정서적 공감척도의 타당화 및 차별기능문항 탐색

초록

The purpose of the present study was to examine the psychometric functions of 33 items of the Emotional Empathy Scale(Mehrabian & Epstein, 1972) using Rating Scale Model(Andrich, 1978). Followings are the three main findings. First, an examination of the person-item target structure for the Emotional Empathy Scale showed a substantial number of items were not closely targeted on the persons(i. e., some items were too easy). This situation was reflected in the person separation reliability. The person separation reliability of 0.65 in this study was considered low. Second, the joint display of the expected and empirical ICC revealed that the data fit the model. If the data fit the model we could expect 19 or more out of 20 x's to be within the 95% confidence interval. The result indicated that most of the dots(x) was located within the 95% confidence interval along the expected score ICC. Third, an examination of the rating scale characteristics showed that thresholds increased monotonically across the rating scale and the item fit statistics indicated that all 33 items were located within the criteria (0.5<fit<1.5). Fourth, the results of DIF(Differential Item Functioning) indicated that almost half items of all 33 items in the analysis of counselor variable were detected as biased items. However, fewer number of items in the analysis of gender variable were detected as biased items. In conclusion, the Korean version Emotional Empathy Scale needs changes in item level to utilize in the future studies. In particular, the low reliability requires a cautious use of this scale.

키워드