Assessing secondary schooling performance by applying multivariate analysis methodologies to the PISA and OECD data

• Mario García Muñoz ^[1] ; Fernando Martínez-Santos ^[1]
  1. [1] Universidad CEU San Pablo

     Universidad CEU San Pablo

     Madrid, España

     
• Localización: BEIO, Boletín de Estadística e Investigación Operativa, ISSN 1889-3805, Vol. 38, Nº. 1, 2022
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • The aim of this paper is to explain the differences between countries in the competency of Science in secondary education. Using the PISA and OECD data of 2016, we have considered three different regression models. We find that the method of ordinary least squares, with a stepwise forward selection of the variables, performs better than the other two regression models obtained through machine learning techniques, these are the support vector machine and the Gaussian regression models. The most sigThe aim of this paper is to explain the differences between countries in the competency of Science in secondary education. Using the PISA and OECD data of 2016, we have considered three different regression models. We find that the method of ordinary least squares, with a stepwise forward selection of the variables, performs better than the other two regression models obtained through machine learning techniques, these are the support vector machine and the Gaussian regression models. The most significant drivers of students’ scores are the GDP per capita and the percentage of young people who are “Not in Education, Employment or Training” (NEET). Moreover, we find that introducing the interaction of both variables has a significant impact on the scores of secondary schooling.nificant drivers of students’ scores are the GDP per capita and the percentage of young people who are “Not in Education, Employment or Training” (NEET). Moreover, we find that introducing the interaction of both variables has a significant impact on the scores of secondary schooling.