應用概化理論估計運動技能測驗多層面變異來源的信度

The Applications of Generalizability The Ory to Estimate The Reliability of Multifacet Source of Variation in Psychomotor Test

本研究的目的是應用概化理論估計運動技能測驗多層面的變異來源。研究對象為113位國小學生，在經過四週的羽球教學課程之後實施羽球正手發高遠球測驗；實驗設計採「受試者」、「試做次數」和「評分者」等三個重要層面的重複設計（P ｘ T ｘ J），以「受試者」為固定水準模式（測量目標），「試做次數」和「評分者」為隨機水準模式的概化理論「二層面重複設計」（two-facet crossed-design），經由概化理論專用電腦程式GENOVA進行多層面變異來源的測量誤差估計。由概化研究的結果得知測驗的真實變異佔總變異成份的46.71﹪，為最主要的變異來源；「受試者和評分者」以及「受試者和評分者和試做」的交互作用，總共佔總變異成份的43.71﹪，為最主要的誤差變異來源；其餘的變異來源所佔的比率很小。決斷研究的結果發現適度的增加評分者人數可以提高測驗的概化係數。本研究根據研究結果提出下列結論：應用概化理論可以得到運動技能測驗中屬於多層面變異來源的整體測驗的概化信度係數；概化研究可以了解測驗中各種變異的來源及所佔的比率，使用者可以以實際情況決定最適當的測驗條件配置。The purpose of this study was to apply Generalizability (G) theory estimating the reliability of multifacet source of variation in the psychomotor test. The hitting performance of badminton test was administered to 113 elementary school students from NEHS. A Two-facet crossed-design of this study was identified as subjects, trials, and raters(S×T×J). The assessments were computed by using GENOVA. The results of G-study were: 1.The true variance of tests were the main source of variation and the percentage of total variance 46.71﹪. 2. The main sources of variation were the interaction of 'S ×J ×T' and 'S ×J' with 43.71﹪of the total variation. 3. The other sources of variation were small. The results of D-study were: The number of more rater and more trial advanced the G coefficient. The conclusions were: 1. The reliability of G coefficient was an overall for test that was to apply G theory on multifacet source of variation in the psychomotor test. 2. The source of variation was understudied by G-study in the percentage of total variance, and the best decisions of test condition were depended on the need of user.