工作記憶對學業表現的預測力：臺灣樣本的後設分析   免費試閱

The Prediction of Working Memory on Academic Performance: A Meta-Analysis of Taiwanese Samples

本研究採用後設分析，檢驗工作記憶對學業表現的預測力，並探討此預測力是否有發展及學科差異。以在臺灣發表的學位論文及期刊論文為樣本蒐集對象，計有12篇期刊論文、34篇碩士論文與7篇碩士在職專班論文及5篇博士論文符合篩選標準，共獲得到250筆效果量。先採用固定效果模式計算平均效果量，若未能通過同質性檢定，再以隨機效果模式計算效果量。以z 考驗檢定不同條件下效果量的差異顯著性。本研究發現，工作記憶及短期記憶皆能顯著預測學業表現，但納入本研究分析的效果量具有出版偏差，博士論文的平均效果量未達顯著，期刊論文、碩士論文及碩士在職專班論文的平均效果量則達顯著。其次，區隔出僅有儲存功能的短期記憶，以及具有儲存及處理功能的工作記憶。在小學階段，短期記憶的預測力顯著低於工作記憶，特別是低年級樣本；在中學階段，短期記憶顯著高於工作記憶。工作記憶的視覺空間模板可顯著預測數學表現，但對國語文表現則無顯著預測力；工作記憶的語音迴圈及執行功能對國語文及數學表現皆有顯著預測力。此結果意味視覺空間模板的預測力具有領域特定性，語音迴圈及執行功能的預測力具有領域一般性。最後討論本研究限制並提出未來研究方向之建議。

This study examined the relationship of working memory and academic performance by means of meta-analysis. The researches published in Taiwan journals and the dissertations finished by the Taiwanese graduates were collected. Two hundred and fifty effect sizes were identified from 12 journal papers, 41 master theses, and 5 doctoral dissertations. Fixed-effect model and random-effect model were conducted. We performed z tests to examine the statistical significance of the difference between two average effect sizes. It was found working memory and short-term memory were significantly predictive of academic performance. However there was publication bias, that is, the mean of the effect sizes from journal papers was significantly higher than those from Masters’ theses. Those from doctoral dissertations were not significant. The difference between short-term memory and working memory was recognized. Information is stored in these two memory systems, but is processed only in working memory. It was found that in elementary schooling, the prediction of short-term memory was lower than the one of working memory, especially in the first and the second grades. In high schooling, short-term memory was more predictive than working memory. Working memory has four elements: the central executive, phonological loop, the visuospatial sketchpad, and the episodic buffer. We found that the visuospatial sketchpad was predictive of math performance, not Chinese; the central executive and the phonological loop were predictive of Chinese and math performance. It was implied that the prediction of the visuospatial sketchpad was domain specific and the one of the central executive and the phonological loop was domain general. Finally the implication and suggestions were discussed.