+++ Abstract submission is closed +++ Registration to participate is open +++

Poster Session

The poster session is scheduled for Thursday, December 10, starting at 2:15 pm. However, all posters will be exhibited throughout the entire NEPS Conference, so that an exchange can also take place during the breaks. The contact persons for the posters are marked with an asterisk in the following list. Please click on ≡ to open the respective abstract.

• Psychological resources for a successful transition into adult life
  Annalisa Schnitzler* (BIBB Bonn)
  ≡ Abstract

  Young people transitioning into the world of work and into a self-determined life are faced with numerous challenges, for example unclear perspectives concerning their entry into work after finishing their education and stressors affecting their mental health (e.g. for the context of dual Vocational Education and Training (VET), DGB, 2026). Tenacity and flexibility in goal pursuit, internal locus of control and self-control can be considered as valuable psychological resources for coping resiliently with challenges (cf. Hobfoll, 2002; Schäfer et al., 2024) and have proven to positively affect various life outcomes, for example in the case of tenacious and/ or flexible goal pursuit older people’s life satisfaction (Zhang, 2020) and their depression symptoms and self-rated health (Bailly et al., 2016), as well as school-leavers’ success in transitioning to VET (Theuer & Weßling, 2025). Self-control has been shown to positively impact educational attainment and income (Converse et al., 2014), school and work performance and well-being (de Ridder et al., 2012). Internal locus of control has demonstrated positive relationships with inter alia career success (Smidt et al., 2018), employees’ vocational satisfaction (Abdel Hadi et al., 2023) and less psychopathological symptoms (Brosschot et al., 1994).
  This proposal investigates whether these psychological resources are also positively related to job and personal outcomes in young people entering adult life and includes outcomes from the areas of health, satisfaction and the entrance into work life. I resort to data from NEPS Starting Cohort 3 and make use of several survey waves, on the one hand because the constructs of interest were assessed in different waves and on the other hand to measure outcomes at a later point in time (i.e., waves 12 and 13) than the psychological resources wherever possible.
  In a first step, bivariate correlational analyses show significant relationships for all four constructs with outcomes from all areas. Some outcomes, like general life satisfaction, are related to all four constructs, while other areas show a different pattern of relationships with specific outcomes depending on the construct. For example, physical health is only related to internal locus of control, whereas perceived chances to be taken on by the training company after completing VET are only related to self-control. Next, I will conduct regression analyses to control for socio-demographic variables and also to look at the interplay of the constructs in their relationship to the outcomes.



• Vocational self-selection into craft apprenticeships: Regional craft opportunity structures, social background, and individual fit among German school leavers
  Johannes Ennenbach* (LIfBi Bamberg)
  ≡ Abstract

  Germany´s craft sector has faced persistent recruitment shortages for nearly three decades, intensifying skilled-worker shortages (BIBB, 2025). Understanding why young people choose or reject craft apprenticeships matters for recruitment and the sector´s future. Craft entry is shaped by distal contexts, including family background reflecting linked lives across generations (Elder et al., 2003) and regional structures affecting the visibility of craft opportunities (Weßling et al., 2015; Malin & Jacob, 2019; Flohr et al., 2020), which may shape proximal dispositions. Following Holland´s (1997) fit model, vocational self-selection reflects occupational interests. Practical-technical interests favor manual-technical occupations, whereas social interests favor person-oriented environments. Gottfredon´s (1981; 2005) compromise principle adds that choices are constrained by gender and status. Generally, vocational entry is linked to social background, qualifications, interests and career aspirations (Schleer & Calmbach, 2022; Busse et al., 2024), but their role in craft entry remains unclear (Mischler, 2017).
  This study examines how regional structures, family background, vocational interests and aspirations relate to craft entry among German school leavers. Using NEPS Starting Cohort 3 (Blossfeld & Roßbach, 2019; NEPS Network, 2025; N=4,604; mean school-leaving age≈17.8), discrete-time event-history models with piecewise-constant hazards estimate craft apprenticeship entry (ISCO-08). Models sequentially add regional craft density, gender, migration background, parental SES, craft occupation, qualification, then vocational interests are added, and finally prestige aspiration and craft internship experience. Fit improves with each added block, and Cox-Snell diagnostics show no clear misspecification (Blossfeld & Rohwer, 2019). SE´s are district-clustered and missing data are multiply imputed.
  Results show 13.6% of school leavers enter a craft apprenticeship, with entry peaking shortly after school leaving and smaller peaks one and two years later. Practical-technical interest is the strongest predictor, nearly quadrupling the hazard at the highest level, while social interests and university qualifications reduce it (all p<.01). Parental craft occupation and internship increase entry. Girls and migrant youth have lower hazards across specifications, even after controlling for interests and aspirations (all p<.01). Regional craft density predicts entry before aspirations and experience are added but loses significance thereafter. Occupational aspirations both below and above craft-prestige range are associated with lower hazards, strongest for upper-secondary/university-level occupations.
  Vocational interests, gender, family background and occupational exposure emerge as key recruitment factors, consistent with Holland´s (1997) Realistic-Social distinction and Gottfredson´s (1981; 2005) compromise principle. Career guidance could target students with realistic interest and craft-related experience, particularly underrepresented girls and migrant youth.



