean comprehensive schools revealed how individualist and collectivist cultures value different aspects of agency, seeing agency as individual or collective capacity. Yet, both systems can have implications on democratic schooling (Yoon & Rönnlund, 2021). Therefore, the culturally comparative aspect of agency is a promising vein of studies on student agency.

To our knowledge, cross-cultural large-scale studies on student agency are rare. There is the Agency of University Students’ Scale which has been validated in the Finnish and Spanish context (Jääskelä et al., 2023) but there are no validated scales available for comparing secondary education students’ agency in different cultural settings, including in majority and minority schools. We have filled this gap by developing a research instrument for comparing student agency in two different cultural settings. The case of the Estonian education system is interesting in this regard since it is a bilingual and multi-cultural environment which allows to compare student agency in schools with Estonian and Russian as a language of instruction (henceforth, EIL and RIL schools). Different experiences of student agency which we expect to find reflect different cultural traditions and attitudes of people in these schools. Since adolescent agency is also deemed to be an important predictor of social mobility and status attainment in adulthood (Schoon & Cook, 2020), it is useful to study any inequalities in this respect in order to design interventions that would prevent the negative spiral of social helplessness. This can be particularly important for minority students.

There is reason to be concerned in Estonia due to a large achievement gap found between the PISA test and national exam results of students in EIL and RIL schools which show that Russian speaking students are lagging behind by 42 points in all tested domains. This is an equivalent of one school year worth of knowledge (Täht et al., 2018). Furthermore, the lack of sufficient Estonian as state language skills impedes the further education and career of Russian speaking students which is documented by their lower representation in upper secondary and higher education as well as disadvantages on the labour market (Kunitsõn & Kalev, 2021; Täht et al., 2018). The language problem is currently being addressed by Estonian government’s decision from October 13^th 2022 which requires the transition of all RIL schools and kindergartens to Estonian instructional language by 2030 because facilitating the teaching of Estonian as a second language has not led to sufficient language skills of students in RIL schools (Eurydice, 2023).

Previous research has shown quite significant differences in how teachers interact with students in EIL and RIL schools. For example, in the PISA study of 2018, none of the teachers in RIL schools mentioned having discussions with students as a common method in their teaching practice while teachers in EIL schools made frequent use of this method (Täht et al., 2018). PISA 2018 study also showed that students in RIL schools perceive less support from their teachers, a lower level of belonging and more bullying by peers which result in lower level of well-being compared to students in EIL schools (Tire et al., 2019). These differences in the perceived school culture can be studied quite convincingly through the lens of student agency.

It is important to note that students do not develop agency just in school. In general education contexts, the discussion of student agency is usually confined to students’ capacities to act agentically and proactively in the classroom or school settings and teachers’ practices in supporting this type of agency (Mameli et al., 2021; Rajala et al., 2016; Reeve & Tseng, 2011). However, secondary school students’ learning and work experiences outside the formal education are rarely examined through the lens of agency development. Studies exist which explore the (professional) agency development of higher and vocational education students during internships and workplace learning (Brauer et al., 2021; Smith & Dalton, 2004; Soini et al., 2015) and of novice teachers during their first years of working (Eteläpelto et al., 2015). Based on the research on higher and vocational education students, it is reasonable to believe that secondary education students develop competences and attitudes through extracurricular work experiences which may also contribute to their agency capacity in school.

2. Context

The segregation of schools by Estonian or Russian mother tongue dates back to the Soviet occupation of Estonia after the WWII when large numbers of immigrants from Soviet Russia and other Soviet states were placed in Estonia as part of the Soviet armed forces or moved here to seek work as industrial workers. During this period (1940-1991) Russian was the official language of the Soviet Union although education could also be given in national languages such as Estonian. After Estonia regained its independence in 1991, the only official language became Estonian because the Republic of Estonia was not a new country appearing on the world map but it was reinstated as the legal follower of the Republic of Estonia which existed from 1918-1940 as a nation state. In 1941 ethnic Estonians comprised 90,8% of the whole population of Estonia (Maiste & Puur, 2018) and the forced dramatic change in the ethnic composition of Estonia during the Soviet occupation (by 1989 Estonians made up only 61% of the population) left many Estonians with the fear of becoming a minority in their own country. The decision to re-establish nation states in the Baltic states with just one state language in each was sometimes criticised internationally and viewed by Russian residents as linguistic discrimination (Włodarska-Frykowska, 2016). However, from the point of view of preserving the small national languages of Estonia, Latvia and Lithuania, it was an inevitable step, especially given the previous negative experience of Russification of public life, including schooling, in these countries which had put the indigenous languages in a very unequal position (Kemppainen, 2004).

A quick integration of Russian and Estonian students has proven difficult due to the large numbers of Russian speakers (25% in the country) who are concentrated in the industrial towns of the North-Eastern part of Estonia and in the capital city of Tallinn, often physically segregated from their Estonian peers in different neighbourhoods. In the school year 2021/2022 Estonia had 73 general education schools (15% of all schools) which had Russian or a combination of Estonian and Russian as instructional language (Eurydice, 2023).

