Data from the Swiss TREE Panel Study (Transitions from Education to Employment)

Research objectives and theoretical background

TREE’s objective is to observe long-term trajectories and transitions within and between the (post-compulsory) education and the labour market systems in Switzerland, with particular focus on the individual development of young people during these phases.

Analysing long-term trajectories of multiple cohorts calls for a theoretical framing that best addresses the specificities of different biographical phases as well as contextual changes at the macro and meso levels that may affect the observed cohorts. Furthermore, the multi-disciplinary character of TREE requires a framework that relates to the relevant discourses in the involved research fields. Therefore, it must be open to further development and adaptation of the survey programme – be it due to biographical and life-cycle developments, social changes between cohorts, or methodological and theoretical innovations in the respective research fields.

As a general theoretical framework, the life course research paradigm as initially proposed by Elder (1975) addresses these needs. In recent years, the paradigm has gained importance in a number of disciplines of the social sciences, thus lending itself to an overarching frame for inter-disciplinary use of the TREE data (Bernardi et al., 2019). It draws on both sociological (Elder, 1975) and psychological (Baltes et al., 2007) research traditions that are considered to be relevant for the transitions under scrutiny.¹ As Diewald & Mayer (2009) emphazise both traditions assume that the embeddedness of individual developmental processes in historical and local contexts is crucial (e.g., age effects, big events such as COVID or societal trends as digitisation). In addition, there is widespread agreement that an individual’s life course and development are embedded in social contexts (significant others, institutions), and that transitions are developmental tasks (Havighurst, 1972 [1948]) in the face of which the individual must make decisions. Understanding individual agency therefore has to account for different life contexts in which adolescents develop and with which adolescents and their families must deal with (Bernardi et al., 2019) as well as for the timing of different events and stages within the life course. Furthermore, individual action is seen as guided by bounded rationality (Bernardi et al., 2019, Esser, 2004; Stocké et al., 2011).

While life course sociology sees the individual life course as strongly influenced by opportunity structures, institutional settings, and decision-making situations, life course psychology focuses primarily on intra-individual development and the individual’s adaptation to external conditions (Bernardi et al., 2019). With regard to the concepts and scales adopted in the TREE surveys, we refer to sociological theories of status reproduction on the one hand (Bourdieu, 1977; Bowles & Gintis, 2002; for a critical overview of the current state of research, see also Draelants & Ballatore; Winkler-Wagner, 2010) and rational choice approaches (Baumert & Schümer, 2002; Blossfeld & Shavit, 1993; Boudon, 1974; Breen & Goldthorpe, 1997) on the other. These theories model the mechanisms by which social origin influences educational decisions at each point of transition in the education system.

In the fields of psychology and educational sciences, TREE draws on developmental theories that deal with educational and occupational socialisation (Heinz, 1984; Ulich, 1991) as well as with the ways in which youths cope with, in Bronfenbrenner’s (1981) terms, “ecological transitions”. The underlying assumption is that individuals need certain skills to successfully manage transitions, such as the ability to set and pursue goals. Motivational aspects (Deci & Ryan, 1993; Wild, 2001), self-efficacy as well as problem-solving and decision-making skills are held to be the central characteristics that favour successful educational and acquisition pathways (Schoon & Silbereisen, 2009; Silbereisen et al., 2006). Eccles (2006) and Gottfredson (1981) underline that not only skills in their own right, but also their intra-individual representation (Greve, 2000) are crucial for educational choices and success. Other concepts used to explain educational pathways derive from theories of institutional resources and constraints (which may lead to individual stress; Semmer, 1997), wellbeing (e.g., self-esteem, depression, attitude towards life, personal values; see Hascher, 2004) and critical life events (e.g., relocation, divorce or separation of parents, illness, death of a relative; see Filipp, 1995).

When it comes to analysing labour market entry, TREE relies on concepts and instruments used by classical human capital theory (Becker, 1964), signal(ling) and social capital theories (Coleman, 1988; Spence, 1973), theories of discrimination (Arrow, 1994; Becker, 1957/1971) and school-to-work transition models based on labour market economics (OECD, 2000; Ryan, 2001; Ryan, 2004).

The Swiss education System as a specific context of school-to-work transitions

The Swiss education system is characterized by a federal, small-knit structure and a pronounced horizontal and vertical stratification from lower secondary level onwards. Furthermore, there are historically and culturally determined differences between cantons and language regions which reflect varying institutional structures of opportunity (Glauser & Becker, 2016; for a detailed institutional self-description, see www.edk.ch).

