結構方程模型(structural equation model, SEM)有逐年越來越多被應用的趨勢。在統計檢定方面,仍需加強報導及檢測能力。本研究以MacCallum, Browne, and Sugawara(1996)近似均方根誤差(root mean square error of approximation, RMSEA)估計法計算統計檢定力與樣本數。運用R 語言程式,分析《大專體育學刊》有關於SEM論文的應用概況及統計檢定力與樣本數決定。研究結果發現SEM的應用大部分的模式統計檢定力與樣本數還算理想,但仍需有加強。本研究提供未來研究在研究設計、SEM統計檢定力計算及樣本數決定運用程序等的實用建議,以作為體育領域學者應用SEM研究的參考。
Structural equation modeling (SEM) has been applied widely in the researches over the past few years. However, there are the places for improvement in both reporting and testing in the aspect of statistical testing. This research adopted root mean square error of approximation (RMSEA), developed by MacCallum, Browne, and Sugawara (1996), to calculate statistical power and sample size, and used R programming language to analyze the papers with SEM application that were published in ”Sports & Exercise Research”. The findings were shown that the statistical power calculation and sample size for most papers were fairly satisfactory, yet required to improve. The suggestions on research design and the application of statistical power calculation and sample size determination of SEM can be referenced by future researches with SEM application in sports and physical education field.