• CAP-VET: Choices, aspirations and perceptions in vocational education and training. Educational pathways, social perceptions and institutional experiences in Brazil’s Federal Institutes – A German-Brazilian research collaboration
  Silvia Annen* (University of Bamberg)
  Melanie Hochmuth (University of Bamberg)
  ≡ Abstract

  Vocational education and training (VET) provides an important route into employment and further education, yet educational choices are also shaped by the relative status attached to vocational and academic pathways (Cantwell et al., 2018; OECD, 2015). This tension is particularly salient in Brazil, where the share of upper-secondary students enrolled in vocational programmes remains comparatively low: among 15- to 19-year-olds, 11% are enrolled in vocational programmes, compared with an OECD average of 37% (OECD, 2023). At the same time, Brazil’s Federal Institutes combine technical, vocational and higher education and have expanded access to publicly funded education across different regions (Annen et al., 2024).
  CAP-VET examines how educational aspirations, family expectations, perceptions of VET and institutional experiences interact in shaping educational and occupational pathways, and how these relationships differ between Brazil and Germany. The theoretical framework combines perspectives on educational stratification and status hierarchies with a multi-level understanding of educational participation, linking socioeconomic and family background to perceived opportunities and institutional conditions as tracking and specificity (Boeren et al., 2010; Bol & Van de Werfhorst, 2013; Marginson, 2016).
  The comparative design uses Germany as a longitudinal reference context. Starting Cohort 4 of the National Educational Panel Study (NEPS) follows educational pathways from Grade 9 through transitions into upper-secondary, vocational and higher education and the labour market. More broadly, NEPS provides a longitudinal framework for analysing educational processes and decisions across the life course (Blossfeld & Roßbach, 2019). CAP-VET draws on conceptually corresponding dimensions, particularly ideal and realistic educational aspirations, parental expectations and choices between vocational and academic pathways. Rather than replicating NEPS directly, the project develops a construct-oriented comparison that accounts for the different institutional configurations of VET in Germany and Brazil.
  The Brazilian empirical phase will establish the corresponding database through a quantitative survey of students and alumni of Federal Institutes, covering educational and occupational aspirations, parental expectations, institutional preferences, social recognition of VET, institutional experiences and perceived educational and occupational returns. Qualitative interviews with teachers and selected institutional actors will complement the survey by contextualising individual-level patterns. Brazilian data collection is currently being prepared. The first wave will establish the empirical basis for subsequent comparative and longitudinal analyses of how family expectations, institutional contexts and the perceived status of vocational education contribute to pathway formation in two structurally different (vocational) education systems.



• Who trains and who gains? Training, task composition, and labour market stratification in the green transition
  Jana Kern* (IAB Nuremberg)
  Markus Janser (IAB Nuremberg)
  ≡ Abstract

