組織研究的中介檢測：緣起、爭議、研究設計和分析

Testing Mediation in Organizational Research: Origins, Challenges, Research Design, and Analysis

在組織研究領域裡，中介模式的檢驗非常普遍，然而有關中介變項的概念和驗證方法令學者疑惑之處仍多。本研究針對跨越八十年間，散見於不同文獻之重要中介論文及專書內容加以彙總及評述。首先，我們追溯中介變項概念的歷史源頭，從1930 年代心理學家的論戰談到迄今的發展。其次，研究者常以Baron 和Kenny (1986)所提出的判斷條件作為中介效果的檢測，但該方法近年所受到甚多的挑戰和質疑，我們在此作一重點整理。第三，從研究設計的角度，我們討論進行中介研究應注意的要項。第四部分則是中介模式的驗證分析程序，我們整理了五大類較有代表性的中介模式，評析各模式驗證邏輯、步驟和優缺點。有志於中介研究的組織學者，可將本研究作為實用的參考工具。

Despite that testing mediation has been prevalent in organizational research for decades, there remain much confusion in theories, concepts, and analytical practices. This study highlights and comments on existing discourses that span 80 years in different literatures and presents them in four sections. First, we traced back the origins of mediation research in psychology in the 1930s and in sociology in the 1950s and linked the discourses to the present. Second, we summarized the major challenges to the most frequently used concepts and methods introduced in the Baron and Kenny (1986) study. Third, we proposed research design issues pertaining to mediation testing. Finally, we highlighted five types of fundamental mediation models and discussed the pros and cons of various analytical techniques in testing these models. This study hopes to serve as a useful guide for mediation researchers.