A Likert-type Data Analysis Using the Partial Credit Model

Abstract

This study is about examining the possibility of using the partial credit model to solve several problems that occur when we analyze and interpret Likert-type data by traditional methods. The problems are as follows: (i) scores are not directly interpretable and must be examined in the light of a criterion group; (ii) the absence of a zero point handicaps the direct use of individual scores; and (iii) the adequacy of integer scoring, resting upon the Validity of the assumption of equal distances between response categories, is not often verified. This study shows that the partial credit model (PCM) solves these problems. In addition, the PCM provides several advantages in the analysis and interpretation for Likert-type data (e.g., item response maps, person and item fit statistics). The PCM also might help to implement the computerized adaptive testing for Likert-type scales.