Risky decline? Exploring the determinants of pupil's proficiency development over time

The early identification of at-risk students and of factors influencing proficiency development are crucial in education policy. This study investigates how students' fluctuations in academic performance during primary school impact the probability of being at-risk in Mathematics and Reading by the end of lower secondary school. Leveraging longitudinal data from 278,765 students in Italy, this study employs a multilevel multinomial model with Markov chain Monte Carlo (MCMC) techniques to explore personal, class, and school-related factors steering shifts between proficiency and non-proficiency. Findings highlight the importance of these transitions in predicting at-risk students at grade 8, emphasizing the role of peers' academic achievements and the variation across schools in shaping proficiency development.