The recent decision of the Estonian government on the gradual transition of all schools and kindergartens to Estonian instructional language by 2030 in order to ensure a good Estonian language competence and a common information room for everybody (Republic of Estonia Ministry of Education and Research, 2022) is actually not new. Attempts to create a common Estonian school system have failed over the last 30 years due to resistance of the Russian community (Kunitsõn & Kalev, 2021). Nevertheless, the Estonian Integration Monitoring (Estonian Integration Montoring (EIM) 2017) showed already in 2017 an increasing support of other ethnic groups for early start of teaching Estonian in kindergartens (79%) and for teaching more subjects in Estonian (78%). The willingness of minority parents to ensure a good Estonian language competence for their children is evident by the increasing enrolment numbers of Russian speaking students in EIL schools and language immersion groups (Estonian Integration Monitoring (EIM) 2020). Furthermore, the reform does not impede the ethnic minorities from teaching their own language and culture in schools if at least 15 students express a wish for that. Despite that, the reform will still be very challenging to implement.

From the historical point of view, it is important to note that most teachers who currently teach in RIL schools were educated in Russia instead of Estonia where teacher education is only provided in Estonian language. For this reason, the pedagogical practices and the school culture in RIL schools tend to follow a different ethos from the Estonian proclaimed constructivist and student-centred approach (National Curriculum for Basic Schools, 2014), resembling the Soviet education tradition (Carnoy et al., 2015; Zaichenko, 2021).

3. Theoretical framework

The theoretical framework is based on the concepts of student agency that informed the development of the agency scale used in this study. Further, the types of agency as related to learning at work are discussed. This section also has the purpose of introducing previous research that informed the hypotheses of this study.

3.1. The concept of student agency

The research on student agency has so far been usually qualitative, concentrating on aspects such as student voice and choice (Vaughn, 2021), student-centred approach to learning, self-determination, autonomy, motivation and engagement (Goodman & Eren, 2013; Moore, 2022). Quantitative research has been using many different theories, particularly emphasising individual aspects of agency: social-cognitive theory (Bandura, 1977) has focused on self-efficacy beliefs; basic psychological needs theory within the self-determination theory (Deci & Ryan, 2000) has stressed the importance of satisfying the three basic needs of humans which are competence, autonomy and relatedness. Self-regulation theory also highlights the importance of self-efficacy beliefs, self-regulatory skills and goal-setting in self-regulated learning as an attribute of agency (Zimmerman & Cleary, 2006; Zimmerman et al., 2015).

However, mixed-methods research on student agency is less developed and theories developed within qualitative research such as the ecological approach to agency are rarely applied to quantitative research. The unique theoretical contribution of this study is using the ecological approach to agency in a sequential mixed methods study as a meta-theory which informed the development of a quantitative research instrument for student agency. In this study, the ecological agency theory (Emirbayer & Mische, 1998) was first applied in a qualitative focus group study of EIL and RIL high school students in 2021 (results reported in Erss, 2023). This study gave input to developing a quantitative research instrument for studying student agency.

Although the instrument is based on some psychological concepts such as agentic engagement (Reeve & Shin, 2020), resistance to perceived injustice (Engestrom, 2014; Mameli et al., 2019a; Mameli et al., 2021; Rajala et al., 2016), teacher autonomy support (Reeve & Shin, 2020) and perseverance (Dweck, 2006; Vaughn, 2021), there is an important distinction in how the statements are stated, following the ecological approach. The main difference is that ecological approach does not see agency as action itself but as capacity for self-determined action which may differ in various contexts, times and situations. One common flaw in the psychologically founded and action-based research instruments is to phrase items in such a way that they suggest a habitual or repeated action as i.e. ‘I let my teacher know what I need and want.’ In a real school context, it can depend on the teacher, the day, the activity involved or student’s mood on that day, so it is difficult to generalise. According to the ecological approach, agency can be achieved by combining personal efforts with the ‘affordances and constraints of the environment’ (Priestley et al., 2015). It requires conscious deliberation of the alternatives when making decisions while simultaneously drawing from the past experience, current resources and anticipating the future. These temporal aspects of agency are known as iterational, practical-evaluative and projective dimensions of agency (Emirbayer & Mische, 1998).

In the classroom, student agency is often expressed in the agentic engagement that is the capacity of students to actively influence their learning environment to make it more motivationally conducive. ‘It is what students say and do to create a more motivationally supportive learning environment for themselves’ (Reeve & Shin, 2020, 151). For example, students can let their teacher know what they are interested in or what they need or want. Teachers can either support the expression of students’ personal interests or goals or not. Usually, supporting student agency and voice leads to more student engagement in learning (Reeve & Shin, 2020).

