Keywords

1 Introduction

The aim of the interdisciplinary BiKS study “Educational Processes, Competence Development, and the Formation of Selection Decisions in Preschool and School Age” is the longitudinal investigation of educational and competence development processes. For this purpose, two panel studies (BiKS-3-18; BiKS-8-18) with surveys in the German states of Bavaria and Hesse have been conducted since 2005. This was done against the background of two weaknesses of the German education system, which came to light in the wake of international comparative school performance studies such as PISA 2000 (see e.g., Baumert et al. 2001): On the one hand, students at German schools showed an unexpectedly low level of competence in an international comparison. This was especially true in the lower performance groups and across different competence areas. On the other hand, particularly pronounced disparities in educational participation and skill acquisition had become apparent for students of different social origins and nationalities or migration statuses.

In BiKS-8-18, beginning in 2006, a sample of primary school children in Grade 3 was accompanied for the next ten years of formal education. The transition from primary school to lower secondary school is arguably still the crucial point in the German educational system, and it is critical for the development of social disparities (Maaz et al. 2006). Therefore, the longitudinal study BiKS-8-18 focused on the development of competencies and interests, the formation of educational decisions before and after the transition to lower secondary schooling as well as the continuation of developmental trajectories in lower secondary schooling. Using a multi-method, multi-perspective and multi-level design, the effects of the learning environments of family, primary school, and secondary school, and the interactions between these factors on child development were examined over a period of ten years. On the other hand, BiKS-8-18 focused on the transition from upper secondary school to the vocational training sector and the labor market by surveying adolescents across additional follow-up surveys.

The remainder of this chapter is structured as follows: First, we describe the study design of BiKS-8-18. Then, we describe the sampling and different sub-sample tracking strategies in detail to understand their limitations and potentials. After that, we provide an overview of the study instruments we used to provide a rich multi-level data set. Finally, we briefly discuss the research potentials of the data by illustrating research that has previously been done with the BiKS longitudinal data.

2 Study Design

BiKS-8-18 can be structured in three phases. The first phase consists of the data collection when the sampled children were in the second half of primary school in Grade 3 until they transitioned into lower secondary schooling after Grade 4. The second phase covers the period of lower secondary schooling until the transition into either upper secondary schooling on their path to tertiary education or into vocational education and training (VET), i.e., from Grade 5 through Grade 9 and 10, respectively. The third phase contains students’ time in upper secondary schooling or VET and their transition into either tertiary education or labor market entrance.

2.1 Multi-Perspective Panel Design

The surveys of the first phase of the BiKS-8-18 took place in the states of Bavaria and Hesse and began in spring 2006 in the second half of third grade in 155 classes in 82 primary schools (see Sect. 2.2). The children of the 2,395 participating families were surveyed in three panel waves in six-month intervals. Data collection included the assessment of the children’s competencies. These children transferred to a secondary school in the fall of 2007. In the second phase, these children and their parents were followed at annual intervals in five panel waves until the end of ninth grade. In the last phase, the students’ transition into upper secondary school to the vocational training sector or the labor market was examined in three additional panel waves (see Fig. 1).

Fig. 1
A study design depicts the transition into lower secondary schooling, upper secondary schooling, or vocational education and training, and the transition into tertiary education or the labor market from 2005 to 2016. Data are collected from the various interviews with students, parents, and teachers.

Study design of BiKS-8-18

Full size image

The first and second phases followed a multi-perspective design, which included in addition to student questionnaire and competence tests on the individual level also parent interviews and questionnaires for teachers, thus covering central contexts of the family and institutional learning environment (for further details, see von Maurice et al. 2007). The surveys in the third phase exclusively focused on the individual students.

