The research team, based at the Institute of Educational Sciences in Bucharest and working within the Horizon Europe REVERS-ED project, exploited a resource few countries can offer at this scale: the Romanian Integrated Information System of Education, known as SIIIR, which assigns each student a unique identifier and links school enrollment records with national examination results. Using extracts provided by the Romanian Ministry of Education in November 2024, the researchers compared three birth cohorts—1999, 2004, and 2009—across six transition-specific analyses organized into two overlapping historical windows. The lower-secondary and Grade 9 transitions were compared between the 2004 and 2009 cohorts, while upper-secondary completion, Baccalaureate entry, and Baccalaureate success were compared between the 1999 and 2004 cohorts. Inequality was examined descriptively along four administrative indicators: recorded sex, the Roma administrative marker derived from the nationality field, first language, and the location of the school attended at each transition origin.

The headline cohort gains were substantial. The probability of promoted Grade 8 completion by the expected endpoint rose from 88.87 percent in the 2004 cohort to 92.98 percent in the 2009 cohort, a gain of 4.11 percentage points. Promoted Grade 12 completion within day-school high school increased by 7.16 percentage points between the 1999 and 2004 cohorts, from 86.05 to 93.21 percent. Baccalaureate entry in the examination year of documented high-school completion rose by 2.33 percentage points, and cumulative Baccalaureate success within the first two examination opportunities climbed by 7.60 percentage points, from 74.54 to 82.14 percent. Administrative observability of a subsequent Grade 9 state among promoted Grade 8 completers was nearly universal in both cohorts, at roughly 98 percent, but the distribution across the four parallel Grade 9 pathways—theoretical high school, technological high school, vocational and aptitude-based high school, and the three-year professional vocational route—shifted between cohorts, with professional VET declining by 4.46 percentage points in the later cohort.

Between-cohort changes in inequality were strikingly uneven across indicators and transitions. Female–male differences were smaller in the later cohort at all four favorable milestone endpoints, with reductions ranging from 0.90 percentage points for Grade 12 completion to 3.11 points for Baccalaureate success. The non-Roma–Roma difference narrowed markedly at Grade 8 and Grade 12 completion, by 12.18 and 11.27 percentage points respectively, although sizeable Roma-related gaps remained within the later cohorts. The urban–rural contrast followed a more stage-specific pattern: it changed little for Grade 8 completion, narrowed by 5.53 points for Grade 12 completion and 6.36 points for Baccalaureate success, but widened by 3.36 points at Baccalaureate entry. First-language contrasts were comparatively small and directionally heterogeneous, even reversing sign at some transitions.

The second analytical perspective followed a single linked population of 122,344 students from the 2004 cohort, each with a documented Grade 5 origin and no administrative special-education marker at that baseline, through a progressive stage-based multi-state framework. Rather than multiplying conditional transition probabilities from separate populations, the researchers propagated fitted probabilities through the modeled Grade 9 pathway branches and subsequent transitions, then averaged over the empirical joint distribution of the four Grade 5 baseline indicators using marginal standardization. Uncertainty was estimated by direct person-level nonparametric bootstrap, refitting the complete six-model transition system in each of 2,000 successful replications. The standardized results were sobering: from the common Grade 5 origin, 90.54 percent of students reached promoted Grade 8 completion, 64.59 percent reached documented high-school completion, 60.21 percent entered the Baccalaureate examination sequence, and only 50.72 percent achieved Baccalaureate success across the first four examination opportunities.

Within this linked trajectory, every signed group difference was larger at Baccalaureate success than at Grade 8 completion, although not all gaps evolved monotonically. The female–male difference grew from a negligible 0.67 percentage points at Grade 8 completion to 15.08 points by the fourth examination opportunity. The non-Roma–Roma difference, already a formidable 30.75 points at Grade 8 completion, surged to 50.60 points at documented high-school completion, receded to 48.20 points at Baccalaureate entry and 39.31 points at the first success opportunity, then climbed again to 43.15 points through four opportunities. By that final horizon, the standardized probability of Baccalaureate success was 51.90 percent for students without the Roma marker and just 8.75 percent for those with it. The Romanian–different-first-language contrast widened from 0.79 to roughly 6.2 points across the Baccalaureate horizons, and the urban–rural Grade 5-origin difference ballooned from 2.32 points at Grade 8 completion to 30.39 points at Baccalaureate success, where the corresponding standardized probabilities were 65.90 percent for urban-origin students and 35.51 percent for rural-origin students.

