For years, about 20% of primary school graduates in Luxembourg repeat a grade – an internationally high share, raising concerns about the implications for their future development. Two studies are related: Klapproth et al. 2016 apply propensity score matching to assess the effects of grade retention for Luxembourgish secondary students. Vandecandelare et al. 2016 apply propensity weighting to do so for primary students in Flanders. However, while the first study focuses on secondary education and the latter limits to Mathematics, the present study attempts to estimate the causal effect of grade retention on different academic domains and psychosocial wellbeing in primary. It sheds light on the beginning of a child’s educational biography and the role of grade retention in managing heterogeneity. The study draws back on the ÉpStan panel that assess key school competences of each student in Luxembourg every two years beginning in grade 1 and is enriched by background information, allowing unique tracing of individual trajectories. The sample comprises 3,801 grade 1 students in 2015 out of which 545 were retained until grade 3, the rest promoted. First, missing data is imputed by nonparametric missing value imputation using random forest (Stekhoven, Buehlmann 2012). Second, the study establishes comparability for a wide range of student grade 1 baseline characteristics. Multiple gradient boosted logistic regressions and diagnoses identify the propensity score weights with the best balance (McCaErey et al. 2004), resulting in two nearly identical groups differing only in grade 3 repetition. Reweighting instead of matching allows to use the full sample of the 3,801 students. The ATE of retention on achievement (Mathematics, German, French) and wellbeing (general and domain specific self-concept, interest and anxiety) in grade 5 is assessed. Findings indicate: The effect of retention on achievement is significant and positive for Mathematics, negative for German and French reading comprehension. Regarding wellbeing, grade retention significantly raises anxiety in Mathematics, lowers self-concept in general, in German and school interest in general. Grade retention in primary education shows mixed effects for achievement and predominantly negative effects for psychosocial wellbeing. Implications of the findings shall be discussed against the background of increasing heterogeneity among Luxembourgish students stemming from migration and home languages (foreigners share 2001: 37%, 2024: 53%). Increasing linguistic heterogeneity combined with consistently high grade repetition shares, leading to lower language achievement and lower psychosocial wellbeing, raise need for discussion. The results may be of interest for countries facing increasing migration.

McCaErey, D., Ridgeway, G. & Morral, A. (2004). Propensity Score Estimation with Boosted Regression for Evaluating Adolescent Substance Abuse Treatment. Psychological Methods, 9(4), 403-425. Klapproth, F., Schaltz, P., Brunner, M., Keller, U., Fischbach, A., Ugen, S. & Martin, R. (2016). Short-term and medium-term eEects of grade retention in secondary school on academic achievement and psychosocial outcome variables. Learning and Individual DiEerences, 50, 182-194. Stekhoven, D.J. & Buehlmann, P. (2012). MissForest - nonparametric missing value imputation for mixed-type data. Bioinformatics, 28(1), 112-118. Vandecandelaere, M., Vansteelandt, S., De Fraine, B. & Van Damme, J. (2016). The eEects of early grade retention: EEect modification by prior achievement and age. Journal of School Psychology, 54, 77-93.