The effects of educational systems, school-composition, track-level, parental background and immigrants' origin on the achievement of 15-years old native and immigrant students : a reanalysis of PISA 2006

Abstract

The effects of educational systems, school-composition, track-level, parentalbackground and immigrants’ origins on the achievement of 15-years old native andimmigrant students. A reanalysis of PISA 2006.The main research question of this paper is the combined estimation of the effectsof educational systems, school-composition and track-level on the educationalachievement of 15-years-old students. We specifically focus on the effects of socioeconomicand ethnic background on achievement scores and to what extent theseeffects are affected by characteristics of the school, track or educational system thesestudents are in. In doing so, we examine the ‘sorting’ mechanisms of schools and tracksin highly stratified, moderately stratified and comprehensive education systems. Weuse data from the 2006 PISA (Programme for International Student Assessment) wave.Compared to previous research in this area the main contribution of this paper is thatwe explicitly include track-level and school-level as separate units of analyses, whichleads to less biased results of the effects of characteristics of the educational system.The results highlight the importance of including track-level and school-level factorsin the debate of educational inequality of opportunity for students in differenteducation contexts. The findings clearly indicate that the effects of educationalsystem characteristics are flawed if the analysis uses only a country and a studentlevel and ignores the track- and school-level characteristics. Moreover the inclusionof the track-level is necessary to avoid overestimation of the school-compositioneffect, especially in stratified educational systems. From a policy perspective, the mostimportant finding is that educational system are not uniformly ‘good’ or ‘bad’, butthey have different consequences for different groups. Some groups are better offin comprehensive systems, while other groups are better off in moderately or highlystratified systems.