  The green transition reshapes work less through the entry and exit of whole occupations than through the recomposition of tasks within occupations. Regulation, technology, and demand expand green tasks—environmentally friendly activities—while rendering brown tasks—environmentally harmful activities—obsolete. Whether workers keep pace can hinge on further training, yet training access is strongly stratified. This tension is especially salient in Germany, where an industry-heavy structure generates high adjustment pressure while initial vocational education largely fixes career paths, shifting the adjustment burden onto further training.
  We bridge two separate literatures: green transition research examines exposure to green and brown tasks but brackets training, while training research analyses stratified participation but ignores task profiles. Our guiding question is: Who trains when tasks turn green, and who gains from training? If access is unequal across task profiles, the transition amplifies existing inequality rather than reducing it.
  Combining the task-based approach, human capital theory, and segmentation theory, we derive three predictions: participation is higher in green and lower in brown profiles (H1); wage returns are stronger in green and weaker in brown (H2); training stabilises employment, most in brown profiles (H3).
  We link NEPS SC6 survey data with IAB administrative biographies and the Greenness of Jobs Index (GOJI), which measures the net share of green minus brown tasks per occupation. The person-year panel covers 8,647 individuals and about 57,000 person-years over 2012–2024 (ages 25–64). We estimate a two-part model for participation and intensity (H1), a Mundlak fixed-effects wage model (H2), and a Cox competing-risks model for employment stability (H3).
  Findings are sharper than the hypotheses. H1 holds only halfway: a clear brown penalty but no green premium, and the gap is almost entirely an employer gap, not worker motivation. H2 holds (green > white > brown; firm-sponsored training pays more). H3 is refuted in favour of segmentation: training stabilises employment cumulatively (about 12, 25, and 35 additional days after one, two, and three years), yet the safety net does not catch those whose tasks are disappearing.
  The transition does not lift green profiles; it leaves brown profiles behind. Inequality sits at the gate to training, and that gate is the employer's, making firm-sponsored access the central policy lever.



• Aspiring and attaining: Social networks, well-being and occupational trajectories
  Natalie Päßler* (LIfBi Bamberg)
  ≡ Abstract

  Aspirations are a key indicator of career development, shaping career decisions and providing a benchmark against which we can evaluate subsequent labour market outcomes, particularly subjective well-being. When occupational attainment does not match these aspirations, the resulting discrepancy, the Aspiration-Attainment Gap (AA-Gap), has been linked to lower well-being at work (Nießen et al., 2021). However, the AA-Gap itself has rarely been the focus of research, mostly being examined as an outcome of sociodemographic background or in the context of uncertainty within the formation of aspirations. Considering the evidence linking the AA-Gap to subjective well-being, closer attention should be given to how it emerges and unfolds over time.
  To explain how the AA-Gap emerges, the extended status attainment model posits that significant others, particularly the expectations of parents and peers, shape adolescents' aspirations and subsequent attainment (Sewell et al., 1969), so that even adolescents with similar aspirations may diverge in attainment depending on the support they receive. As the AA-Gap is defined by its relative discrepancy rather than an absolute outcome, its consequences for well-being are best understood through frameworks that likewise emphasise discrepancy over absolute position, such as Person-Environment Fit Theory (Holland, 1997) and Self-Discrepancy Theory (Higgins, 1987), both of which assume that well-being depends not on one's absolute position but on its deviation from one's aspirations.
  Building on this, the dissertation examines the AA-Gap across three related research questions, drawing on data from NEPS Starting Cohort 4. One question addresses how significant others shape adolescents' career attainment both directly and indirectly, with the adolescents’ occupational aspirations functioning as a mediator. A second turns to the consequences of the AA-Gap for subjective well-being, examining whether the costs of underachievement and the benefits of overachievement are comparable in size or diverge in magnitude. Another point of interest is whether the AA-Gap affects occupational stability, both directly and indirectly via job satisfaction, with a greater gap expected to increase the likelihood of career change.
  This dissertation conceptualises the AA-Gap as a continuous process, from its emergence and network influences during the aspiration-to-attainment conversion stage, to its long-term consequences for subjective wellbeing and occupational retention.



• Unemployment and mental health: The role of gender and gender role attitudes
  Felix M. Hafner* (University of Bamberg)
  Diana Steger (LIfBi Bamberg)
  Jan Lenhart (University of Bamberg)
  ≡ Abstract