In most cantons, after grade 6 of primary school (at around 12 years of age), students are streamed to two up to four different tracks of lower secondary school, which differ according to their level of academic requirements. The extent and form of tracking varies greatly from canton to canton. Approximately 30% of all students are assigned to tracks officially termed to fulfil “basic (academic) requirements”, with inter-cantonal variation ranging from 10 to 40 percent. Even though educational policy postulates permeability between tracks, initial allocation to a given track remains de facto largely irreversible (Bayard, 2018; Oesch, 2017). The tracking not only influences further skills development (Angelone & Ramseier, 2012; Baumert et al., 2006; Tomasik et al., 2018) but also further education trajectories (Gomensoro & Meyer, 2021; Scharenberg et al., 2016) and has proven to be a major factor reinforcing social inequality (Terrin & Triventi, 2022). Compulsory schooling ends after three years of schooling at lower secondary level (i.e., after nine school years).²

Upper secondary education in Switzerland is dominated by firm-based vocational education and training (VET): the large majority of Swiss compulsory school leavers (65 to 70%) attends VET programmes, which are provided in over 200 training professions, with varying duration (two and four years) and academic requirements. Most programmes are provided in a “dual” form, i.e., learners attend a (usually larger) part of their training “on the job” in a training company, while the remaining time is spent in (vocational) school. VET trainees in this type of programme sign an apprenticeship contract with the training firm and receive a (modest) salary. Vocational school attendance varies greatly by training profession, number of lessons and type of curriculum (for more detail, see e.g., Gronning & Kriesi, 2022). Upon completion of their training, they obtain a VET diploma which grants access to specific occupational segments of the labour market and/or to further training at post-secondary and tertiary levels. In particular, VET graduates may acquire a professional baccalaureate, which entitles them to access Universities of applied sciences.

Only about one third of all students enrolled in upper secondary level education attend academically oriented general education programmes.

Completion rates at tertiary level education is relatively low by international standards, particularly in the VET sector (for a general outline regarding VET and higher education, see Nikolai & Ebner, 2011).

2.1 Study design

TREE (Transitions from Education to Employment) is a multi-disciplinary longitudinal large-scale survey on educational and occupational pathways in Switzerland for the use within the scientific community at large. The source of the data are two panel surveys of school leavers that both start at the end of compulsory school, at respondents’ age of approximately 15 to 16 years (see Figure 1), drawing on national large-scale assessments as baseline surveys (see Section 2.4.3 for detail); for further documentation, see the two cohorts’ detailed study designs (Gomensoro & Meyer, 2017; Hupka-Brunner et al., 2023; TREE, 2016d).

Figure 1

Since the launch of the first TREE cohort in 2000, the macro context in which educational pathways and transitions are embedded has changed. This includes major reforms to the education system that are likely to affect the educational pathways of young people. Moreover, global developments in the domains of economy, communications technology, environment, science and demographics pose major challenges to education systems worldwide (Blossfeld et al., 2007). The Swiss education system is no exception and is confronted with challenges such as the ongoing economisation and pluralisation of society (changing family forms, changing fertility birth patterns, increased international and national migration) and the increasing computerisation and virtualisation of everyday life (Akademien der Wissenschaften Schweiz, 2009; Hupka-Brunner et al., 2022).

Therefore, a second cohort has been launched in 2016 to examine the extent to which these changes affect the individual trajectories of youth. TREE2 adopts essential characteristics of the first cohort’s (TREE1) design, which allows for inter-cohort comparison of how school-to-work transitions have changed over time (Gomensoro & Meyer, 2017; TREE, 2016d). The two cohorts are comparable insofar as

• – they both are followed up on their pathways through upper secondary and tertiary education into gainful employment and young to middle adulthood;

• – they both are followed up at yearly intervals up to age 22–23, and at looser intervals (2–5 years) later on, thus providing a dense and comprehensive observation of all relevant types of activities during school-to work transitions;

• – they both draw on a baseline survey including elaborate assessments of cognitive skills acquired by the end of lower secondary education;

• – the baseline surveys of both cohorts provide comprehensive measurements of students’ starting conditions held to be relevant for their later educational and labour market pathways;

• – they both draw on a large, nationally and regionally representative sample of students in their last year of compulsory education (i.e., at the end of lower secondary education).

