Influence of Contextual Variables on Educational Performance: A Study Using Hierarchical Segmentation Trees

Abstract

The general objective of this study is to explore the relationship between students’ contextual characteristics and their performance in mathematical reasoning (MR) and linguistic comprehension (LC) skills. The census data from the ESCALA (EScritura, CAlculo y Lectura en Andalucía) tests developed by Agencia Andaluza de Evaluación Educativa (AGAEVE) in 2017 were used. These tests are carried out in the second year of primary school in the Autonomous Community of Andalusia (Spain). These data have been analysed through the data mining technique known as segmentation trees, using the CRT (Classification and regression trees) algorithm for each of the skills. This has allowed the detection of the high influence of social and cultural status (ESCS) and familial expectations regarding academic performance in both tests. In addition, it allows us to point out that there are different interactions between contextual characteristics and their relationship to performance in MR and LC. These results have made it possible to establish groups of students who may be at risk of not reaching the minimum required levels. Some characteristics of at-risk students are low ESCS, low family expectations or being born in the last six months of the year. The detection of at-risk profiles could contribute to the optimisation of the performance of these groups by creating specific plans.