School tracking in Luxembourg: the longitudinal impact of student characteristics and school composition

Research question: The current study aimed to investigate the influence of student and school level factors on school tracking in secondary education. We were especially interested in the association between student characteristics and school composition in Grade 3 and school track in Grade 9.
Data source: Data were collected as part of the Luxembourg school monitoring programme “Épreuves Standardisées” (ÉpStan; Fischbach et al., 2014). The study cohort include all students enrolled in the Luxembourg public education system in Grade 3 in November 2013 combined with data from the same students in Grade 9 in November 2017-2019 for students following advanced or regular educational pathways, completed with data from November 2020 and 2021 for students that repeated once or twice (N≈3600).
Theoretical approach: The study draws upon theoretical frameworks and empirical findings (e.g., Boudon, 1974; Bourdieu, 1984), that have demonstrated students´ socio-demographic characteristics are associated with (dis)advantages for specific groups of students in education systems as well as more recent work focusing on school composition (e.g., Baumert et al., 2006), especially as tracked school systems are known to be prone to social segregation (e.g., Hadjar & Gross, 2016). To date, most research on school segregation in tracked education systems such as Luxembourg has focused on individual student´s characteristics. However, with increasing heterogeneity of student cohorts and known differences in educational opportunities related to the social and ethnic composition of the school’s student body (e.g., Thrupp et al., 2002), the current research extents the existing literature by considering both individual (including prior academic achievement and socio-demographic characteristics) and school level factors (mean academic level and percentage of students from lower socio-economic and migration background) in predicting school track placement.
Main findings: Results of a multilevel random effect logistic regression analysis in which we estimated marginal effects on the probability to be placed in the highest, middle or lowest track in Luxembourg show that even after controlling for student´s academic achievement, track placement is affected by the gender and socio-economic background of the student, whereby boys and students from low SES families have less chance to be placed in the highest track. The association with socio-economic background is not only visible on the student level but also on school level, whereby students attending primary schools with a higher percentage of low SES families have less chance to be orientated to the higher track compared to the middle track, regardless of the student´ individual academic performance.