AUTHOR=Gutiérrez-de-Rozas Belén , López-Martín Esther , Carpintero Molina Elvira 

TITLE=Defining the profile of students with low academic achievement: A cross-country analysis through PISA 2018 data

JOURNAL=Frontiers in Education

VOLUME=Volume 7 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/education/articles/10.3389/feduc.2022.910039

DOI=10.3389/feduc.2022.910039

ISSN=2504-284X

ABSTRACT=The explanation of underachievement and the search for its associated factors have been of constant interest in educational research. In this regard, the number of variables that have been involved in its description and explanation has increased over the years, as so has the number of studies that have been published at the international level. Although much research has been focused on the identification of the personal, family and school aspects that exert the greatest influence on students' low academic performance, the literature shows the need to study the differential effects of said variables according to the countries in which the studies are conducted. Due to this, the objective of this paper is to analyse cross-national differences in the effect of personal, family and school characteristics on students' academic underachievement based on data derived from the PISA 2018 assessment. Furthermore, it aims to identify the profile that characterises students with low academic performance and to estimate the importance of the selected variables in explaining low achievement across countries. To reach these purposes, the multivariate technique of decision trees through the binary CART (Classification and Regression Trees) algorithm was used, allowing the estimation of both a global model and nine specific models for each of the selected countries. The results show that, although there are slight differences between the countries analysed, the variables that define the general profile of students with low achievement and which have shown the strongest predictive capacity for low performance are mainly linked to the students themselves. These variables are followed in importance by family aspects, which present great differences between the territories that compose the sample. Finally, teacher and school variables have shown to have a low explanatory capacity in this study. Therefore, it can be concluded that, although personal characteristics continue to be those that best explain academic performance, there seem to be a series of contextual variables, especially related with families, that influence academic achievement differentially and that may even hide or annul certain personal characteristics.