In a recent qualitative study on student agency which included 37 students from four EIL and four RIL schools, differences were found in how students perceived relationships with their teachers, their school culture and, depending on that, their capacity for agency (Erss, 2023). Expressing one’s opinion freely often led to conflicts between older teachers and students in RIL schools which students perceived as conflicts of generations indicating different values that each generation holds. For this reason several Russian speaking students preferred to keep a low profile and do as they were told. Students in EIL schools felt that their opinions and initiative were more welcomed and they described their school culture more positively (Erss, 2023). These differences have not been explored in a large-scale quantitative study except in an unpublished master thesis (Ots, 2023) which used the same student agency scale developed by the first author (Erss) of the current study. Ots found in her regression analysis of 8320 students in EIL schools and 826 students in RIL schools that Russian speaking students evaluated their agency and teachers’ support for their agency lower than students in EIL schools. She also ascertained that teacher agency support positively predicted student agency. Another dimension of agency in the developed scale is resistance to perceived injustice. Adolescents are particularly sensitive to questions of justice, therefore student agency can take the form of resistance to the ideas or actions of others, particularly to perceived injustice from teachers (Engestrom, 2014; Mameli et al., 2019a; Mameli et al., 2021; Rajala et al., 2016). Although students alone are in an unequal power relationship with teachers, they know that tables can be turned around if they join their forces with other students. Emirbayer and Mische (1998) have mentioned this phenomenon of agency emerging from unequal power relationships referring to Scott’s (1985) research about oppressed people or groups who exercise agency by subverting rules perceived as unjust. Mameli et al. (2019b) found in their study of Italian high school students that perceived interpersonal justice positively predicts responsibility and indirectly, through the student responsibility for their learning, also the career decision-making self-efficacy. This means that being treated fairly by teachers is an important motivation factor and can increase the level of responsibility that students take for their learning, thereby affecting their achievement, and their self-confidence in making career decisions.

On the other hand, perceived injustice often evokes strong negative emotions in students which can be expressed by active agentic actions or suppressed, leading to loss of agency (Mameli et al., 2021). When students perceive that their initiatives and transformative actions are contradicted and refuted, teachers may get to see the ‘stormy side’ of agency. The latter is correlated with lower achievement and higher intentions of dropout (Mameli et al., 2021). This may be the case if resistance is seen as only reactionary, or from teachers’ point of view, pointless disruptive behaviour. Most resistance theorists see resistance as a reaction to structural domination and oppression (De Certeau 1984; Giroux, 1983; Scott, 1985) which students often experience in school. However, resistance also has a transformative potential for student agency which can be used by skilful teachers to advance dialogic forms of pedagogy where students’ voices are heard and also negative emotions are allowed. Besides, some authors claim that what appears to be resistance on the first sight may actually reflect students’ pursuit of other goals meaningful for them (Lanas & Corbett, 2011). We think that the ability to resist injustice is an important dimension of agency also in the adult life of students and as citizens in a democratic society. In this research instrument, we do not focus on how justly students generally feel that their teachers treat them unlike in the instrument of Mameli et al. (2019b) but on students’ hypothetical capacity to resist injustice. This also matters in relationships with peers when there is a need to resist negative peer pressure.

Finally, students’ own resolve to pursue self-determined goals even while facing adversities emerged from the focus group interviews as an important assumption for agency. The resilience of continuing work even when experiencing drawbacks is according to Dweck’s mindset theory an important asset of achieving one’s goals and is dependent on one’s attitudes towards mistakes and failure. Persistence in pursuits or perseverance as an aspect of agency has also been mentioned by Vaughn (2021) and is related to Bandura’s notion of self-efficacy (1977) as a bundle of beliefs in one’s own capabilities to achieve the desired goals.

3.2. The role of work experience in achieving agency

It is widely acknowledged that work experience is a rich and powerful source of learning (Billet, 2011; Eraut, 2004; Zitter et al., 2020) and the spectrum of capacities, such as skills and competences, that people can acquire informally through work is very wide (Eraut, 2011, Evans et al., 2011). For example, Eraut (2004) presented a typology consisting of eight clusters of competences that generally are learned at work, namely task performance, awareness and understanding, personal development, teamwork, role performance, academic knowledge, decision making and problem solving and judgement.

In the context of postmodernity, when the lifelong learning has become an imperative (Evans et al., 2011), an increasing attention needs to be paid to the learning of schoolchildren that happens in the contexts outside of the formal learning, such as sports, arts, music and particularly while participating in the working life. More research is needed in regard to what students learn while working alongside the school, particularly in general education whether the work is paid, voluntary or other kind of occasional work. We suppose that the activity and experience-based feature of practice-based informal learning from work is (inter)linked with the issue of agency arising in situations where learning is not the main aim of activities.

Studies on the influence of early work experience on development of student agency and competences in the context of general education are scarce. There is some evidence from the USA that students who work during high school exercise their agency as it reflects their future goals by building human capital either through education or work. Students who are academically engaged, usually keep their workload moderate, using work mainly as a ‘side line’ or even means to save for college, while those with less interest in academic studies tend to work longer hours and take more responsibilities at work at the expense of school work (Mortimer, 2019). This may have implications on students’ agentic engagement at school.

3.3. Agency as a generic competence needed at work

Since our student survey also contained questions about what students thought they were learning in work situations, it is necessary to relate these competences to agency and expertise. There are various ways how agency has been conceptualised related to work and professionalism. One way is to look at specific generic competences as related to expertise and working life skills that are often in high demand by employers.

