THE ACCESS TO EDUCATION OF THE HUNGARIAN MINORITY IN ROMANIA. A MULTIVARIATE ANALYSIS Adrian HATOS–Krisztina BERNÁTH University of Oradea str. Universitãþii nr. 1, 410087 Oradea, Romania; e-mail: ahatos@gmail.com Partium Christian University str. Primãriei nr. 27, 410209 Oradea, Romania; e-mail: krisztinabernath@gmail.com Abstract: During the transition the topic of Hungarian higher education in Romania has been very intensely debated as its polemic has been often characterized by ethnicist and nationalist arguments. One major issue of the debate is, however, that in Romania Hungarians have lower chances of access to tertiary education diplomas than the Romanian majority. The present study aims at clarifying the distribution and the covariates of educational and subsequently social chances among citizens of Romanian and Hungarian ethnicity in Romania. Using the consolidated data of the Romanian Public Opinion Barometer of 1998–2004 (N=11040) our multivariate analyses test three models of educational achievement: an ethnic determination hypothesis, a model of regional discrepancies in educational achievement and the basic social reproduction hypothesis. Though bivariate analyses indicate that there is a significant gap in educational chances for Hungarians, measured through educational achievement, the multivariate analysis proves that the thesis of lower educational chances of Hungarians from Transylvania does not hold. However, educational inequalities seem to be more a regional issue: Szeklerland, with an overwhelming majority of Hungarians provides significantly lower chances of attaining a diploma of higher education than the rest of Transylvania. One explanation for this could rely on the modernization deficit of the region. The topic requires further research in order to clarify the impact of the diverse covariates on career decisions and status attainment. Keywords: higher education, social capital, minority, social reproduction, regional differences Review of Sociology 15 (2009) Review of Sociology Vol. 15 (2009) 1, 40–60 * Previous versions of the present material have benefited from the remarks of several colleagues. Special thanks to Professor Dumitru Sandu for his detailed observations referring especially to the elaboration of the multivariate models and statistics we used. Therefore the authors are entirely responsible for every deficiency that would affect the material.