Sexual Orientation-Based Bullying at Apprenticeship Start Predicts Swiss Training Dropout

Apprentices in the Swiss vocational training system who reported being victimized because of their sexual orientation at the very start of their training were significantly more likely to disappear from their class registers the following year, according to a longitudinal study published in Empirical Research in Vocational Education and Training. The finding held even after the researchers accounted for assigned sex, the gender-typing of the training field, and whether students concealed their sexual orientation at school. In contrast, victimization based on sex showed no significant association with training discontinuity at either of the two time points examined.

The study, led by Lavinia Gianettoni and colleagues, followed a representative cohort of 1,370 first-year apprentices enrolled in six vocational training centers in the canton of Geneva. Data were collected across three annual waves between 2020 and 2023, with questionnaires administered between November and February of each year. At the first wave, apprentices had a mean age of 18 years, and 97 percent were between 14 and 26 years old. Response rates declined over time, from 82 percent in the first wave to 70 percent in the second and 61 percent in the third. The study included only learners enrolled in three- or four-year programs leading to a Federal Certificate of Capacity; students in two-year programs were excluded.

Training discontinuity, the outcome of interest, was defined conservatively as absence from the class registers in the following academic year. At the second wave, 18 percent of apprentices were absent from the registers, indicating discontinuity, and this proportion stood at 15 percent among the remaining sample of 1,122 at the third wave. The authors stress that this measure does not distinguish among contract termination, grade repetition, delayed progression, or continuation of training in another canton or institution, so it captures instability in training trajectories rather than formally verified premature termination.

Victimization was measured with a combined indicator covering four forms of violence: physical, verbal, psychological, and online. Apprentices were asked whether they had ever been physically assaulted, verbally attacked, targeted with negative rumors or lies, or subjected to rumors through social media or electronic communication because of their sexual orientation, with parallel questions referring to experiences occurring because of their sex. Response categories ranged from never to several times every week, and participants were classified as having experienced victimization if they reported at least one form. At baseline, 11 percent of apprentices reported at least one experience of sex-based violence since the beginning of their apprenticeship, rising to 19 percent at the second wave and 20 percent at the third among respondents with available data. Sexual orientation-based victimization was reported by 5.5 percent at the first wave, 8.8 percent at the second, and 10 percent at the third.

The analytical approach used discrete-time survival analysis with logistic regression, reshaping the data into a person-period format in which each participant contributed one observation per time point until discontinuity occurred or the observation period ended. Because the dataset contained substantial missing data, analyses were conducted on 15 multiply imputed datasets generated with predictive mean matching, and results were pooled according to Rubin’s rules. The imputation model included all study variables, and convergence diagnostics such as trace plots and the potential scale reduction factor indicated satisfactory convergence. Notably, missingness was itself related to the phenomenon under study: apprentices who had experienced sex-based or sexual orientation-based violence, men, and those in gender-mixed training fields were significantly overrepresented among cases with missing data, and data collection could not continue once apprentices had left their class, producing structurally missing observations at later waves.

The central result concerned timing. Sexual orientation-based victimization reported at the first wave was significantly associated with being absent from the class registers at the second wave. In the first model, the coefficient was 0.82 on the log-odds scale, with a 95 percent confidence interval from 0.05 to 1.59 and a p-value of .04. The association remained significant after controlling for administrative sex and the sex-typing of the training field, with a coefficient of 0.81, and after further adjustment for concealment of sexual orientation, with a coefficient of 0.75 and a 95 percent confidence interval from 0.08 to 1.42. By contrast, sexual orientation-based victimization reported at the second wave was not significantly associated with discontinuity at the third wave in any model, and sex-based victimization at either time point showed no significant association with subsequent discontinuity. Control variables, including assigned sex, sex-typing of the field, and concealment, were not statistically significant in any model.

The authors interpret the pattern through the lens of timing in vocational trajectories. Drawing on the life course perspective, they suggest that entry into apprenticeship is a critical transition period during which roles, expectations, and commitments have not yet stabilized, making apprentices particularly receptive to contextual signals about their fit and long-term viability in the training environment. Experiences of sexual orientation-based victimization at this early stage may function as turning points that increase the likelihood of subsequent discontinuity. The absence of a significant effect at later stages may reflect the consolidation of trajectories, or it may reflect selection: apprentices who remain in training at the second wave may constitute a more resilient subgroup, either because they have developed effective coping strategies or because those most affected by victimization have already exited the system.

The divergence between the two forms of victimization also invited interpretation. The authors note that sex-based discrimination in education has gained substantial public, institutional, and political visibility in recent decades, particularly through debates on gender equality and women’s underrepresentation in male-dominated fields, which may make claims related to sex-based victimization more readily acknowledged and addressed. Sexual orientation-based victimization, they argue, remains comparatively less visible and less consistently recognized within vocational training settings, potentially limiting access to formal and informal sources of support. On this reading, the absence of a significant association for sex-based victimization may reflect buffering factors rather than an absence of discriminatory experiences, an interpretation the authors emphasize requires empirical testing in future research.

The study sits within a broader evidence base suggesting that sexual orientation-based victimization undermines students’ attention and motivation and is associated with higher rates of truancy and disengagement. Prior longitudinal work has found that non-heterosexual youth report more excused and unexcused absences than heterosexual peers, and analyses of U.S. longitudinal data have linked lower high-school graduation and college attendance rates among non-heterosexual youth to experiences of victimization and homophobic school climates. The new findings extend this literature into the apprenticeship period, a stage the authors note had not previously been examined longitudinally for both sex-based and sexual orientation-based victimization in relation to training discontinuity.

Several limitations qualify the conclusions. The set of control variables was limited by data availability, and factors commonly included in models of educational persistence, such as prior academic performance, socioeconomic background, mental health, and family resources, were not available, which limits causal interpretation and internal validity. Concealment of sexual orientation was measured only in relation to the school context, although apprentices alternate between school and workplace settings, and self-reported measures may underestimate the prevalence of sexual orientation-based victimization, particularly in strongly male-dominated fields where heterosexist norms may discourage disclosure. The study also did not include measures of victimization based on gender identity or gender expression. The authors suggest that efforts to reduce training discontinuity may benefit from greater attention to sexual orientation-based victimization at the beginning of the apprenticeship pathway, and that inclusive policies, supportive adults, and explicit anti-discrimination measures may help buffer the negative effects of victimization on student engagement. With roughly one quarter of apprenticeship contracts in Switzerland terminated prematurely according to national data, the findings underscore that the earliest months of vocational training may be a decisive window for prevention.

Subject of Research: The longitudinal association between sex-based and sexual orientation-based victimization and training discontinuity among Swiss apprentices

Article Title: Impact of sex-based and sexual orientation-based victimization on training discontinuity among swiss apprentices: a longitudinal study

Article References: Gianettoni, L., Blondé, J., Guilley, E., & Lacalamita, M. (2026). Impact of sex-based and sexual orientation-based victimization on training discontinuity among swiss apprentices: a longitudinal study. Empirical Research in Vocational Education and Training, 18(1), Article 7. https://doi.org/10.1186/s40461-026-00211-0

Image Credits: AI Generated

DOI: 10.1186/s40461-026-00211-0

Keywords: vocational education and training, apprenticeship, Switzerland, sexual orientation-based victimization, sex-based victimization, training discontinuity, dropout, longitudinal study, minority stress, Geneva, logistic regression, school climate