本研究取自 TIMSS 2003中的台灣資料，以課堂活動為外衍變項，利用結構方程式模型探討影響我國國二學生數學興趣偏低，卻依然展現亮眼成就之背後可能原因。再者，為使本研究所建立的模型具有模型穩定之證據，因此，將有效樣本隨機分割為建模樣本與驗證樣本兩組，以進行最終模型的交叉驗證，研究有效樣本共計3,929位。
研究結果顯示，本研究所建立之模型具有良好的交叉驗證效度，建構式教學對能力知覺與數學有效性具有直接且正面的影響，並透過能力知覺與數學有效性間接影響數學學習興趣與成就；教師中心式教學則透過能力知覺間接影響數學有效性與數學學習興趣，並對數學成就有直接正向影響。換句話說，建構式教學對於促進學生能力知覺、數學有效性與數學學習興趣等心理特質較具有正向影響力，而教師中心式教學則是對增進數學成就較具影響力，兩者間各有優劣。最後，本研究亦根據研究結果，提出在教育實務應用與未來繼續研究上的建議。

The data of this study were retrieved from TIMSS 2003 database. We took the clas-sroom activities as exogenous variables, and used the structural equation modeling (SEM) to explore why Taiwan eighth-grade students showed lower learning interests but outstanding performances inMathematics. Afterwards, through cross-validation the present study had verified the model stability. The valid samples were 3,929 eighth-gra-de students.
According to the results, the model that we constructed showed a well cross-validity.Constructivist instruction had direct and positive effect on ability beliefs and mathemat-ics usefulness. Teacher-centered instruction had indirect effect on mathematics usefulness and mathematics learning interests through ability beliefs and had a direct and positive influence on mathematics achievements. In other words, constructivist in-
struction could improve these positive psychometric characteristics that including abil-ity beliefs, mathematics usefulness, and mathematics learning interests. Teacher-cen-tered instruction could advance mathematics achievements. There were varied features in different pedagogies. Finally, according to the findings, some suggestions for future
applications were proposed on the basis of the discussions provided in this study.