探索斯騰伯格智力三元能力測驗、一般智力測驗以及學業成就之間的關係

Exploring the Relationships between the Sternberg Triarchic Abilities Test (STAT)%2C the Test of General Intelligence%2C and Academic Achievement

本研究修訂斯騰伯格智力三元能力測驗(STAT)，並將此測驗對200位國中二年級學生進行施測，同時蒐集這些學生的一般智力測驗成績與學期學業成就資料，並針對它們與本測驗結果之間的關係加以探討。本研究雖涵蓋測研編製歷程中的效度研究，但重點主要在於修訂後測驗的應用研究。研究結果顯示：STAT分測驗(分析、實用、創造)分數、STAT總分、一般智力測驗分數與學業成績彼此間有顯著的正相關；三種智力分測驗中，分析部分與智力測驗成績以及學業成績的相關為最高；在排除智力的解釋力後，分析部分、實用部分、創造部分、STAT總分皆能有效解釋學業成績的變異量；在對分析部分、實用部分、創造部分、智商與學業成績進行因素分析後，僅抽離出一個特徵值大於1的因素；本研究進一步以學業成績為效標變項，智商、分析部分、實用部分與創造部分為預測變項，進行逐步多元迴歸分析，結果僅留下分析部分與實用部分。本研究發現在這些研究結果中，有些支持斯騰伯格的理論觀點，有些卻與其相左。

In this study， the Sternberg Triarchic Abilities Test (STAT) was modified and the modified test was administered to 200 8th graders. Meanwhile， these students' scores on the general intelligence test and the academic achievement of a whole semester were also collected. The relationships between these scores and the outcome of the modified STAT were then carefully examined. The focus of this research study was thus on the applications of the revised measure. Results are shown as follows. STAT analytical， practical， and creative subtest scores， total score， the score on the general intelligence test， and the academic achievement were all significantly inter-correlated. Of all the three subtests， the analytical subtest had the highest correlation with the general intelligence test and the academic achievement. With the variance due to the general intelligence test removed， STAT analytical， practical， and creative subtest scores and total score all can explain the variance of academic performance substantially. A factor analysis including STAT analytical， practical， and creative subtest scores， the score on the general intelligence test， and the academic achievement indicated that only one factor whose eigenvalue was larger than 1 emerged. A stepwise multiple regression analysis with academic achievement as a criterion variable and with IQ， analytical subtest， practical subtest， creative subtest as predictive variables indicated that analytical and practical subtests were most powerful predictors. Some results supported Sternberg's claims but others did not.