  Traditionally, men are expected to be the primary breadwinners in their families. Consequently, unemployment may be particularly detrimental to men’s psychological well-being because it challenges this gender role. Beyond its economic function, employment provides psychosocial benefits such as activity and structured daily routine (Jahoda, 1981). Therefore, it is not surprising that across longitudinal and cross-sectional studies unemployment has been consistently associated with lower well-being (McKee-Ryan et al., 2005) and poorer mental health (Paul & Moser, 2009). Furthermore, gender has been identified as a moderator (Gedikli et al., 2023), with unemployment being more strongly associated with poorer mental health among men than women. Gender role attitudes may contribute to explaining this pattern. In countries characterized by more traditional gender role attitudes gender differences appear to be more pronounced (Tattarini & Grotti, 2022). As traditional and egalitarian attitudes coexist in Germany (Kleinschrot, 2024), individual differences in gender role attitudes may help to explain the moderating role of gender. The present study therefore examines how gender and gender role attitudes contribute to the association between unemployment and mental health.
  We used data from Starting Cohort 6, Wave 4 of the German National Educational Panel Study (Blossfeld & Roßbach, 2019). We conducted an ANCOVA with the mental health component of the SF-12 as the outcome variable and binary gender, current employment status (employed/unemployed), and dichotomized gender role attitudes as factors, controlling for educational level, age, and the presence of children in the household. Main effects were significant and consistent with previous findings: Men, employed participants and egalitarian participants reported better mental health. Significant interactions were found between gender and unemployment, gender role attitudes and unemployment, as well as a significant three-way interaction. Contrary to our expectations, traditional unemployed women showed the lowest levels of mental health, whereas traditional unemployed men reported the highest levels among unemployed participants.
  Several explanations may account for these unexpected findings. First, self-reported mental health may be influenced by gender role attitudes, potentially contributing to differences between men and women. Thus, we plan to examine additional health-related measures such as alcohol consumption that are less influenced by social desirability. Second, the low mental health of unemployed women with traditional gender role attitudes may reflect pre-existing differences in mental health. Longitudinal analyses incorporating baseline mental health will therefore be conducted to examine whether these differences precede unemployment and to determine the temporal relationship between unemployment, gender role attitudes, and mental health.




• Exploring science-related activities: Findings from NEPS Starting Cohort 8
  Kim Heder* (LIfBi Bamberg)
  Jan Lenhart (University of Bamberg)
  Ilka Wolter (LIfBi Bamberg)
  ≡ Abstract

  Science-related activities as one component of STEM education may offer important opportunities for STEM learning. Previous research has primarily examined the potential benefits and design of organized STEM programmes (Öndeş, 2024). However, less is known about who engages in STEM activities, how these activities relate to broader activity patterns, and in which contexts they take place.
  Using data from Starting Cohort 8 of the National Educational Panel Study (NEPS Network, 2026), this poster presents exploratory findings on what characterizes students’ engagement in science-related activities in terms of frequency, activity breadth, context, and continuity, drawing on a multidimensional conceptualization of youth activity involvement (Bohnert et al., 2010).
  Of the 4,188 students in Grade 6 (Wave 2) with complete information across all six activity domains (painting, making music, writing, craft, sports, science), 38.2% engaged in science-related activities at least several times during the previous month. Initial analyses indicated that science participation was typically embedded in broader activity patterns: Only 0.6% of science participants engaged exclusively in science, whereas 70.3% of science participants engaged regularly in at least five of the six domains.
  Girls and boys did not differ significantly in whether they engaged in science-related activities. Daily engagement was the only frequency category showing a significant gender difference, with boys being more likely than girls to report daily science engagement.
  Among 1,480 science participants with valid context information, 53.6% engaged in self-organized science activities, 34.5% in organized settings within school, and 11.9% in organized out-of-school settings. Although self-organized science activities were most often pursued alone (63.4%), students also reported engaging in them with parents, peers, siblings, or other adults.
  Initial longitudinal analyses indicated considerable fluctuation. Among students who participated regularly in science activities in Grade 5 (Wave 1), only 54.3% remained regular participants in Grade 6. Conversely, 26.6% of students who had participated rarely or not at all in Grade 5 became regular participants in Grade 6. Overall, these initial findings suggest that students’ involvement in science-related activities may be embedded within broader activity patterns rather than representing an isolated domain of STEM-specific engagement. The patterns of continuation and uptake further indicate that this involvement is dynamic. Moreover, the predominance of self-organized activities highlights the importance of considering student-initiated forms of science engagement, which may be overlooked when research focuses primarily on organized STEM programmes.



• Reading competence development from grade 5 to grade 12: A latent growth curve approach with NEPS data
  Kerstin Schwenkschuster* (LIfBi Bamberg)
  Simon Tiffin-Richards (IPN Kiel)
  Jan Lenhart (University of Bamberg)
  ≡ Abstract