In primary school, educational trajectories are typically linear and uniform, i.e., with infrequent class or school changes (Bellenberg 2020). With the transition to the various types of secondary school, there is a clear pluralization of school and educational trajectories in Germany: In addition to local school changes, there were also changes in school types and grade level as high performing students could change to a more demanding school type or low performing students may repeat a grade level or even change to a less demanding school type (Bellenberg 2020). Therefore, already in the second phase, surveys in the school context could not be continued equally for all students to cope with this complexity on a practical level. This led to different survey and testing strategies (see Sect. 2.3). Although students were the main target persons, the contact and interviewing strategy differed in the study phases. In the first phase, students were primarily interviewed and tested in the school context, and their parents were the primary contact persons. In the second phase, most students were interviewed in the school context, and some students were surveyed individually. In the last phase, the focus was entirely on the students themselves. Data collection was individualized and without the inclusion of parents or learning context.

To investigate the different research questions longitudinally, 11 panel waves were conducted with a multi-method design in which different survey and test instruments were used. In the first and second phases, the parent interviews were conducted as computer-assisted telephone interviews (CATI). Students and teachers received paper-and-pencil questionnaires (PAPI), and students’ competence assessments were also paper-based (PBA). In the third phase, all survey instruments were either CATI or CAWI (computer-assisted web interview). While the CATI was the primary mode in this phase, the CAWI were mainly used to interview students who were difficult to reach by telephone.

2.2 Sampling Process in Two Federal States

The sampling of BiKS-8-18 followed a stratified multi-step process (for details on the sampling process, see Kurz et al. 2007). In the first step, two federal states, Bavaria and Hesse, were selected. The guiding principle for selecting these two states was to vary relevant contextual factors that determine individual educational decisions systematically. On the one hand, these were the state-specific differences regarding the transition regulations from primary school to secondary schools, in which different emphasis was given to the parents’ free choice and the school track recommendation. According to the regulations of the state of Bavaria, school track recommendation given by the teacher(s) at the end of primary school was of primary importance, and students without a track recommendation have to pass an additional entrance examination. In Hesse, on the other hand, parents were ultimately free to choose the school type in secondary education (see Faust 2005; Secretariat of the Standing Conference of the Ministers of Education and Cultural Affairs of the Länder in the Federal Republic of Germany 2010). Second, the two states significantly differed in terms of the structure of secondary schooling. In Bavaria, the choice of schools included Hauptschulen (lower track schools), Realschulen (middle track schools), Gymnasien (higher or academic track schools), and in some few cases, Gesamtschulen (comprehensive schools). In contrast, integrated and cooperative Gesamtschulen (comprehensive and multi-track schools) were additional regular school types in Hesse.

In the second step, specific survey regions within the federal states were selected based on similarities and differences in the respective opportunity structures and socioeconomic conditions (e.g., presence of different school forms, accessibility, employed persons by economic sectors). This led to the selection of one large city (Bavaria: Nuremberg, Hesse: Frankfurt), one medium-sized city (Bavaria: Bamberg, Hesse: Darmstadt), and two rural districts (Bavaria: Bamberg and Forchheim, Hesse: Bergstrasse and Odenwaldkreis).

In the third step, primary schools were recruited. The following disproportional distributions of participating schools were targeted: First, disproportional stratification by federal state with a ratio of Bavarian and Hessian schools of 60:40, and second, disproportionate stratification by major cities: One-third of schools each of Bavarian and Hessian schools were to come from the metropolitan regions of Nuremberg and Frankfurt respectively. For practical reasons, this was done to link BiKS-8-18 to the other cohort BiKS-3-18 (see Homuth, Lehrl, et al. this volume). The linkage was supposed to be established by recruiting primary schools named by the preschool teachers of BiKS-3-18 as their most important schools where their students would transition to.

In the last step, parents of students in the participating schools were asked to participate in the survey. This resulted in the initial sample of n = 2,395 participating children in wave 1.

2.3 Different Survey and Test Strategies

BiKS-8-18 started in 2006 with 2,395 participating primary school students and their parents. In the fall of 2007, the majority of the children moved on to secondary schools. The aim of the second phase was to continue to follow the remaining participants of the initial sample (n = 2,104, corresponding to 88%) and to expand the sample to include the class context by including their classmates in the study. Since the number of receiving secondary schools was too large to continue to study all participants in the class context together with their new classmates, and not all children transitioned to secondary schools located within the BiKS survey regions, three different survey and test strategies were used (see Fig. 2):

Fig. 2
A flow diagram. It includes initial sampling and competence assessment in primary schooling, a survey in individual and school contexts, competence assessment, and class complement sampling in the lowest secondary schooling, and individual assessment in the upper secondary schooling and tertiary education.