The integrative pedagogy (IP) model, elaborated by Virtanen and Tynjälä (2019) is useful in categorising the generic competences, which should be developed in every educational and work setting. The IP model combines constructivist teaching with the development of expertise claiming that expertise is based on four basic elements: 1) theoretical, conceptual knowledge; 2) practical, experiential knowledge; 3) self-regulative knowledge; and 4) sociocultural knowledge. The IP model suggests that all four elements of expertise or knowledge should be present in learning situations since they are integrated with each other in expert knowledge (Virtanen & Tynjälä, 2019, 882). From these four elements, Brauer et al. (2021) associated agency related competences in their study of Finnish higher education students’ opinions of generic skills needed in working life particularly with self-regulative competences (i.e. self-management, critical thinking, self-development, creativity and self-awareness) and sociocultural knowledge (i.e. respect for others, collaboration skills).

4. Purpose of the study, research questions and hypotheses

The purpose of this study is to ascertain how students’ personal non-cognitive traits such as perseverance, school-level factors such as perceived agency support by teachers and extracurricular activities such as work experience outside of school affect student agency in EIL and RIL schools while controlling for socio-demographic variables such as gender, school stage and socio-economic status (SES). Based on the problems indicated above and the purpose of the study, the research questions and the hypotheses for this study are:

1. How are students’ perceived agency support by teachers and students’ non-cognitive skills such as perseverance related to their agency at school?

   Based on previous research we hypothesise that (Hypothesis 1) student agency is positively influenced by perceived teacher support for student agency (Reeve & Shin, 2020), students’ own perseverance (Dweck, 2006; Erss, 2023; Heckman, 2007; Ots, 2023; Vaughn, 2021) and work experience (Brauer et al. 20121; Eraut, 2004; Mortimer, 2019; Virtanen & Tynjälä, 2019).

2. How is students’ previous and current work experience related to their agency?

   Since working is regarded as a way for young people to gain financial independence and is thus in line with the developmental goals of the youth (Friedrich Ebert Stiftung (FES), 2022), we think that students who have more different kinds of work experience (paid or unpaid) have higher agency scores than students with no or limited work experience (Hypothesis 1.1).

3. How is the school’s language of instruction related to students’ perceived agency support by teachers, their perseverance and agency, as well as their work experience?

   We expect to find differences between students in EIL and RIL school in their agency scores, perceived teacher support (see Erss, 2023; Ots, 2023) and work experience (Hypothesis 1.2). The Friedrich Ebert Stiftung study on the Baltic youth (Friedrich Ebert Stiftung (FES), 2022) ascertained that Russian speaking youth in Estonia have experienced difficulties in finding work if they are not fluent in Estonian. Therefore, it is likely that school-aged Russian speaking youth have less opportunities to get work experience.

4. How are the competences that students develop through work related to the development of agency?

   Since the data for answering this question comes from open answers, we did not set a hypothesis for this question.

   There is also some previous research evidence that gender (Mameli & Passini, 2017), grade level (Reeve & Tseng, 2011), and socio-economic background (Schoon & Cook 2020) have an influence on student agency although their effects may vary in different contexts. Therefore, these variables are treated as control variables in the following structural equation model.

5. Method

5.1. Participants and procedure

The main survey took place in January and February 2022 as part of a larger student survey containing different aspects of student’s experiences and opinions regarding their learning and school life. Invitations for participation in the study were sent to all schools in Estonia which have the grades 6-12. The participation was voluntary for schools; all schools were promised confidentiality and student data was collected anonymously. 55 EIL and 4 RIL schools decided to participate including 8510 students from EIL and 697 students from RIL schools from grades 6-12. The total number of participants in the study was 9307 students. However, some inadequate responses were excluded. The exclusion criteria were the same values (usually extremely low or high values or the medium value) for at least three different scales. We also excluded one bilingual school which had a different pedagogical approach and differed from all other schools, reducing the number of participating RIL schools to 3. Therefore, the final number of included responses to the analysis was 9060 (8380 in EIL schools and 680 in RIL schools). The ethics of the research was discussed with school principals who distributed informed consent letters to students and parents. While preparing and conducting the study, the ethical guidelines of Estonian Code of Conduct for Research Integrity (2017) were followed.

5.2. Measures

1. Student agency. Agency was measured by a five-item scale adapted from Reeve, 2013 and loosely inspired by Mameli et al. (2019b). The adaption and development process of the scale will be discussed in more detail in the authors’ next article. The scale items can be found in the Appendix.

2. Perceived agency support by teachers. Perceived support was measured by a three-item questionnaire adapted from Reeve & Shin, 2020.

3. Perseverance. Perseverance was measured by a three-item questionnaire created by Erss based of Vaughn (2021) and Dweck (2006). Seven-point Likert scale (1 - I don’t agree at all … 7 – I totally agree) was used in answering all three scales.

4. Students’ work experience. Two types of data were collected about work experience. First, six yes-no questions were developed about different types of work experience which secondary school students may have experienced. The questions included options such as:

   • I am currently having a paid job alongside school.

   • I am currently not working but I have had a paid job before.

   • I have worked during school holidays for pay.

   • I have participated in student work camps and received a pay for that.

   • I have done other paid work (i.e. worked in my own business).

   • I have done voluntary work during my studies for no pay.

These six options were aggregated into one variable ‘work experience’ with values from 0-6 based on the exposure to different kinds of work experience.

Additionally, students were asked three open-ended questions about their work experience: what kind of work they have done in parallel to studies at school; which work tasks they have completed; what skills they have acquired while working. 5) Language of instruction. Schools were divided based on their reported language of instruction in Estonian Education Information System EHIS into EIL and RIL schools. Although at high school level (grades 10-12) all RIL gymnasia have to offer at least 60% of the subjects in Estonian, these schools still count as RIL schools which teach part of the curriculum in Russian. Two schools which reported being bilingual were left out of the sample.

5.

Control variables. Students’ gender, school stage and socioeconomic status (as measured in the number of books at home and having one’s own room) were used as control variables in the analyses. Two dummy variables were used as school stage indicators - grades 8-9 and grades 10-12, while grades 6-7 was the reference category. Number of books at home in three categories - 1 = 0-25, 2 = 26-200 and 3 = more than 200 - were used as a socioeconomic status (SES) indicator. Number of books at home and having one’s own room are widely used as SES indicators in large-scale international studies such as PISA studies.

5.3. Data analysis

First, the measurement invariance of the three questionnaires answered in Estonian and Russian language was established. All eleven items loaded into three subscales as expected. Three correlations between the residual variances were allowed to improve the model fit. Measurement invariance was tested by χ2 difference between the less-constrained model and the more-constrained model using Satorra–Bentler’s calculation (Satorra & Bentler, 2010). We managed to prove a similar factor structure (configural invariance) and factor loadings (metric invariance). Establishing scalar invariance by constraining item intercepts by two groups of students first failed. We next tested the partial invariance with letting the intercept of one item in perceived agency support scale (‘My teachers encourage me to ask questions’) as well as one residual correlation to be estimated freely between groups.

Standardised factor loadings ranged between .44 and .72 for students’ agency scale, between .65 and .69 for perceived agency support, and between .54 and .79 for perseverance scale. All three scales were used as latent variables estimated by single items in the structural equation modelling (SEM). In the SEM model the students’ agency was predicted by perceived agency support, students’ perseverance and their work experience. Language of instruction was predicting all four above mentioned variables: student agency, perceived agency support by teachers, perseverance and work experience. Three control variables, gender, school stage and SES were also entered into the model to predict these variables. The statistical programme MPlus 8.8 was used for both estimating the measurement invariance and testing the hypothesised structural model (Muthen & Muthen, 1998–2022).

In order to analyse the work experience of students, the aggregated variable of the existing work experience measured on the scale of 0-6 was included in a SEM analysis to predict agency. Additionally, crosstabs and a χ2 test were used to analyse the differences between the work experience of students in EIL and RIL schools regarding different types of work. Then, a content analysis was performed where categories were first formed qualitatively regarding the competences and skills that students mentioned in their open answers. The analysis was a combination of inductive and deductive procedure where some categories emerged from data with no underlying theory and some followed loosely the competences of the IP model (Virtanen & Tynjälä, 2019). Next, these categories were quantified to show the frequencies in a pivot table.

6. Results

The result section is organised by model characteristics, research questions and hypotheses.

6.1. Model fit

In order to test our hypotheses we estimated a SEM model presented in Figure 1. To evaluate the model fit characteristics Root Mean Square Error of Approximation (RMSEA), Tucker-Lewis Index (TLI), Comparative Fit Index (CFI) and Standardised Root Mean Square Residual (SRMR) were used. Hu & Bentler (1999) have suggested that values below 0.06 for RMSEA and 0.08 for SRMR can be considered as indicating a good fit between the hypothesised model and the observed data. For TLI and CFI values approaching 1.0 suggest a better fit and values over .90 an adequate fit (Keith, 2019). The model fit the data well (CFI = .92, TLI = .89, RMSEA = .05, SRMR = .03).

Figure 1. The SEM model of student agency.

6.2. Control variables as common causes

Next to the language of instruction, all dependent variables were controlled for students’ gender, school stage and SES as common causes for agency. Boys perceived somewhat more agency support from teachers (β = -.04, p < .001), had higher perseverance (β = -.04, p = .005) and more agency (β = -.04, p < .001). They also had more work experience compared to girls (β = -.05, p < .001). Students in Grades 8-9 had higher perseverance (β = .04, p = .005) and more agency (β = .04, p = .001), and also more work experience (β = .15, p < .001) compared to students in Grades 6-7. Students in Grades 10-12 perceived more agency support from teachers (β = .18, p < .001), had higher perseverance (β = .17, p < .001) and more agency (β = .03, p = .007), and also a lot more work experience (β = .35, p < .001) compared to students in Grades 6-7. Students’ higher SES was related to perceiving more agency support from teachers (β = .08, p < .001), having higher perseverance (β = .10, p < .001) and more agency (β = .03, p = .006). Students’ SES was not related to their work experience (β = -.01, p = .294).

