Beläggsmodellen i svenska nationella prov

Abstract

In the Swedish national tests, the so-called “Beläggsmodell” is used to generate test grades in seven school subjects. This article examines the origin and effects of the “Beläggsmodell” on test scores in one of these subjects, Religious studies. Classical methods for evaluating the quality of tests are used in combination with confirmatory factor analysis. The results show that the tests have a reliability above 0.92 and that the “Beläggsmodell” reduces this in unpredictable ways for individual students. Confirmatory factor analysis shows that the tests are mainly one-dimensional, which has the consequence that the “Beläggsmodell” does not fulfill any meaningful function. Instead, it has major consequences for different students' test grades. There are examples of students who at the same total score on the test receive everything from D to A as a test grade. The large differences arise through differences in performance on a very small number of items. If, instead, the total sum is used as a basis for generating test grades, the possibilities for fair and equivalent test grades increase, as shown by examples. Responsible authorities should as soon as possible consider switching to a system where the total sum points is the basis for the test grades.