Survey, interview, and test strategy in BiKS-8-18

Full size image

In the first strategy, participants were no longer surveyed in the school context but individually outside their schools. This concerned n = 802 children who either attended schools outside the BiKS survey regions or schools with fewer than three children of the initial sample. Furthermore, children were moved to individual surveying if no information was available about the secondary schools they attended or if their school generally refused to participate.

In the second strategy, students and teachers were only interviewed via paper questionnaires that were administered by the class teachers within the school context. A total of n = 382 children at comprehensive schools, special-needs schools, and schools that did not extend their participation, as well as children at schools in which fewer than three children of the initial sample attended the same class were accompanied this way.

In the third strategy, the remaining n = 920 students in schools with at least one class with three or more children of the initial sample were included. For this subgroup, an attempt was made to include the class context, i.e., all their classmates, in the study. This way, n = 879 children could be recruited additionally as class complement sample. In these schools, all children of the initial sample and the class complement sample were surveyed and tested in class.

If schools no longer agreed to study participation in class, they were asked to switch to the second strategy without competence testing in class. If a school was also not (anymore) willing to participate in this strategy, all children of these schools finally switched to the individual survey context. Furthermore, it happened that individual children switched from one of the two school survey variants to the individual survey variant due to a change of school or, in rare cases, vice versa, switched to a school that participated in the BiKS surveys.

Additionally, to enhance the class complement sample in lower secondary schools (Hauptschulen), n = 14 additional students in these schools were included in the sample.

With the beginning of the third phase of the study, all remaining participating students were surveyed and tested in individual contexts.

3 Sample Development and Panel Participation Rates

Figure 3 shows the evolution of the BiKS-8-18 (gross) panel sample over time and panel participation ratesFootnote 1 differentiated by the initial sample and the class complement sample. Of the n = 2,395 primary school children sampled initially in wave 1 in 2006, n = 1,157 (48.3%) were still part of the panel sample at the end of the study in 2016. Of the n = 879 participants who were included in the study as part of the class complement sample in the fifth grade (see Schmidt et al. 2009), n = 436 (49.6%) remained until the end of the study. Both the initial and class complement samples consistently showed high participation rates. This could be attributed mainly to effective panel management, which consisted mainly of personal contact, the sending of information material, regular feedback on results, and the effective use of incentives (for a detailed description of the incentivization strategy, see Mudiappa and Schmitt 2010).

Fig. 3
A combined bar and line graph plots the sample development and the panel participation rates of the initial sample and the class complement sample. The highest sample development for the initial and the class complement samples are 2395 in wave 1 and 879 in wave 4 respectively.

Sample development of BiKS-8–18 (absolute) and panel participation rates (in percent)

Full size image

Figure 3 also shows that in wave 7, panel participation rates significantly dropped below the average, with only about 68% for the class complement sample and 77% for the initial sample. The drop in the panel participation rate at wave 7 can be attributed to the fact that it became necessary for the Bavarian subsample to obtain written panel consents again. Prior to wave 7 and due to regulatory change in Bavaria, it was necessary to obtain again written consent from the Bavarian families being interviewed in the school context. Since the participating adolescents were already 14 years old at this time, they had to explicitly consent in addition to their parents. Only if both parents and adolescents had given their consent, the family could be interviewed further. Only those families who had actively provided their consent were allowed to remain in the study.

This not only led to a decline in the panel participation rate but also to an aggravation of the selective panel mortality typical for panel studies (for further details, see Homuth et al. 2017). Analyses showed that the high dropout at wave 7 was precisely due to the affected subgroup of Bavarian families in the school survey. In contrast, Hessian and Bavarian families in the individual survey strategy variant did not show a disproportionately high tendency to leave the panel (Homuth et al. 2017).