• – Moreover, response for both baseline surveys is extraordinarily high (PISA 2000: 95%; AES 2016: 93%; see BFS & EDK, 2002; Verner & Helbling, 2019 for details), which substantially facilitates measures to correct for non-response bias due to panel attrition.

The two cohorts differ primarily in the granularity of the observed trajectories, which is finer for the second cohort in the early stages of the panel, as well as a number of topical foci that seemed more relevant at the time of the 2^nd cohort’s launch (e.g., digitisation; for an overview of the two surveys’ topical foci and their comparability, see also Table 5).

2.2 Time of data collection

The first TREE cohort (TREE1) was launched in 2000 and draws on a large national (compulsory) school leavers’ sample (n = 6343) tested and surveyed on the occasion of Switzerland’s then first-time participation in PISA. Since then, the sample has been followed up by means of ten panel waves, the most recent one conducted in 2019/20. Data were collected during the last three to four months of the Swiss school year (March through June/July) to ensure comparability of educational situations across panel waves. As of panel wave 8, data collection started somewhat later in some waves and was extended to a longer period (i.e., March/April to September/October), which is justifiable as pathways at that stage are substantially less closely structured by the rhythm of the school year. Further panel waves are planned at five-years intervals. To date, TREE1 respondents have reached an average age of approximately 40 and been surveyed for a period of over 20 years, spanning from early adolescence up to early middle-age. The study thus has gradually grown into a full-blown life course survey.

The second TREE panel (TREE2) covers a comparable population of school leavers who completed compulsory education in 2016 (n = 8429). Since then, the second cohort has been surveyed six times at yearly intervals. Data collection was conducted in the same period of year as with the first cohort. Further panel waves are planned at looser intervals (two to five years).

2.3 Location of data collection

With respect to the first TREE cohort (TREE1) the baseline survey (PISA 2000) was administered in a proctored classroom setting within the schools drawn for the test sample. As to the subsequent TREE1 panel waves, there were no instructions or restrictions as to where or in which setting respondents were to complete the survey. It may be assumed that most respondents completed the questionnaires (both CATI and self-administered) at their homes. In later panel waves (and in step with the increasing use of cellular telephones and smartphones in the past decade), we have to assume a wide variety of locations where the survey was completed (at home, in school, at work, while commuting, etc.).

With regard to the second TREE cohort (TREE2), the administration of the baseline survey also took place in a proctored classroom setting within the schools drawn for the test sample. As smartphone use had become largely commonplace by the time of TREE2’s launch (2016), we have to assume – much like in the later panel waves of the first cohort — a wide variety of locations where the survey was completed (at home, in school, at work, while commuting, etc.).

2.4 Sampling, sample and data collection

2.5 Survey instruments

Along with detailed student background characteristics, the baseline surveys of both TREE cohorts provide elaborate measurements of general literacy skills⁸ that are at the respondents’ command at the end of their compulsory schooling (9th grade). Documentation of these tests is under the responsibility of the OECD (in the case of TREE1’s baseline survey PISA 2000) and the Swiss Conference of Cantonal Ministers of Education (EDK, in the case of TREE2’s baseline survey AES). The respective documentations are provided along with the TREE data releases. It should be noted that the tests themselves must not be published for licensing and confidentiality reasons. The subsequent TREE panel waves then collect detailed data on education and labour market trajectories, which are contextualised by a rich set of complementary information on various life domains that have been identified in previous research as factors relevant for the respondents’ later transitions from education into working and adult life. Detailed documentation of survey instruments are provided for both cohorts and available when downloading the data from SWISSUbase (TREE, 2019, 2023).

2.6 Quality Control

TREE guarantees high standards in terms of quality control through all stages of the survey-data cycle. Field work is mandated to a professional survey institute⁹ and thoroughly supervised by the TREE survey management. The institute has a solid reputation for the implementation of large-scale, multi-language national scientific surveys in Switzerland. Survey instruments, documentation and interviewer training materials are designed and developed in close cooperation between the institute and TREE’s survey management. Ongoing quality control during field work is ensured by means of TREE staff being present in the survey institute’s CATI call centres, listening in and providing feedback on the conducted interviews. Furthermore, any adjustments of survey instruments, survey modes or field operation procedures are carefully tested on pretest samples prior to main field operation.

Before the TREE data are made available to the research community, they are subjected to thorough quality and consistency checks by the TREE data management (von Rotz et al., 2023).

