Examining the contextual factors of science effectiveness: a machine learning-based approach

ABSTRACT This research intended to identify key contextual factors that synergistically influence high- and low-performing students’ science outcomes by drawing upon a dynamic model of educational effectiveness. The dataset, the Programme for International Student Assessment (PISA) 2015, consisted of 79,963 science scores for secondary students (49,924 high performers at proficiency Level 6 and 30,039 low performers at proficiency Levels 1a and 1b) from 53 countries/economies along with students’ and school principals’ responses to the PISA questionnaires. By applying a support vector machine (SVM) and SVM-recursive feature elimination (SVM-RFE) sequentially, this study successfully (a) identified 30 key factors of the total 127 contextual factors at the school, classroom, and student levels that synergistically differentiate high and low achievers and (b) provided evidence to support the validity of the dynamic model of educational effectiveness by recognizing the multidimensionality of the contextual factors.