Importancia de algunos supuestos «psicométricos» en la calibración y equiparación de las pruebas usadas en la evaluación de los sistemas educativos. Estudio del caso de los «Estándares Nacionales» de México

Authors

José Luis GaviriaSalto de línea Covadonga Ruiz de MiguelSalto de línea Universidad Complutense de Madrid.

Abstract

Importance of Specific ‘Psychometric’ Assumptions in the Calibration and Comparison of the Tests Used for the Assessment of Education Systems. Mexico's ‘National Standards’ CaseSalto de línea Salto de línea The study of the evolution of students' performance through time seems to be especially significant in the assessment of national education systems. In order to make comparisons, tests of different years or different forms of the same year should be put on the same level, using for this purpose suitable ‘psychometric’ models. These models assume certain assumptions, and in the case of not being taking into account, they may introduce variations in the results which do not correspond to the variation of students' performances, but they turn to be a kind of design's artful devise. In this article we explain how the design and estimate procedures of individual scores, as well as variations in the dimensional complexity of the tests, may affect comparisons, both horizontally and vertically, thus giving rise to relevant inconsistencies in the results. As an example, this article analyses the specific case of the Mexican programme on ‘National Standards’.

Keywords

Education system assessment, item response theory, comparison assumptions, factorial structure, ‘one-dimensionality’ assumption.