4 Sample Description

4.1 Basic Composition

Table 1 provides a supplementary overview of the basic sample characteristics and distributions at the beginning and end of each phase. The distribution across the states shifted slightly in favor of Bavarian families from waves 1 to 4 due to the higher proportion of Bavarian families as part of the class complement sample. Classmates in comprehensive schools were not included in the class complement sample, so the lower proportion of Hessian children is due to such study design decisions. In the third phase, in waves 8 to 11, the described selective dropout of Bavarian participants led to a relatively equal distribution across states.

Table 1 Selected sample characteristics of the longitudinal study BiKS-8-18
Full size table

The proportion of girls in the complete sample after the inclusion of the class complement sample was also significantly higher than in the initial sample. Regarding the sample composition in terms of social origin characteristics, shifts in the sample composition from the first to the second phase could be observed in the direction of more highly educated parents as well as parents with an upper socioeconomic position and children without a migration background still participating. This can be explained by the comparatively higher participation rates of families with a high socioeconomic position, of parents with a academic educational background, and of families without an immigrant background in the class complement sample. Higher response rates at academic track schools (Gymnasium) in comparison to other school types are the primary explanation for this change in the sample composition. Accordingly, a disproportionate number of children from this type of school were added. Thus, a larger share of children from homes with these characteristics was newly included in the study.

4.2 Selectivity

Like in any other longitudinal study, selective panel attrition is a paramount concern (Rendtel 1995). One major question is whether the dropouts were neutral concerning central sample indicators. Overall, continued participation was not systematically biased on the dimensions gender, age of the participants, region, social origin, parental education, and migration background, which taken together accounted for only 0–10% of the total variance in participation in each panel wave (cf. adjusted R2 in Table A1 in the Appendix).

As expected in longitudinal studies (e.g., Behr et al. 2005), the proportion of higher-educated families increased over time as lower-educated families had on average a higher dropout risk. However, the dropouts of participants with less-educated parents mostly occur at specific times (waves 2, 3, 8, and 11) and not each wave. It seems that the parents of the class complement sample were on average significantly higher educated across all waves due to higher participation rates in academic track schools (Gymnasium). Concerning the socioeconomic status and the migration status of the participants, no specific attrition patterns can be identified. Mean HISEI and standard deviations vary only slightly both in the group children of the initial sample and the class complement sample. Again, only the differences between the groups are substantial but not between waves.

This also applies to the migration background: At the beginning of the study, 23.6% of the initial sample and 18.8% of the class complement sample had at least one parent who was not born in Germany. At the end of the study (wave 11), the share of migrant students was 23.3%, which is about the same as at the beginning of phase two of the study at the beginning of lower secondary schooling (wave 4).

There were significant changes in composition in the third study phase as the dropout between these last waves was selective along the dimensions of gender, social status, and educational background. Mainly male participants and those from lower education and low social status households left the sample. The dropout was not significantly correlated to migrant background or spatial distribution (federal state).

4.3 Educational Trajectories

In the first phase of BiKS-8-18, all students by design attended primary school. With the beginning of the second phase, i.e., the transition into secondary school, the educational trajectories of the sample start to diverge. Figure 4 shows the educational status of the sample at the respective wave.

Fig. 4
A stacked bar graph plots the educational status of primary schooling, lower, middle, comprehensive, and higher tracks of the lower secondary schooling, vocational education and training, labor market participation, vocational preparation, higher secondary schooling, and tertiary education from wave 1 to wave 11.

Educational status over time

Full size image

Due to the selective participation of classmates of the initial sample as part of the class complement sample, the proportion of students attending the higher academic track of secondary schooling was the largest from wave 4 onwards. Due to a change in the survey and assessment strategies, there was a group of participants for whom we do not have the information about their respective educational status or attended school type, respectively.

Educational trajectories further diverged in the third phase of the study. This phase encompassed the transition from lower secondary schooling into either upper secondary schooling or vocational education and training (VET) and beyond. At the end of study phase three, there were still participants who attended a general education school, while others had already entered tertiary education or the labor market.

