A longitudinal perspective on educational inequalities in Romania: between-cohort comparisons and within-cohort progression across stages and pathways

Abstract

Introduction Progression through educational milestones and inequality need not change in parallel: group differences may narrow at one transition, widen at another, or shift into differentiated pathways. Methods Using Romanian administrative records, we compared the 1999, 2004, and 2009 birth cohorts across six transition-specific analyses in two overlapping historical windows. A linked stage-based multi-state analysis followed 122,344 students in the 2004 cohort from a documented Grade 5 origin through Baccalaureate success within four examination opportunities. We estimated model-implied probabilities and probability differences by sex, the Roma administrative marker, first language, and school setting. Results The later cohort had higher probabilities at four milestones: promoted Grade 8 completion [2009 vs. 2004: +4.11 percentage points (pp)], promoted Grade 12 completion within day-school high school (2004 vs. 1999: +7.16 pp), Baccalaureate entry (+2.33 pp), and Baccalaureate success within the first two examination opportunities (+7.60 pp). Group differences changed unevenly: female–male differences were smaller in the later cohort at all four milestones, non-Roma–Roma differences narrowed at Grade 8 and Grade 12 completion, whereas the urban–rural difference narrowed at Grade 12 completion and Baccalaureate success but widened at Baccalaureate entry. Grade 9 pathway composition varied by cohort and group. Within the linked 2004 cohort, standardized probabilities from Grade 5 were 90.54% for promoted Grade 8 completion, 64.59% for documented high-school completion, 60.21% for Baccalaureate entry, and 50.72% for Baccalaureate success through four examination opportunities. Between Grade 8 completion and Baccalaureate success within the first four examination opportunities, the female–male gap increased from 0.67 to 15.08 pp, the non-Roma–Roma gap from 30.75 to 43.15 pp, the Romanian–different-first-language gap from 0.79 to 6.21 pp, and the urban–rural Grade 5-origin gap from 2.32 to 30.39 pp; not all gaps evolved monotonically across stages. Discussion Aggregate progression, between-cohort change in inequality, and inequality development across stages within a cohort are related but distinct processes. A single endpoint can obscure convergence at some transitions alongside larger or reconfigured inequalities elsewhere in the educational trajectory. Stage-sensitive longitudinal evidence can identify where inequalities become most pronounced and guide research on their mechanisms and intervention effectiveness.