本研究目的在使用後設分析以統合國內合作學習相關閱讀教學研究，經對照國內、外相似主題之研究統合文獻，了解彼此存在的一致或不一致性，為研究結果提供更穩定的聚斂或區別效度，而研究者以數個關鍵字就國內資料庫進行檢索，經對照納入與排除準則後，就93篇文獻進行訊息編碼與後設分析。整體而言，分析結果及相關文獻的對照，顯示此教學模式對於閱讀理解的立即及延宕成效大致具有介於小至中等程度的教學成效，其中，以交互教學同時具有較穩定的研究證據支持，而在調節因素探討方面，本研究發現兩組間前測表現差異校正與否是影響彼此統合結果不一致的關鍵。最後，研究者於文末提出幾點建議供未來實務工作者、研究者參考。

Reading instruction has evolved into more interactional and active forms in recent decades, and past evidences from meta-analyses showed that students comprehend more in reading when they learn together than when they learn alone. But some conflicting results across different meta analyses regarding the same theme were sometimes ignored, and this may also lead to be overconfident on the effect of instruction. The first goal of this study is to meta-analyze the effect of cooperative model of reading from Taiwan articles. The second goal is to compare the results derived from this study to findings from other previous relevant review studies. The procedure through this research was undertaken via systematic review. After few keywords were made in searching Taiwan database, 93 articles were finally coded and meta-analyzed on the basis of five inclusion and exclusion rules. Overall speaking, an average effect size between lower and medium level was explored in my studies. Furthermore, moderator analysis showed quality of studies or adjusted for pretest difference or not may have impact on the inconsistency between past meta analyses and this study. Finally, some suggestion and implication were proposed for future usage.