效果量在臺灣心理與教育期刊的應用：回顧與再思

Effect Size Reporting Practices in Taiwanese Psychology and Education Journals: Review and Beyond

雖然效果量在量化實徵研究裡十分重要，但過去尚未有研究探討效果量在臺灣心理學與教育學的應用情況。本研 究系統性地回顧了 2017 年與 2018 年發表在臺灣具高評價的八本心理學與九本教育學期刊，一共 268 篇文章，旨 在探討四個報告效果量的面向：（1） 效果量報告的比例、（2） 效果量報告的類型、（3） 效果量的解釋，以及 （4） 作者 如何處理統計顯著性與效果量強度的落差。結果顯示：72% 的文章報告至少一個效果量，超過 65% 的效果量是 r 類型（如：Pearson相關係數或 ）。在報告效果量的文章中，55%解釋了效果量，80%以上對效果量的解釋僅以小、 中或大的標籤標註。在同時報告統計顯著性與效果量的文章裡，50% 呈現兩者間有落差的問題，其中僅有 35% 對 落差提供解釋。就兩學門的比較而言，儘管心理學期刊文章報告效果量的比例顯著地低於教育學，然而教育學文 章使用標籤化的方式解釋效果量之比例卻顯著地高於心理學。整體而言，大多數的作者報告了效果量，卻未必對 效果量的意涵提供完整的解釋。本文提出五點效果量報告之建議與四個效果量報告的範例，希冀能幫助讀者在研 究報告中正確地闡述效果量的意涵，並促進效果量在量化研究之廣泛應用。

The importance of reporting effect sizes (ESs) in quantitative empirical studies has been emphasized in the literature. However, no published study to date has shed light on current ES reporting practices in Taiwanese psychology and education journals. To fill this gap, the present study systematically reviewed 268 articles published in eight Taiwanese psychology journals and nine education journals during 2017 and 2018. All of these 17 journals were highly ranked in their respective fields. Four aspects of ES reporting practices were investigated: (A) the ES reporting rate, (B) the ES type, (C) the ES interpretation, and (D) the resolution of discrepancies between the ES magnitude and statistical significance. The results revealed that 72% of articles reported at least one ES, and more than 65% of ESs reported were the r-type, such as Pearson's r and . Of the studies that reported ESs, 55% also interpreted the ESs. More than 80% of these interpretations were the mere labeling of an ES as small, medium, or large, according to established benchmarks. Approximately 50% of the articles showed a discrepancy between the magnitude of an ES and its corresponding statistical significance, but only 35% of these articles attempted to explain or resolve the discrepancy. When the data for psychology and education articles were analyzed separately, the psychology articles exhibited a lower rate of both ES reporting and ES interpretation by labeling. In sum, the majority of articles reported at least one ES, but few interpreted ES fully or meaningfully. To assist authors with a full and meaningful ES reporting, we offer five suggestions and one exemplary ES reporting in the Extended Abstract. It is hoped that this paper contributes to an increased practice of meaningfully reporting ES(s) in empirical quantitative studies in Taiwan.