For better readability, the figure only contains regression coefficients explaining agency. The coefficients explaining perceived support, perseverance and work experience are mentioned in the text.

6.3. How are students’ perceived agency support by teachers and students’ non-cognitive skills such as perseverance related to their agency at school?

In answer to the first research question, our model showed that the more agency support students perceive the higher is their own agency (β = .16, p < .001). Further, the students’ perseverance substantially explained their differences in agency - the more perseverance they have, the more agentic they are (β = .49, p < .001).

6.4. How is students’ previous and current work experience related to their agency?

SEM model confirmed the hypothesis 1.1. that students’ with more work experience have higher agency (β = .15, p < .001). In total, our proposed model explained 8% of variance in perceived agency support from teachers, 4% variance in perseverance, 11% variance in students’ work experience and 41% of variance in their agency. As hypothesised, the SEM analysis confirmed that teacher agency support, students’ perseverance and work experience have a positive effect on student agency (Hypothesis 1).

To examine the effect of different kinds of work experience on student agency in more detail, a linear regression analysis was conducted using the enter method. Using paid work experience at the time of taking the test as a reference category, the following factors were entered in the model: paid work experience in the past, paid work on school holidays, paid work in student work camps, other paid work (e.g. in one’s own business) and voluntary unpaid work. The results show that only paid work on school holidays (β=.126, t = 11.068, p <,001), other paid work (β=.098, t = 9,169, p<.001) and voluntary unpaid work (β=.059, t = 5,625, p<.001) are significant predictors of agency at the α=.01 level and student work camp (β=.026, t = 2,499, p =.012) at α=.05 level. Having a current paid work during the studies was not a significant predictor of agency.

6.5. How is the school’s language of instruction related to students’ perceived agency support by teachers, their perseverance and agency, as well as their work experience?

To answer our third research question, the school’s language of instruction was set to explain perceived agency support, students’ perseverance and agency, and their work experience. We expected to find differences between students in EIL and RIL schools in their agency scores, perceived teacher support, and work experience (Hypothesis 1.2). We found small to moderate effects of language of instruction on three of our four dependent variables. Students from EIL schools perceived more agency support from teachers (β = -.19, p < .001), had higher agency (β = -.03, p = .008), and had somewhat more work experience (β = -.08, p < .001) compared to students from RIL schools. Students’ perseverance did not differ between schools of different language of instruction (β = -.03, p = .068).

The independent T-Test indicated that the difference in work experience between EIL and RIL schools was significant for current work, t(9058)= -3,04, p=.001; work on holidays, t(9058)=6,39, p= <.001; student work camps, t(9058)=7,39, p<.001; other paid work, t(9058)=4,13, p<.001 and volunteering, t(9058) =8,66, p<.001. While Russian speaking youth tended to have more frequently current paid jobs, Estonian youth had more work experience in the past, during holidays, in work camps and volunteering.

The frequencies of different types of work experience for students in EIL and RIL schools are in Table 1.

Table 1. The frequency of students’ work experience by schools’ instructional language.

Group Statistics
Instructional language		N	Mean Percentage
Work now	EIL schools	8380	7%
	RIL schools	680	10%
Work before	EIL schools	8380	14%
	RIL schools	680	12%
Work on holidays	EIL schools	8380	43%
	RIL schools	680	30%
Work in work camps	EIL schools	8380	17%
	RIL schools	680	6%
Other paid work	EIL schools	8380	21%
	RIL schools	680	15%
Volunteering	EIL schools	8380	28%
	RIL schools	680	13%

6.6. How are the competences that students develop through work related to the development of agency?

From 9060 students involved in the survey 3593 (39,6%) answered the question, ‘What have you gained from working, what skills have you acquired through work?’

The main categories as a result of qualitative categorisation of open answers of students are visible in the following Table 2. Many responses contained several categories.

Table 2. Students’ work-related skills and competences.

Categories	%	Count
Social skills (communication skills, team work, collaboration)	32%	1149
Specific skills (practical skills related to a specific field or subject, incl. sports, financial literacy)	26%	940
Self-regulation (time management, resilience, patience, handling stress, goal orientation, persistence, concentration, metacognition, incl. thinking about own cognition which helps to regulate behaviour	19%	673
Generic work skills, incl. discipline, work ethics, incl. following rules/requirements	12%	445
Self-awareness and understanding/realisations and other life experience	9%	320
Self-confidence (courage, independence)	7%	254
Responsibility and dutifulness	7%	243
Did not gain anything + angry responses	6%	201
Other responses	3%	102

Students perceived that they have learned at work mostly social and field- or subject-specific/academic skills and competences. This is understandable because they mainly completed tasks related to various kinds of services, such as client services, jobs in restaurants, cleaning, waitressing, working as assistants in shops etc. Students also performed other jobs, requiring more specific skills, such as construction work, bookkeeping, helping in advertising, working with Photoshop to edit pictures, working as a security guard etc. Most jobs were of temporary nature.

Generally, all kinds of skills and competences they learned through working are highly appreciated general skills and competences at the labour market.

