以統計檢定力檢測結構方程模式運用在商管研究的正確性

By Using Statistical Power to Test Correction of Applications of Structural Equation Modeling on Business and Management Research

周子敬(Tzu-Chin Rejoice Chou)

《中原學報》 32卷4期 (2004/12) Pp. 619-633

https://doi.org/10.6358/JCYU.200412.0619

摘要

結構方程模式(SEM)因其分析的特性有在國內有逐年越來越多被應用的趨勢。在統計檢定方面，不管是國內、外方面仍需加強報導及檢測能力。本研究藉由MacCallum等人的SAS程式，檢視了國內TSSCI期刊『管理學報』最近5年有關於SEM的應用概況及統計檢定力。研究結果發現SEM的應用有一定的程度，雖然大部分的模式檢定力還算理想，但仍需有加強的空間，本研究綜合文獻內容分析及個案分析結果提出將『統計檢定力』併為未來在檢測模式參考標準的可行性。透過文獻內容分析本研究整理在SEM原理方面及報導時的重要項目：模式規格、鑑定力、資料及估計、適配度、參數及其標準差以及替換模式。

關鍵字

統計檢定力 ； 結構方程模式 ； 商管研究

並列摘要

There are widespread applications for structural equation modeling (SEM) owing to its special characteristics as years gone by. Statistical power is still needed to increase the ability on reporting and testing, no matter what in the nation or abroad. This study further test the recent 5-year articles concerning with applications of SEM and statistical power of ”Journal of Management” in Taiwan TSSCI. The results shown that SEM in Journals held out some degrees of application, although most of the power was satisfied enough, but it still have degrees for enhancement. This study finally integrated results from content analysis, and case study, to propose combining the ”statistical power” test as checklist for future model test reference. This study by using content analysis, and integrate some crucial items while applying the SEM: (a) model specification, (b) identifiability, (c) data and estimation, (d) goodness of fit, (e) parameters and their standard errors, and (f) alternative model.

並列關鍵字

Statistical Power ； Structural Equation Modeling ； Business and Management Research