  Reading competence is foundational for academic success, yet long-term trajectories across secondary education and beyond remain understudied. Prior longitudinal studies have predominantly focused on shorter period or younger samples (e.g., Richmond et al., 2024; Ricketts et al., 2020). Integrative models such as the Direct and Indirect Effects Model of Reading (DIER; Kim, 2017) conceptualize reading competence as emerging from the interplay of cognitive, metacognitive, linguistic, and motivational components, which may contribute differentially to initial reading competence and its subsequent growth. This study examined reading competence development from Grade 5 to Grade 12, its cognitive, linguistic, metacognitive, and motivational predictors, and whether the development and predictor associations differ between students who left the school system before Grade 12 (school leavers) and those still attending school.
  We used longitudinal data from Starting Cohort 3 of the National Educational Panel Study (NEPS; N=4,395), with reading competence assessed at four measurement points. Latent growth curve models were estimated in Mplus. A quadratic growth specification, selected based on model comparison, was regressed on 15 predictors including reading speed, reasoning, vocabulary, declarative metacognition, reading motivation, reading amount, reading self-concept, and socioeconomic and demographic covariates. A multi-group model compared school leavers and non-leavers.
  Reading competence showed substantial average growth (M=0.36, SE=0,01, p<.001), with a decelerating pattern (M=-0.01, SE=0,00, p<.001). The intercept variance was significant (variance=1.33, SE=0.08, p<.001), indicating substantial interindividual differences in initial reading competence. Significant positive predictors of the intercept were vocabulary, reading speed, reasoning, declarative metacognition, reading self-concept, reading amount, intrinsic reading motivation, gender, parental education, and school type; fewer and smaller effects emerged for the linear and the quadratic slope. Non-leavers showed a significantly higher initial reading competence (M=0.48, SE=0.04) than school leavers (M=-0.39, SE=0.05), χ²(1)=238.40, p<.001. Growth-related parameters and predictor associations showed no robust group differences after correction for multiple comparisons.
  These findings suggest that cognitive and metacognitive resources, alongside sociodemographic factors primarily shape initial reading competence rather than its subsequent growth. The intercept gap between school leavers and non-leavers points to early reading competence as a marker of later educational pathways, suggesting that interventions may be most effective early in secondary schooling.




• Highlighting – The new item format in the reading competence tests in the NEPS
  Karin Gehrer* (LIfBi Bamberg)
  Madlen Mangold (LIfBi Bamberg)
  Kerstin Schwenkschuster (LIfBi Bamberg)
  ≡ Abstract

  Among other topics, the National Educational Panel Study (NEPS) investigates the development of various competencies across the life span (Weinert et al, 2019; Kompetenzsäule, 2020). One of the competencies assessed longitudinally and coherently across all NEPS starting cohorts is reading competence. Reading competence is defined as the functional ability to comprehend written texts (Gehrer et al., 2012, 2013). The stage-adapted test forms used to assess reading competence comprise five text types serving different functions and three types of cognitive requirements.
  With regard to the computer-based assessments introduced in NEPS Starting Cohort 6 (SC6) from 2016 onward, research and development have been conducted on new item formats. Alongside the implementation of the Text Enrichment Task (TET; Rohm et al., 2019), the transition to computer-based assessment opened up new possibilities for item development. Text highlighting is a widely used learning strategy from primary school through higher education (e.g., Wade, Trathen & Schraw, 1990). Consequently, electronic highlighting can be expected to be intuitive and easy to explain to respondents.
  Accordingly, our recent research focused on the development of the new task format called “Highlighting” (HL; Gnambs & Gehrer, 2026; cf. Heyne et al., 2020). To answer a specific question about the text, HL requires respondents to identify and mark the appropriate passages in the text. From a participants’ perspective, one purpose of HL was to make the assessments more engaging by introducing a new format that uses colors and allows more direct interaction with the text. Furthermore, highlighting electronic texts resembles individuals’ natural highlighting behavior when reading and studying printed texts and books.
  HL items as new item format were first introduced in Starting Cohort 6 (adults) and Starting Cohort 4 (young adults) in 2024/2025 (wave 17). The reading competence test consisted of 37 items representing different cognitive requirements and text types. The items were distributed across two partially overlapping booklets: an easy booklet (27 items) and a difficult booklet (26 items). The test was administered to a total of 5,812 participants (49% women) from SC4 (N=2,251) and SC6 (N=3,561). The easy booklet contained two HL items, whereas the difficult booklet contained three. Overall, item fit was satisfactory (Gnambs & Gehrer, 2026). HL proves to be a stimulating addition to the assessment, while presenting specific difficulties. The introduction of innovative item formats represents a significant development toward enhancing the currency and relevance of reading literacy measurement.



