This study examines the key predictors of mathematics performance among Grade 10 (age 15–16) Italian secondary school students using data from the INVALSI 2025 national assessment. We investigate how language proficiency, digital competencies (measured via the DigComp 2.2 framework), and demographic factors (gender, socioeconomic status, school type, and region) influence mathematics achievement. Using multiple regression, machine learning, and network analysis, we identify the most significant predictors and their structural relationships. Results confirm Italian proficiency as the strongest predictor, followed by overall digital competence. Machine learning models highlight these factors’ dominance, while network analysis reveals digital competencies form a tightly interconnected cluster, distinct from demographic influences. Students in Northern Italy and academic-track schools (i.e., lyceums) outperformed peers from other geogprahical regions and school typologies. Gender differences persist but are modest compared to socioeconomic effects. Our findings suggest that while language and digital skills are central to mathematics performance, structural inequalities, linked to school type and geography, remain persistent barriers.
v Darjo (Mon,) studied this question.