For example, when asking what students learned they provided short answers,

I learned…

…communication with people, communication with clients, to stay cool while communicating with clients

…to concentrate

…much self-confidence

… a lot of courage, motivation

…dutifulness, correctness, discipline, work ethics

…time management and valuing moments of free time

…how difficult it is to earn money and how much work has to be done to earn money

According to the concept of ecological agency, agency is not a personal trait but can be viewed rather as bounded to a ‘temporal-relational context’ (Emirbayer & Mische, 1998, 1004). This means that the competences acquired or developed at work are highly dependent on the specific context and affordances of work students have performed. However, from the perspective of achieving agency in the work environment, generally, the development of self-regulation, self-confidence and self-awareness can be considered as highly important. Also, social skills are important for agency since effective communication, asking for help and receiving it, also collaborating with others are often necessary competences when pursuing one’s goals. From many answers it was clear that students do develop agency related competences through work, although these competences are not always transferred to the school context.

7. Discussion

The main results of the study point at the complexity of the student agency phenomenon which can be defined in different ways and predicted only in combination of various context-related, demographic, socio-economic and personal factors. In this case, students’ capacity for agentic behaviour in Estonia, as measured by agentic engagement and resistance to perceived injustice, was positively affected, as we predicted, by perceived agency support by teachers, by students’ own perseverance and various work experience. From the control variables Estonian instructional language, being male, studying at a higher school stage and having more than 200 books at home as a proxy for SES also proved to be positively correlated with agency.

The fact that school’s instructional language predicted agency points at differences in the pedagogy and teacher-student relationships in EIL and RIL schools. These differences have been discussed in previous research (Tire et al. 2019; Erss, 2023). One of the most important factors predicting student agency besides gender, school stage and instructional language is therefore teacher agency support which could explain the differences between EIL and RIL schools since Russian students rated the support of their teachers to student agency lower than Estonian students. Research on teachers in RIL schools indicates that the prevailing beliefs among older generation teachers often reinforce the obedience culture known from the Soviet time where teachers ‘demand discipline to the point where you can hear a pin drop’ (Zaichenko, 2023, 12, 14) and are less willing to allow students to express their own opinions (Erss, 2023). In the light of the planned language reform of RIL schools which was supposed to alleviate the inequalities between the EIL and RIL schools and bring everybody in the same information and value zone (Republic of Estonia Ministry of Education and Research, 2022), these findings cast some doubt on the benefits of the reform. Changing the instructional language into Estonian in RIL schools does not automatically entail a change of values. It seems that there are much deeper layers of school culture which need to be addressed concerning school leaders’ and teachers’ values regarding democratic practices in school and an empowering attitude of teachers towards students. This can be decisive if the aim is to educate confident, well-coping and agentic citizens for a democratic society.

Gender differences in agentic engagement have been previously found also in the Italian context where boys’ scores were higher than girls’ in this regard (Mameli & Passini, 2017) which may well be connected with cultural gender norms in both Italian and Estonian society. Boys are usually more expected to behave proactively than girls. Nevertheless, girl may develop already in kindergarten age different attributes of agency such as better self-regulation skills (Størksen et al., 2014).

Although work experience was identified as one of the factors predicting student agency, the quantitative study did not shed any light on the causality. It is possible that more agentic students were more likely to have work experience. Yet, from the open answers, we could still detect some dimensions of agency which students felt that they acquired as competencies at work. Therefore, it is more likely to assume that agency related competences can be developed through work experience.

It was interesting to learn that not all forms of work experience predicted student agency. The fact that current or previous paid work parallel to school was not a significant predictor of student agency is puzzling. One explanation could be that students who are more agentically engaged in school are also more academically interested students. Choosing to work during the studies may indicate a lack of academic interests (Mortimer, 2019). It can also be assumed that for some students it might be an economic necessity, not a choice to work and study in parallel.

Other forms of work such as working in school holidays, doing other paid work (like working in one’s own business or doing household chores), or doing unpaid voluntary work strongly predicted agency. Working in student work camps was also significant but a slightly weaker predictor of agency. It is likely that these forms of work have additional benefits for students beyond earning money, which was one of the most expressed reasons for working parallel to school. For example, work camps are also important places for the youth to socialise and spend their recess time together. Volunteering usually requires a strong motivation to contribute to a personally important cause, which can easily be transformed into agentic engagement in the classroom. Previous research has found positive effects of volunteering on civic engagement, psychological well-being, social skills and educational attainment (Kim & Morgül, 2017). Moreover, volunteering has been found to be a way to increase social cohesion and integration in cities (The State of the World’s Volunteerism Report, 2018).

The comparison of students in EIL and RIL school samples indicates a gap in their work experience. Estonian speaking students are more likely to have acquired some work experience during their school time than Russian speaking students. This probably points to the different opportunities (see Friedrich Ebert Stiftung (FES) 2022) and, perhaps, also divergent cultural expectations that students in distinct ethnic communities have. It is probably easier for Estonian speaking students to find a summer job, join a student work camp, work in their own or their parents’ business and find meaningful ways of contributing as volunteers to different causes. The differences between Estonian and Russian speaking youth may also relate to individual choices which ’are based on what an individual considers to be options and follow the preferences shaped by their social position and context’ (Kogler et al., 2023). The obstacles for Russian speaking students for finding paid work or to volunteer may be their language barrier and limited social networks. This is concerning because all these forms of work that Estonians were more actively experiencing are also predictors of agency.

