Multivariate mixed models for assessing equity and efficacy in education. An analysis over time using EU15 PISA data

This work investigates upon the determinants of students’ literacy in mathematics and reading by analysing data collected in EU15 countries in three different waves (2006, 2009, 2012) by the Program for International Student Assessment ran by OECD. The main aim of our analysis is: (i) to assess the associations between students’ skills in mathematics and reading and their socio-economics and cultural background and to investigate how these associations vary across countries and waves; (ii) to suggest value-added measures of performance at the different levels of analysis in the time span considered; (iii) to advance a system of indicators at country level which allows to investigate the performance of different countries in terms of efficacy and equity in the tree waves. A 4-level multilevel regression model with a bivariate latent structure and random slopes at country level and school level has been adopted. Implications in terms of trends of efficacy and equity in comparisons across countries are discussed.