A Comparison Between the Rating Scale Model and Dual Scaling for Likert Scales

The fundamental requirement of using Likert to measure affective behaviors-that all must use the ordered response of all scale statements in the same -is examined. Scaling problems arise when statements within a Likert scale are preference data, and when the position spacing of the ordered response categories are differently across scale statements and respondents. The assignment of a neutral category in the center of the response is questionable. These problems pertain to of the existence of an interval scale, the of the trait, and patterns of item functions. An attempt is made to resolve problems using two contrasting scaling -item response theory modeling and dual. The primary result is that conformity of to the item response theory model and targeting of scale statements and ordered response categories to the respondents dual scaling are vital for a resolution of problems. This study establishes the similarity the two scaling methods. Index terms: scaling, items response theory, jackknife technique, scaling, Rasch model, rating scale model.