• When is short too short? A longitudinal evaluation of long, short, and single-item measures of self-esteem
  Diana Steger* (LIfBi Bamberg)
  Elena Wittmann (LIfBi Bamberg)
  Jenny Wagner (University of Hamburg)
  Tanja M. Gerlach (LIfBi Bamberg)
  ≡ Abstract

  Self-esteem is a central construct in psychological research and is routinely assessed in longitudinal studies due to its associations with a wide range of developmental, educational, and health-related outcomes. As these studies increasingly rely on large, multi-topic panel designs, questionnaire space becomes a scarce resource, creating a need for brief yet psychometrically sound measures. Although the established Rosenberg Self-Esteem Scale (RSES) is already relatively concise with ten items, even shorter instruments, as well as single-item measures, are often used in practice. However, although the psychometric consequences of shortening scales have primarily been evaluated in cross-sectional settings, considerably less is known about whether brief measures yield the same conclusions in the context of longitudinal analyses. For longitudinal research, the relevant question may therefore be not only whether shorter measures reproduce the psychometric properties of their longer counterparts, but whether they preserve the substantive conclusions researchers draw about change, individual differences in developmental trajectories, and the correlates of these trajectories.
  To this end, we use data from NEPS Starting Cohort 3 (covering waves 1 to 10, N=2,964) to compare longitudinal patterns of the original 10-item version of the RSES with a recently developed 4-item short form and their single items to investigate the impact of scale shortening. Specifically, we model these different operationalizations as latent growth curve models using self-esteem data assessed at wave 1, 5 and 9/10 (during the final year of school) to compare model fit, measurement precision, and estimated developmental trajectories. Moreover, we integrate relevant covariates from the domains of academic achievement, social relationships, and well-being to assess whether substantive conclusions regarding correlational patterns are retained for the shorter operationalizations as well. Drawing on the findings of this study, we offer recommendations on the operationalization of global self-esteem in longitudinal settings, as well as on scale shortening procedures more generally.



• The proxy problem in the assessment of child health: Exploring determinants of parent-child (dis)agreement in self- and proxy assessments of children’s general health status
  Julia Tuppat* (Leipzig University)
  Jacqueline Kroh (LIfBi Bamberg)
  Donata Perakis (LIfBi Bamberg)
  ≡ Abstract

  Measurements of self-rated health (SRH) highlight the proxy problem in assessing subjective health. When parents assess their children’s health on their behalf, systematic discrepancies may arise between parental proxy reports and children’s self-assessments. This is particularly relevant when the choice of informant can affect estimates of social inequalities in health.
  This study therefore examines which sociodemographic characteristics are associated with discrepancies between children’s self-reports and parental proxy assessments of children’s self-rated health, and whether systematic patterns of overestimation and underestimation of children’s health emerge. Based on Shields’ (2000) theory, we expect to find differences by gender, parental education, family structure (single parenthood), and migration background.
  Using data of the NEPS Starting Cohort 1, we examine these relationships through bivariate analyses and regression models. First, the results show a rather low degree of agreement between children’s self-reports and parents’ proxy reports, suggesting that the two reflect different perspectives on children’s health and should not be considered interchangeable. Second, parents are more likely to overestimate their children’s health than to underestimate it. Third, the degree of agreement increases as children’s self-assessments improve. Fourth, there is only weak evidence of systematic differences by sociodemographic characteristics, particularly parental education and, on an exploratory basis, certain parent-child gender combinations. For the remaining characteristics, there is little evidence of systematic variation.
  The findings illustrate that the use of parental proxy reports may affect not only the measurement of children's health but also, potentially, the representation of social inequalities in health.



• Running out of resources? Premature termination due to time-out in the student questionnaire of NEPS Starting Cohort 8
  Jonathan Tönnessen* (University of Bamberg)
  ≡ Abstract

