Variation in Likert scale of the mathematics usefulness test

Abstract

The aim of this work was to compare the results obtained in validity and reliability studies of a mathematics usefulness scale when the Likert format of the items varies. The test measured beliefs about the importance attributed to mathematics for career progress and future professional development. The 939 Psychology students (81% female) who participated responded to the items using 3, 5 and 6 categories. The effect of the order of exposure of individuals to each format was controlled and other tests were also included to reduce answer memorization. Likert scales with more categories increased test reliability at the extreme levels of the trait, but at the expense of compromising the internal structure validity evidences (Confirmatory Factor Analysis and Partial Credit Model of the Item Response Theory). The relative efficiency function revealed that similar information is obtained for all levels of the trait when using 5 or 6-point scales. The number of Likert categories did not substantially affect the relationship between usefulness and other variables.

Keywords

• psychological tests
• psychometrics
• mathematics education
• university students