Grade 9 emerged as the system’s great branching point, where quantitative access and qualitative differentiation visibly diverged. Theoretical high school showed particularly large differences by the Roma marker and Grade 5 school setting, at 26.98 and 24.06 percentage points respectively, alongside an 11.78-point female advantage. Professional VET displayed the opposite direction for these contrasts: it was 11.76 points more probable among males, 7.51 points more probable among students with the Roma marker, and 17.38 points more probable among students with a rural Grade 5 origin. The researchers interpret this pattern through the lens of Lucas’s theory of effectively maintained inequality, which holds that when quantitative access to an educational level becomes widespread, advantage may shift toward qualitative distinctions within that level, such as tracks and destinations. They caution, however, that the study does not measure socioeconomic origin and therefore does not constitute a direct test of the theory.

The spatial dimension of the Grade 9 transition proved especially pronounced. Among students whose Grade 8 school was rural and whose subsequent Grade 9 setting was known, roughly 85 percent in both cohorts were observed in an urban Grade 9 school, with urban destinations predominating across all four pathways. For these students, entry into upper-secondary education was frequently a geographical as well as an educational transition. This finding resonates with prior GIS-based research showing that 84.1 percent of Romanian public high schools are located in urban areas although approximately half of the high-school-aged population lives rurally, leaving many rural communities with substantially lower accessibility and fewer high-school options. The authors stress that school setting in their data refers to the location of the recorded school rather than students’ residence, and that the observational design cannot disentangle distance, transportation, school availability, family resources, prior achievement, or residential mobility as mechanisms.

The cohorts also encountered different institutional arrangements that situate the contrasts without explaining them. Upper-secondary admission rules changed across the periods studied, with the weighting of National Evaluation results shifting from 75 percent for the 1999 cohort to 80 percent for the 2004 cohort and to the National Evaluation alone for the 2009 cohort. The three-year state professional VET route was introduced from Grade 9 in the 2014–2015 school year, and the cohorts intersected the COVID-19 pandemic at different educational stages. The authors are explicit that these historical differences cannot be treated as discrete exposures and that the observed between-cohort differences cannot be attributed to particular reforms, admission rules, or pandemic disruption. The administrative records themselves impose further limits: they do not provide complete annual grade-by-grade histories, absence of a later record cannot be interpreted as dropout, the Roma indicator is an administrative classification whose misclassification could not be validated, and the first-language variable is a broad administrative indicator rather than a measure of any specific ethnolinguistic group.

The study’s central contribution is conceptual as much as empirical. Aggregate progression, between-cohort change in inequality, and the development of inequality across stages within a cohort are related but distinct processes, and a single endpoint can obscure convergence at some transitions alongside larger or reconfigured inequalities elsewhere in the trajectory. Among 161 longitudinal studies of K–12 educational inequality in a recent European scoping review, only one used a Romanian sample, making this analysis a rare population-level account from an underrepresented national context. By locating the transitions and pathway junctures at which inequalities change in magnitude or form, the researchers provide a map for subsequent work on the mechanisms that generate these differences and on whether particular interventions can alter educational progression or pathway distributions, for whom, and under what implementation conditions.

Subject of Research: Education

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

Article References: Simionescu, D. P., Dalu, A.-M., Goia, D., Iftode, O., Noveanu, G. N., & Plăeșu, A. (2026). A longitudinal perspective on educational inequalities in Romania: between-cohort comparisons and within-cohort progression across stages and pathways. Frontiers in Education, 11, Article 1924720. https://doi.org/10.3389/feduc.2026.1924720

Image Credits: AI Generated

DOI: 10.3389/feduc.2026.1924720

Keywords: educational inequality, Romania, longitudinal study, educational transitions, multi-state analysis, Roma students, rural education, Baccalaureate, educational pathways, cohort comparison, longitudinal, perspective