  Within large-scale assessments like the National Educational Panel Study (NEPS), student questionnaires are often administered under tight time constraints, meaning items at its end are systematically at risk of premature termination due to time out. So far, related research has concentrated mostly on competence tests even though student questionnaires supply most of the background variables analysts rely on. This study therefore asks how student- and test-group level factors explain premature termination in the student questionnaire of NEPS Starting Cohort 8 (grade 5).
  Survey Response Theory and Satisficing Theory describe responding as a resource-dependent process which relies on ability and motivation. Cognitive Load Theory and Expectancy-Value Theory further establish two expectations. First, limited cognitive and motivational resources increase the share of not-reached items due to slower speed or lack of engagement. Second, demanding conditions in the separate test groups the questionnaire has been administered in amplify these effects. For the analyses, I use wave 1 of NEPS SC8 (3,732 students nested in 330 testing session groups). I test both hypotheses in clustered two-level linear regression models with random slopes and cross-level interactions. The adjustment set based on a causal graph with the outcome being the proportion of not-reached items relative to the individually filtered questionnaire length.
  The results show 27.7 percent of the analysis sample failed to reach at least one item. Among them, an average of 21.3 percent of the questionnaire remains unanswered. Test groups account for 12.9 percent of the variance in the outcome. Contrary to expectations, cognitive and motivational resources do not show a statistically significant effect. Neither do cultural test group composition nor reported classroom discipline moderate these effects. Among the conditioned confounding variables, gender, migration background and parental socioeconomic status and test group size show statistically significant associations across all specifications.
  Premature termination in student questionnaires thus might be less impacted by individual resources than by compositional and organisational conditions of the testing session. Because the missingness shows a systematic pattern concentrated at the end of the instrument, it should not be assumed as ignorable. Analysts working with wave 1 should treat not-reached items in the questionnaire as a selection problem rather than as incidental noise, and future administration should plan for the conditions that produce them.




• Teacher students’ use of and experiences with Artificial Intelligence and their associations with students’ individual characteristics and topics of study courses
  Nora Heyne* (University of Bamberg)
  Maximilian Pfost (University of Bamberg)
  Engelbert Niehaus (RPTU University Kaiserslautern-Landau)
  Walter Kühn (RPTU University Kaiserslautern-Landau)
  ≡ Abstract

  Nowadays, artificial intelligence (AI) is used in almost all areas of life and, despite some risks, offers many opportunities to improve learning and teaching (e.g., Kasneci et al., 2023). In order to adequately prepare adolescents for the demands of everyday interaction with AI and other digital media, their targeted use in schools is necessary, e.g., as part of reading instruction (KMK, 2017, 2022). Therefore, teacher education needs to prepare (prospective) teachers for these challenges in school practice (KMK, 2024). Although initial findings on the use of and experiences with AI among various persons are available (e.g., Gnambs et al., 2025), little is known about teacher students’ use of and experience with AI, and to what extent AI is addressed in teacher education, which therefore is investigated by the KINELL-study (KI-Nutzung und -Erfahrungen für Lehr- und Lernziele). Therein, in addition to the courses offered in the teacher education, various teacher students’ individual characteristics are expected to influence their use of and experiences with AI. In particular, these characteristics include students’ need for cognition (NFC; Beißert et al., 2014), tolerance of ambiguity (TOA; Kischkel, 1984), and academic risk taking (ART; Hübner & Pfost, 2023), which have previously been shown to be linked to various aspects of learning behaviour and success. Furthermore, it is assumed that the use of digital media—e.g., use frequency (MUF; Drossel, 2024a) and self-assessed media literacy (MUL; Drossel, 2024b)—is linked to teacher students’ AI-use and -experiences.
  Against this background, the study examines the frequency with which AI is addressed in teacher education courses, as well as teacher students’ use of and experiences with AI, taking into account the characteristics mentioned above and academic success. The investigation is based on an online survey in summer 2026 with teacher students (N=95) from the University of Bamberg and the Rhineland-Palatinate Technical University Kaiserslautern, by means of scales for the constructs introduced (NFC, TOA, ART, MUF, MUL).
  As the results show, a minority of students learned about AI in their study courses, while most of them used AI regularly in their studies. Moreover, the students’ AI-use and -experiences correlated with several of the individual and contextual characteristics. For example, students’ past AI use correlated negatively with learning success. Furthermore, expected future AI use correlated positively with AI use in study courses. Results and further implications are discussed, and hence, the study informs further practice and research in teacher education.



• Academic risk-taking in neurodivergence – A quantitative study comparing neurotypical and neurodivergent university students
  Christina Schindler* (University of Bamberg)
  ≡ Abstract

