Abstract:

While many theoretical discussions continue on how to name the mid-point label of Likert-type scales, the existence of a small number of empirical studies on this subject is striking. This study aimed to examine the psychometric properties of three verbal forms with varying mid-point labels, and one with end anchored. For this purpose, data were collected from 377 university students with four different scale forms measuring the same feature that consisted of the same items with different attitude labels. The study used a 14-item short form of the Mathematical Attitude Scale which has a unidimensional structure. The data obtained in this descriptive and relational research were tested in terms of validity, reliability, and measurement invariance. The results indicated that the explained variance rates of the four forms were close and the forms had the same and high level of reliability. According to Confirmatory Factor Analysis results, Form 1 with the mid-point label “no idea” was in better compliance with the data. Lastly, the forms only provided structural invariance. The lack of metric invariance shows that item factor loadings differ across the forms. Based on this result, it can be said that individuals perceive different attitude labels in the same category and different type scales in different ways. Within the scope of these findings, researchers can repeat similar studies based on generalizability theory and/or item response theory to put empirical evidence in the field.

Keywords: