Skip to Main content Skip to Navigation
Conference papers

How to predict university performance: a case study of a prestigious Turkish university

Abstract : Turkey's education system is based on a centralized selection system, starting from secondary education level. For post-secondary studies, students are sorted according to their scores in a competitive national exam. By statute, Galatasaray University, a French speaking institution founded in 1992 by Turkey and France, enrolls half of its students from the very best candidates while the other half comes from French speaking high schools (relatively top-ranked). These students must be classified amongst the first 25,000 in the national competition to be admitted as well as passing a competitive exam in French, although their national ranking remains lower than the first group. The first batch has to learn French at university before starting undergraduate studies, whereas French speaking students are entitled to enter the first year directly. Within the public university system, where admission is strictly based on national exam scores, Galatasaray University's differentiated admission scheme can be tested through the respective performance of these two groups of students. Using specific data, we estimate the impact of high school background (public versus private, types of high school) and national exam score on the university performance of students admitted between 1994 and 2011. Although we lack information on family background, the public-private distinction can capture some of the income effect missing in our data. Gender is also taken into consideration to assess the respective performance of females and males. We also use additional controls for selection into graduation (time to complete) and departments. Regional variation is controlled through high school location. If we assume a correlation between initial academic level and final grades, we can measure the trade-off in terms of total academic output linked to the recruitment of French-speaking students through a less demanding specific competition versus the highly selective process of the national competition. Finally, we evaluate the validity of Turkey's national exam in sorting students by ability.
Document type :
Conference papers
Complete list of metadata
Contributor : Bertille Theurel Connect in order to contact the contributor
Submitted on : Friday, December 9, 2016 - 3:41:12 PM
Last modification on : Monday, November 22, 2021 - 9:58:01 AM


  • HAL Id : halshs-01413264, version 1



Jean-Jacques Paul, Sezgin Polat. How to predict university performance: a case study of a prestigious Turkish university. XXV Meeting of the Economics of Education Association, AEDE (Association of Education Economics), Jun 2016, Badajoz, Turkey. ⟨halshs-01413264⟩



Record views