此研究旨在探詢影響本文作者一個行動研究結果的可能因素。該研究，雖然一如假設，證實了［建設性論辯模式］運用在不同程度英語教學上之成效，遠優於「尋求共識模式」，卻也發現了此一經作者參考無數研究，專為EFL學習者研發的教學模式，與一般類似研究結果有所出入。此教學模式在該研究中顯示嘉惠最大的竟然既非高程度也非中高程度的學習者，而是中低程度的學習者。即使是低程度者也多所受益。 為了進一步了解造成這些結果背後所隱含的機制，同樣一批來自不同科系包含112位分四等級的五班學生給予授後問卷調查。每份問卷由三十個問題組成，其後並附有自由表達意見的空間。這三十個問題，首先透過因素分析，集結成八個因素，繼之，運用Wilcoxon Signed Ranks Test及Mann-Whitney Test檢視可能導致該研究結果的重要成因。至於由學生自由發揮的意見，在分類後僅作為輔助材料用。 分析後發現受測者整體的正面態度可能導致他們全面的較優表現，而最可能影響中低程度的學習者表現優於高程度及中高程度的學習者的重要因素，除了此教學模式本身的設計，就是學習者對於建設論辨特質所抱持的態度。分析也發現低程度者可能受限於語言能力，對自己高層次的認知能力缺乏信心，因而在需要較高層次語言及認知能力的「反辯」上表現不佳，但在其他測試項目上，他們顯著的積極態度，可能促使他們在其他方面表現優異。此外，透過分析學生自由發揮的意見，得以推斷出更多研究前未預料到的因子。最後，本研究尚提供了相當多在教學及未來研究可加強或改進的地方，期使此教學模式能誘發更多研究並使其能更有效地運用在英語教學上。

This study in investigated the variables that were likely to have influenced the outcomes of a previous action research of this author's. The research, on the one hand, confirmed, as hypothesized, that the overall efficacy of “EFL Constructive Controversy Procedure” exceeded that of “Concurrence Seeking”. On the other hand, its further finding contradicted what was generally documented in related studies: more proficient learners benefit more greatly from such an approach than less proficient ones. In this author's study, the greatest beneficiaries are the low-intermediate (LM) learners, instead of the high (H) or high-intermediate (HM) learners. Even low-level (L) learners greatly benefited. For a better understanding of the embedded mechanism, the same four levels of students, including 5 classes of 121 students across multiple disciplines, were given a post-instruction survey. The survey consisted of 30 questions and a space for free comments. The 30 questions were further extracted into 8 factors. Then Wilcoxon Signed Ranks Test and Mann-Whitney Test were adopted to examine the impact of the eight factors. The comments were categorized, and mainly used a supplement. The results indicated that an overall positive attitude was correlated with overall better performance, and that the learning mode itself together with a significantly positive attitude towards constructive controversy attributes could have resulted in the LM learners' surpassing other more proficient counterparts. The results also revealed that while L learners might have done better in many areas they did not do so in ‘language’ and ‘refutation’. Lastly, certain unexpected factors may be infered from the free comments.