Science Education and Student Results in Spain: A Multivariate Approach

Abstract

Science education is considered one of the key elements for the development of skills and competencies for the economy and society of the 21st century. Beyond its economic dimension, the quality of science teaching in compulsory education has a fundamental civic dimension: in the new context, it is necessary to adopt scientific reasoning and approaches and be able to position it in the face of political proposals whose evaluation will quite often require future scientific knowledge. In this context, for the first time PISA 2015 provides the results of all the Spanish autonomous communities on statistically representative samples at the regional level. This allows a territorialized diagnosis to be made on different aspects related to science education. In this work, we have identified a set of variables related to instruction whose influence on student performance in Sciences has been significant. For this reason, a multiple linear regression model has been used. This model has first been applied to the national sample and then to the samples from the seventeen autonomous communities. Moreover, based on PISA 2015 data, a dependent variable of a dichotomous nature has been defined, centred on predicting if 15-year-old students will or will not have STEM professions at 30 years of age. An analysis of Multivariate Binary Logistic Regression has been carried out at both the national and autonomous community level. The discussion of the results provides some reflections of interest for educational policies and teaching practices.