ASSESSMENT IN THE AREAS OF MATHEMATICS AND LANGUAGE THROUGH LATENT TRAIT MODELS

To achieve the goal of quality, assessment is needed to diagnose the strengths, problems and difficulties of an educational system with the subsequent aim of trying to correct and improve them. This work focuses on the study of global competencies in the areas of mathematics and language using latent trait model or Structural Equation Modeling (SEM). SEM techniques are widely used to represent unobserved concepts that need to be inferred through measured items. To perform this study, official global results are collected and we had access to a sample of students from two public schools in Andalusia, whose collaboration was invaluable. After checking the validity of the measurement instruments, a structural model that provides information about the internal structure and relationships between the studied skills is proposed. Finally, results derived are compared to the official ones, evidencing the potential advantages of SEM models