本研究以 Shavelson，Hubner， 及 Stanton（ 1976 ）之多向度階層性理論模式 為基礎，再參酌本土化實證研究結果、以及其他相關論述，編製適合臺灣學生使用的「技職 學生自我概念量表」。 本量表包括數學、語文、專業科目、情緒、外貌、體能、同性朋友、異性朋友、父母關係、 師生關係、問題解決能力、道德誠實、生涯規畫等 13 個分量表，以及一個「一般性自我概 念」分量表，採 Likert 九點形成。先調查技術學院、專科、高職之教師與同學意見，再參 酌相關測驗與學者論述形成預試題本，進行二次預試並作項目分析後編成本量表，全量表共 112 題。 量表之 Cronbach α係數在.74 到.94 之間，四週再測信度係數在.66 到.83 之間，以「田 納西自我概念量表」與洪志成之「數學科自我概念問卷」為效標驗證同時效度， 其 r 係數 在.53 到.80 之間，以上 P 值均小於.001 水準，以驗證性因素分析來考驗建構效度，結果 亦佳，綜合而言，本量表之信、效度均良好。最後，隨機抽取 1229 名技術學院學生為標準 化樣本，建立不同性別、年級等八個常模。 本研究之第二部份，是測試自我概念之高階因素建構。首先以驗證性因素分析驗證「基礎模 式」（亦即 13 個相關聯的初階因素模式），得到相當良好的適度指標，接著再以四種可能 形式之高階因素模式進行分析，並比較其與「基礎模式」之差異，結果顯示四個高階模式雖 獲得不錯的適合度，然而階層性並不強，似乎可能有特定領域之變異數未被充分解釋，有待 進一步研究。

　　　　　This study aimed to develop a suitable inventory for vocational students on the measurement of their self-concept. The Shavelson (1976) model of self-concept was adopted and was revised for Chinese students. Items in this study were constructed based on empirical research results and test items derived from related instruments. The SCIVS was composed of 13 subscales and one "general" self concept subscale. The 13 subscales include the following: Mathematics, Verbal ability, Major subjects, Emotion, Appearance, Physical ability, Peers of the same sex, Peers of the opposite sex, Relations with parents, Relations with teachers, Problem solving, Morality/honesty, and Career planning. Likert's 9-point type of scale was adopted in this inventory. Test development procedures were as follows. Questionnaires and surveys were mailed to teachers and students in the vocational education systems to form the item bank. Two pretests and item analyses were then conducted on the item bank in order to reach the final version of the SCIVS, which consists of 112 test items. The averaged Cronbach α of the SCIVS was.84 (ranging from.74 to.94), and the averaged test-retest reliability coefficient was.76.(ranging from.66 to.83) Concurrent correlations with the "Tennessee Self-concept Inventory" scores were ranging from.53 to.80. The construct validity of the Confirmatory factory analysis was also discussed. In general, the SCIVS showed fairly satisfactory measurement results. Finally, test scores of a random sample of 1229 vocational students were collected, from which eight norms were developed according to student gender and grade. The norm statistics were also presented. The sceond part of this study was to validate the higher-order factor structure revised from Marsh and Shavelson's (1985) work. Four higher-order factor analyses were performed and their resulting fit indices were compared to those of the "basic model", in which 13 first-order factors were correlated and indicated by 3 observed variables respectively. The hierarchy, however, was not strong, and much of the variance in subareas of self-concept was unexplained by the higher-order factors. Researchers are encouraged to explore further on more subareas, especially the academic areas and the physical area, of the self-concept.