Qualitative analysis shows that the dimensions of agency measured in school environment did not fully coincide with competences students perceive they acquired at work. This refers probably to the concept of ecological agency that presumes that agency can manifest itself differently in various (learning) environments (Priestley et al., 2015) due to different affordances and constraints.

The skills and competences acquired at work reflect in some extent the general competences, brought out in the IP model (Virtanen & Tynjälä, 2019) and in the typology by Eraut (2004). The most frequently mentioned competences, which refer to self-regulation, self-confidence and self-awareness as well as social skills are indirectly related to the forms of agency as measured in this research. Agentic engagement requires students to contribute to the flow of instruction actively and constructively in order to make it more motivationally conducive (Reeve & Shin, 2020). This requires all of the above-mentioned skills: to distinguish simply disruptive comments from statements that give teachers important input to develop learning situations that are more engaging. It takes courage and self-awareness to speak up, and social skills to keep the conversation civil and motivating for others as well.

In conclusion, the relationship between various forms of work experience that students acquire during their studies in general education schools and the achievement of student agency is still largely unexplored in different contexts. There is a need for further mixed methods studies, including cross-cultural research and studies involving minority students.

Our findings suggest that teacher agency support is the most important factor explaining student agency in school. However, work experience and other extracurricular learning experiences are a good way of contributing to student agency development by acquisition of new skill sets, self-confidence and, essentially, self-worth. So far, it has been mainly left to families and their social capital to ensure access to such work experiences for underage students. A policy suggestion for local administrations and schools would be to collect and spread information on possible job or volunteering opportunities suitable for students so that more students could benefit from these learning experiences. Furthermore, teachers could help students reflect on the experiences and competences they have acquired through work to raise their awareness of what they have learned and how it contributes to their agency development.

7.1. Limitations

As a limitation of the current research the small number of RIL schools who participated in the study must be noted which is why conclusions made about Russian speaking students must be treated as not representing all Russian speaking students in Estonia. Moreover, the research instrument developed to measure student agency did not directly measure the competences that students developed through work. Another limitation may be the one-sided and episodic nature of work that students experienced during their studies at general education, which is usually not supported by mentoring and reflection to develop agency related skills. The problem of vast differences in the quality of mentoring due to lack of financial incentives for trainers was discussed already in 2010 and is characteristic also to VET where workplace learning is compulsory (Ümarik et al., 2010). Further, the discussion on students’ work experience in this research is limited by the nature of data collection. The large-scale survey which contained many different topics did not allow to collect very rich qualitative data. In order to understand the connections between student agency and work experience in depth, qualitative interviews are recommended.

Acknowledgments

The authors want to thank Leena Sirp for her help in quantifying the qualitative categories of work experience.

Disclosure statement

The authors have no conflict of interest to declare.

Additional information

Funding

This work was supported by the Estonian Research Council under Grant PUTJD1031 and European Union’s Horizon Europe research and innovation programme under Grant 101079237.

Notes on contributors

Maria Erss

Maria Erss is a researcher and associate professor of curriculum theory at Tallinn University, Estonia. The current research article is related to her postdoctoral research project ‘Understanding the concept of student agency among Estonian and Russian speaking students: What is the experience and attitude of students towards agency in upper secondary schools in Estonia?’

Krista Loogma

Krista Loogma works at Tallinn University, Estonia as distinguished professor of vocational and professional education. Maria Erss and Krista Loogma are currently collaborating on a Twinning project on ‘Enhancing Research on the Integration of Formal Educational Programmes and Workplace Learning’ (FEWL) by conceptualising formal, non-formal and informal learning as vital factors for students’ agency development in constructing their sustainable individual learning paths.

This article combines both insights from student agency and integration of formal education and workplace learning research.

Anna-Liisa Jõgi

Anna-Liisa Jõgi is a researcher and associate professor at Tallinn University. She has a background in educational sciences and teaches quantitative research methods on masters’ and PhD level.

Appendix

 

The secondary school student agency scale

Capacity for agentic behaviour

1. I am often willing to make suggestions on how to improve the organisation of instruction in the lessons.

2. If I see at school that something is not right, I am capable of intervening to improve the situation.

3. I am always ready to protest if a teacher treats me or my classmates unjustly.

4. I am capable of refusing to follow orders from a teacher if it contradicts what I believe is right or wrong.

5. I can usually resist the peer pressure to do something (to do something that I dislike) even if it makes me unpopular in their eyes.

Teacher agency support

1. I feel that my teachers offer me enough choices and decision-making power.

2. My teachers also listen to different opinions which do not coincide with theirs.

3. My teachers encourage me to ask questions.

Perseverance

1. I can usually continue my work in pursuit of my goal even if I encounter obstacles and backlashes.

2. I can continue pursuing my future goals even if my family disagrees with them.

3. I am able to learn from my mistakes and usually avoid repeating them.