  Theoretical Background: Making errors is often regarded as a necessary step in the learning process (Pösse, Obermeier, Nowak & Gläser-Zikuda, 2025) to consolidate knowledge and foster learning success. This requires students to engage with uncertainty, for example, by engaging with tasks where they feel less confident in their own abilities (Beghetto, 2009), despite the risk of appearing less competent to others. This can be summarized under the construct of academic risk-taking (Clifford, Lan, Chou & Qi, 1989), which has been associated with learning success (Hübner & Pfost, 2025). However, students may perceive uncertainty, errors, and social-evaluative situations in different ways. This may particularly be the case for neurodivergent students (for example Autism Spectrum Disorder or Attention-Deficit/Hyperactivity Disorder), whose cognitive and neurological characteristics differ from those typically considered “neurotypical” (Butcher & Lane, 2025; Durgungoz & Durgungoz, 2025), which may lead to differences in processes such as information processing, risk assessment, and the handling of uncertainty. Addionally, feeling supported by peers or instructors support may provide students with a sense of security, reducing the potential costs of making errors and thereby facilitating academic risk-taking (Pfost & Hübner 2026).
  Objectives: This study therefore aims to examine the extent to which academic risk-taking differs between neurodivergent and neurotypical students. Furthermore, the study investigates whether perceived social support moderates the influence between neurodivergence and academic risk-taking.
  Methods: A sample of 211 university students was analyzed. Neurodivergence status was assessed through self-report (neurotypical, neurodivergent with a confirmed diagnosis, and self-identified as neurodivergent). Analyses of variance were used to determine whether there were significant differences between the three groups. Subsequently, Bonferroni-corrected post-hoc analyses were conducted when significant results were found. For the second research question, linear regression analyses with moderations only including the group of neurotypical students and those with a confirmed diagnosis were conducted.
  Results: Students with a confirmed diagnosis exhibited a higher level of academic risk-taking in front of the whole seminar group than neurotypical students. Conversely, the group of students with a confirmed diagnosis exhibited lower levels of academic risk-taking in peer contexts than neurotypical students. Perceived social support did not moderate the relationship between neurodivergence and academic risk-taking. Conclusions: The findings suggest that academic risk-taking among neurodivergent students is context-dependent rather than generally higher or lower. While seminar group situations may provide greater transparency regarding social expectations and norms, peer contexts may involve implicit social expectations, regarding appropriate participation that could influence their willingness to take academic risks. This research therefore highlights the importance of understanding how neurodivergent students experience learning environments differently, as failing to account for these differences may create or reinforce barriers to their learning and participation.



• Wrong answers welcome? The interplay between error culture, error attitudes, and academic risk-taking
  Vanessa Hübner* (University of Bamberg)
  Maximilian Pfost (University of Bamberg)
  ≡ Abstract

  Theoretical Background: Errors are an inevitable part of learning, and making errors may facilitate learning, for example by creating additional retrieval paths (Pyc & Rawson, 2010). Yet, students often avoid behaviors that involve the possibility of being wrong (Pan et al., 2020). This may be particularly relevant in higher education, where learning requires students to engage with uncertainty, challenge existing assumptions, and develop ideas that are not yet fully formed (Barnett, 2000). Error culture, a multi-dimensional construct encompassing shared beliefs, norms, practices, rituals, and narratives surrounding errors within a certain group (Oser, 1994; Steuer et al., 2013), may shape how students respond to the possibility of making errors. In a constructive error culture, students may feel sufficiently safe to engage in challenging rather than easy learning tasks that involve the risk of making errors, a behavior commonly referred to as academic risk-taking (Beghetto, 2009; Clifford, 1991). At the same time, contextual influences may depend on how well they align with individual orientations, such as students’ attitudes toward errors (Hübner & Pfost, 2022).
  Objectives: The present study therefore examined the interplay between perceived error culture, students’ error attitudes, and their academic risk-taking in higher education from a person-environment fit perspective (Rauthmann, 2021).
  Methods: A sample of 341 university students completed measures of perceived error culture, error attitudes, and academic risk-taking. Perceived error culture was modelled using a latent bifactor-(S-1) approach. Latent regression models were used to examine direct associations. Manifest moderation analyses were conducted to investigate whether students’ error attitudes moderated the relationship between error culture and academic risk-taking.
  Results: Error culture was positively associated with academic risk-taking in peer contexts, even when controlling for students’ error attitudes and background variables. Academic risk-taking in front of the whole seminar group was primarily predicted by students’ error attitudes, with error culture exerting a conditional effect. A more constructive error culture was associated with greater academic risk-taking particularly among students who viewed errors as useful for learning.
  Conclusions: These differentiated findings suggest that error culture can provide a contextual affordance for academic risk-taking, but rather than functioning as a universal driver its relevance depends on the social context, and on students’ individual error attitudes. The study therefore highlights the importance of jointly considering contextual and individual factors when investigating how higher education environments can support students in engaging productively with uncertainty and potential errors.

back to top