The purpose of this study was to examine the construct validity of scores on a Chinese version of an academic self-concept scale as measured by self-, peer, and parent ratings. The multitrait-multimethod approach, general confirmatory factor analysis models, and correlated uniqueness models were utilized. Overall, the results from three analyses supported the convergent validity as well as discriminant validity of scores and the presence of a method effect. However, the evidence for convergent validity, discriminant validity, and method effect associated with scores was somewhat inconsistent. Alternative analysis procedures are recommended to provide safeguards to construct validation.Because self-concept is one of the most important constructs in educational and psychological research, instruments yielding reliable and valid scores are needed for appropriate interpretation in a number of educational settings. Among the myriad measures of self-concept, Dimensions of SelfConcept (DOSC) developed by Michael and Smith (Michael & Smith, 1976;Michael, Smith, & Michael, 1989) is one of the 23 scales recommended by Byrne (1996). The DOSC is widely used for assessing noncognitive factors of academic self-concept. This multidimensional self-concept instrument is available in three forms: Form E for elementary school children, Grades 4 through 6; Form S for secondary school students; and Form H for higher education students. Items in each form represent activities or attitudes toward school learning situations.