At wave 9, the majority of the sample (n = 806) was attending a general education school (mostly higher track school/Gymnasium), n = 139 were undergoing vocational training, and n = 27 indicated that they were neither attending school nor in vocational training; no information was available for n = 934 participants. In wave 10, there were no significant changes; n = 798 participants were still attending a general education school, n = 162 were in training, and n = 25 were classified as participants in vocational preparation (transition system). In wave 11, the picture changes: most participants made a transition. Only n = 274 participants were still attending a general education school. In contrast, n = 267 were in VET, n = 303 had started a tertiary education, and n = 48 participants had entered the labor market. Interestingly, however, a rather large group of n = 207 was still in vocational preparation. Among them were a particularly large number of high school graduates who stated that they wanted to take a “time-out” first.

5 Contents of the Study

5.1 Instruments and Measurement Times

Table 2 provides an overview of the instruments by context, when and in which mode they were employed, and their main contexts.

Table 2 Overview of instruments
Full size table

During the first eight waves, students were interviewed by paper-based questionnaires. In the third phase of the study (waves 9-11), students were the single informant, and the interviews were telephone-based. An additional web interview was conducted in the last two waves to interview participants who could not be reached via telephone interview.

Parental data were obtained by interviewing the parents primarily concerned with the children’s school matters. These parents were interviewed via CATI (Computer-assisted telephone interview) at all waves during the first two study phases (waves 1-8).

During the first two phases of primary and lower secondary schooling, participants’ class teachers were interviewed via a paper questionnaire which included a child-related assessment sheet for each participating student. In the survey group which included students’ competence assessment in waves 4 through 8, teachers completed the additional child-related assessment sheet as well. Teachers were surveyed at waves 4 through 8 in both school-based survey groups (with and without competence assessment), analogous to waves 1 through 3. While, in the intensive version, the class and subject teachers (English as a first foreign language, Mathematics, and German) were included in the survey. In the non-intensive version, only the class teacher was asked to participate.

5.2 Response Rates by Instruments and Waves

Numerous instruments were used in the BiKS-8-18 (see Table 3). The assessment of students’ competencies and student questionnaires were a central component in the first and second phases of the study. In waves 1 to 3, all children participated in competence assessments and answered questionnaires in the classroom context. After the transfer to the secondary schools, only the subgroup of participants in the third survey strategy provided for differentiated testing of the children’s competencies and student surveies were conducted in the interviewer-controlled class context, analogous to the procedure during the primary school period. In the second survey strategy, the response rate was mainly dependent on the cooperation of the participating teachers who handed out the questionnaires themselves. While the response rate in this group was almost as high as in the third survey strategy up to wave 6, response rates declined in waves 7 and 8. In the individual survey strategy, which was highly dependent on the cooperation of the parents, average response rates of about 65% were achieved.

Table 3 Sample sizes and response rates by panel waves and instrument
Full size table

With the transition into the third study phase, participation dropped significantly. When the interviewers contacted them, many participants withdrew their consent to further participation.

The following reasons can be seen as significant for the lower participation rates in Waves 9 to 11 compared to the previous waves:

  1. 1.

    Change of the primary contact person: During the entire duration of the first and second phases of BiKS-8-18, the children’s parents were the primary contact persons. The adolescents were not used to being contacted directly and being responsible for their participation in the study.

  2. 2.

    Change of the survey mode: In the first and second phases of BiKS-8-18, the children were interviewed exclusively by paper-based questionnaires, and their parents were interviewed by telephone. However, it was always possible to participate only by sending back the paper questionnaire so that there had always been persons for whom no telephone numbers were available.

  3. 3.

    Greater time interval between interviews: While all surveys (except wave 2) took place at annual intervals in the second half of the respective school year, two school years elapsed between wave 8 and wave 9.

  4. 4.

    Increased mobility: With the end of compulsory schooling and the move to vocational and tertiary education, the mobility of adolescents is increasing, as relocation due to training and studying is unavoidable for many. Additionally, many adolescents spend time abroad either during upper secondary school or after graduating for voluntary service or work-and-travel stays.

  5. 5.

    Reduced accessibility: Adolescents who complete training or vocational preparation usually work full-time and can only be reached in the late afternoon, evenings, and weekends. In some professions, shift work is added to this, further limiting accessibility by telephone.

6 Research Potentials of BiKS-8-18

The BiKS-8-18 study offers a broad dataset that allows high-quality empirical education research within the German education system from an interdisciplinary perspective. Therefore, the BiKS study has contributed and will contribute substantially to a better understanding of educational decision making, learning and teaching processes, and the educational outcomes of these processes.

For example, in the BiKS-8-18 study, a strong emphasis was placed on assessing students’ academic competencies and tracing their development from primary to secondary school. Thereby, the focus was on general cognitive abilities, mathematics, oral language and reading skills. Furthermore, psychological variables important for students’ academic development such as self-concept, school-subject interests, goals, and motivation were considered. As the BiKS study incorporates the perspectives of students, parents, and teachers, a broad set of variables that may cause individual differences in the development of academic competencies may be explored. Furthermore, as students were assessed within their class contexts, context effects can be analyzed. Some examples for such analyses are provided in Pfost et al. (this volume) and Karing et al. (this volume) of this volume. For instance, in a study by Pfost and Artelt (2013), the effect of attending the upper academic track school (Gymnasium) in comparison to lower and middle track school (Haupt-/Realschule) for reading development between Grade 5 and Grade 7 was analyzed. In another study, Schurtz et al. (2014) analyzed the complex interrelation between students’ academic interests, competencies, and grades between Grade 5 and Grade 6. The empirical analyses were embedded within assumptions of the internal/external frame of reference model (I/E model; Marsh 1986) as well as the big-fish-little-pond-effect model (BFLPE; Marsh 1987). Third, Becker et al. (2017) explored the role of learning environments for students’ goal orientation. Thereby, the development of students’ mastery and performance goals between Grade 5 and Grade 11 was described and related to variables such as the transition from school to vocational training. Or fourth, Karing (2009) analyzed teachers judgement accuracy in reading respectively language arts and mathematics. In her study, judgement accuracy was related to teacher and class characteristics or the whether the judgement refers to cognitive (students’ competencies) or non-cognitive (students’ interests) outcomes. However, besides analyses conducted by researchers within the BiKS research group, the BiKS-8-18 study still offers an extensive range of further possible analyses. For example, analyses that relate individual differences in academic competencies when children were in primary school to education decisions and pathways at the end of secondary school, including the transition into tertiary education, are still scarce.

Another strong emphasis in the BiKS-8-18 study was on the educational decision-making process of parents (and teachers), its surrounding activities, and the influence of social relations before and after the transition from primary to secondary school in Germany. Researchers from the BiKS group examined and compared differences of family background, family aspirations, family burden, family social networks, and institutional differences in the federal states of Bavaria and Hesse on educational outcomes (see Blossfeld et al. this volume): The parental educational aspirations are an essential factor in the decision-making process. Parental aspirations can be understood as representations of parents’ expectations about their children’s possible future educational pathways (Kleine 2014; Kleine et al. 2010, 2013). In addition, Luplow and Schneider (2014) and Luplow (2017) examined the role of tutoring by parents at home or through non-familial institutions during primary school years—as possible tools for parents to pursue their educational goals—on educational outcomes. Further studies addressed the influence of families’ social capital and family burden on their children’s educational success in their analyses (Kleine 2014; Kleine et al. 2013; Luplow 2017; Schmitt 2012; Schmitt and Kleine 2010; Schmitt and Sixt 2014). After the transition to secondary school, the revision and stabilization of previously made school transition decisions, especially with regard to differences in the school type choices in Bavaria and Hesse was analyzed (Zielonka 2017; Zielonka et al. 2013; Zielonka et al. 2014).

In sum, there is still plenty of potential in the data of BiKS-8-18 to further contribute to a better understanding of competence development, educational pathways, and decisions. This rich data set allows researchers to relate these educational trajectories with later transitions into university studies, vocation training or work, taking into account individual differences between students and families from an earlier point in time. Due to its multidisciplinary perspective, the BiKS-8-18 study can still offer interesting empirical answers to open research questions. Furthermore, research that takes the complex interactions of different actors such as parents, teachers, and the student within different educational contexts and across different educational stages into account is still warranted.