合作學習對學生學習成效影響之後設分析－以2005至2012年之學位論文與期刊為範圍

Abstract

本研究採用後設分析法探究合作學習對學生的學習成效有何影響，並進一步探討影響學習成效之變項，並分析其對合作學習的學習成效之影響。 本研究蒐集2005年至2012年及以國小、國中、高中職或大學為研究對象之合作學習相關文獻，經研究報告的七項納入準則進行篩選，最後篩選出76篇進行編碼。本研究採用Hedges 與Olkin（1985）的技術並選定固定效應模式進行後設分析運算，此外，以同質性考驗分析各研究結果之間的顯著水準和效果量大小是否有顯著差異存在，同時也採用Rosenthal（1991）建議 Fail-safe N的方法，檢驗出版偏誤對研究結果之影響。 研究結果顯示合作學習可提升學習成效，其中認知學習和創造力等成效之效果量為0.36、0.33，達到小至中度的效果；而技能、情意、社會行為等學習成效的效果量分別為0.51、0.55、0.63，三者皆達到中度以上之效果。 由於學習成效之效果量間存有異質性，故進一步探討調節變項，經同質性考驗後發現，研究樣本屬性類的四個變項（研究發表類型、學習階段、學習領域、樣本總人數）、教學實驗設計類的五個變項（合作學習的模式、教學實驗期間、每一小組人數、控制組的教學方法、實驗組的教學情境）、學習成效之評量工具類的兩個變項（評量工具產生的方式、評量工具之效度），對學習成效具有調節作用。最後提出建議予實務工作者及未來研究者參考。

A meta-analysis was used to explore how cooperative learning affects students’ learning results in the research, furthermore, to discuss whether the variables such as the character of the research report, the design of the teaching experiment and the evaluation of learning effects adjust learning effects. The relevant references of the cooperative learning based on elementary school, secondary school and college students from 2005 to 2012 were mainly collected. Based on seven inclusion criteria of the study and finally, 76 studies were selected to code. Hedges and Olkin’s (1985) methods were used and the fixed effect model was chosen in this meta-analysis. In addition, the heterogeneity test was used to analyze whether significant levels exist among each study. Meanwhile, Fail-safe N method that Rosenthal (1991) suggested was used to examine how the publication bias influence the research results. The research result indicated that cooperative learning could raise learning effects and the effect size of the cognitive and creative learning effect were 0.36 and 0.33 which achieved the small effects to the medium ones. The effects size of the psychomotor, the affection, the social behavior learning effects were 0.51, 0.55 and 0.63 which achieved the effects above medium. Due to the diversity of the learning effect size, the moderators would have a further discuss. According to the heterogeneity test, the four variables of the sample properties, the five variables of the teaching experiment design, and the two variables of the learning effect examinations, had significant impacts on the learning effects. Finally, some suggestions were provided to the practical workers and researchers in the future.