Differentiation of Self Inventory - short form: creation and Initial evidence of construct validity

Abstract

This study sought to retain the structure and reliability of the four subscales of the Differentiation of Self Inventory - Revised yet make the instrument more efficient by reducing the total number of items within each subscale. The graded response model was applied to data collected from a sample of college students (n = 355) that was primarily Caucasian (73%), and female (55%).The scale items were chosen based on the item characteristic curves of the graded response model and on the individual item content. A second study provided supportive evidence of concurrent criterion validity for the Differentiation of Self - Short Form, demonstrating relationships with symptoms of depression, state anxiety, trait anxiety, perceived stress, and self-esteem. As expected, the Differentiation of Self - Short Form was positively related to self-esteem and negatively related to symptoms of depression, state anxiety, trait anxiety, and perceived stress. Evidence of convergent validity was supported through a positive relationship between the subscales of the Differentiation of Self Inventory -Short Form and the Level of Differentiation of Self Scale. The Differentiation of Self - Short Form was shown in Study 2 to be more efficient than the Differentiation of Self - Revised. The Emotional Control subscale retained 98% of the internal consistency of the full version subscale with 25% of the items; Emotional Reactivity subscale retained 95% of the internal consistency of the full version subscale with 55% of the items; Fusion with Others subscale retained 92% of the internal consistency of the full version subscale with 50% of the items; and the I-Position subscale retained 97% of the internal consistency of the full version subscale with 45% of the items. The full scale score of the Differentiation of Self - Short Form performed similarly retaining 99% of the internal consistency of the full version full scale score with 43% of the items. Finally, in Study 3, preliminary estimates of four-week test-retest reliabilities were established. The test-retest reliabilities ranged from .72 (Fusion with Others subscale) to .85 (Differentiation of Self - Short Form Full Scale). The use of the DSI-SF is recommended in research applications.