2.7 Data anonymisation and ethical issues

When the TREE1 cohort was launched, approval by an ethics committee or review board was not required and not common practice in Switzerland. However, the study has, from its very beginning, explicitly adhered to research methods that strictly follow professional ethical guidelines and good scientific practice. Today, TREE data collection, treatment and publication complies strictly and formally with Swiss ethical and data protection legislation.

Students who participated in PISA 2000 had been asked their explicit consent to be contacted by TREE for a follow-up survey. Students were informed in advance about the purpose and scope of the TREE study and that participation was strictly voluntary. Data are thoroughly anonymised before publication.

Procedures of obtaining panel consent from the participants were largely the same in TREE2 as in TREE1. Due to more rigorous data protection legislation, TREE2 respondents were furthermore called to give their explicit consent that the data collected in the baseline survey (AES) may be linked with the subsequent panel data. A detailed data management plan guarantees strict confidentiality and security standards with regard to collection, treatment and transfer of contact and survey data.

2.8 Data use

TREE is funded as a social science data infrastructure by the Swiss National Science Foundation. Hence, the study’s very “raison d’être” is to provide scholars with empirical data to draw on for their analyses. Over the years, scholars from Switzerland and abroad have taken lively advantage of this opportunity. With a cumulative total of over 500 researchers who have so far downloaded the data from the SWISSUbase data archive, TREE belongs to the country’s most frequently used social scientific datasets.

Accordingly, the scientific output drawing on TREE data has reached a total of approximately 300 publications – and is continually increasing. The full list of publications, most of them available in full text, is available on the study website (see www.tree.unibe.ch/results).

3.1 Repository location

Information on how to access the data, information on the SUFs and the latest data releases is provided at the SWISSUbase repository: www.swissubase.ch/de/catalogue/studies/12476/18017/overview

Furthermore, permanent DOI identifiers have been assigned:

3.2 Object/file name

The TREE data and their documentation are downloadable from the SWISSUbase data archive (see DOI reference numbers in Section 3.1; see also TREE, 2019, 2023). TREE1 data are extensively documented in German, French and English, TREE2 data in English only.

The contents (at single file level) of the most recent available TREE1 data release are listed in Table 6. The contents of the most recent available TREE2 data release are listed in Table 7.

3.3 Data type

The TREE data are basically numerical (at all scale levels). Categorial and open text data are exhaustively numerically coded.

3.4 Format names and versions

The datasets are published in Stata and SPSS formats. Data documentation is provided in PDF and Excel formats. Versioning and releasing of the published data observes international standards and is supervised and implemented by the SWISSUbase repository at FORS center in Lausanne.¹⁰

3.5 Language

TREE is a multi-lingual study collecting data in several survey languages. Therefore, parts of the documentation and data are available in German, French and Italian. For TREE1, the most important documents (study design, concepts and scales, etc.) are provided in English, German, and French. For the second TREE cohort, documentation and labelling of numerical data is available in English only. However, data releases comprise exhaustive documentation of the questionnaires’ wordings in the three national survey languages (German, French and Italian).

3.6 License

The TREE data, as published in the SWISSUbase repository, are basically freely available for scientific use. However, given the sensitivity and comprehensiveness of the collected panel data on the surveyed individuals, there are minimal restrictions to the download of the data (see Section 3.7 for detail).

3.7 Limits to sharing

The TREE data comprise sensitive information. In compliance with data protection regulations, the datasets are available as scientific use files for scientific research and academic teaching only. Access to the data is therefore controlled by the data producers (i.e., TREE). This is custom procedure at the SWISSUbase repository for any sensitive individual data, for which data protection laws prevent open sharing. The data are made available to users upon prior (online) signature of a user agreement and submission of a short description of the type of (scientific) use that the signatories of the agreement plan.

Among other things, the user agreement specifies that

• – the data may only be used for the declared research purpose

• – the data must not be used for commercial purposes

• – users must not attempt to identify individuals

• – the data must be properly cited

• – upon expiration of the agreement, the data and all related materials must be deleted (and deletion confirmed vis-à-vis the data provider).

Duration of the agreement’s validity is limited to a maximum of 24 months. Validity of the agreement may be extended at any time.

3.8 Publication date

The most recent TREE1 data release has been published in May 2019 (TREE, 2019).

The most recent TREE2 data release has been published in June 2023 (TREE, 2023).

3.9 FAIR data/Codebook

Data published in the SWISSUbase data repository – and hence TREE – fully conform to the FAIR principles. Findability is further enhanced by the fact that SWISSUbase datasets are listed in the CESSDA catalogue.¹¹