以多層面Rasch測量評估羽球評分試題

本研究旨在探查主觀評估時評分者發生評分不一致的試題。研究資料是羽球以高遠球做為回擊技術的主觀評分，共有10名受試者，每人試做3次，由五位評分者針對高遠球的9項技術，給予3個等級的評定。研究方法採用多層面Rasch測量（Many-Facet Rasch Measurement，MFRM），先估計各層面的測量品質，發現評分者的評分一致性層面不佳，再利用Facets程式交叉分析評分者和試題的測量差異，探查評分者在試題評分的偏誤。研究結果：一、評分者對試題有7個試題評定偏差。二、第8題有3次偏差最多。三、評分者雖無系統性偏誤，但仍有13筆偏差，宜再提高評分者一致性。四、兩位評分者類型在技術動作過程有系統性偏差，需要加強。結論：MFRM方法可用於小樣本研究，且能夠深入剖析評分者、試題、評分間的交叉影響，可提供運動技術主觀評分時具體的訊息，有助於改善主觀評分的品質。

關鍵字

偏差；評分者一致性；羽球

並列摘要

This study aimed to investigate items rated inconsistently by raters based on their subjective judgements. The research data comprised subjective ratings on high badminton serve performed by 10 participants. Each of the participants performed 3 high serves, which were then rated with respect to 9 techniques involved on a 3-level scale by 5 raters. The many-facet Rasch measurement was used to determine the measurement quality of each facet, which revealed an inconsistency of ratings among raters. The Facets program was then used for a cross-analysis between measurements by raters and items to examine the raters' bias in rating. Results: 1. The raters' ratings were biased in 7 items. 2. Item 8 had 3 biased ratings, which was the most among all items. 3. The raters, despite the absence of systematic bias, produced 13 biased ratings, which demonstrated the need for improvement in the interrater agreement. 4. Systematic bias was observed in the process of two raters rating the techniques, which needed to be improved. Conclusion: The many-facet Rasch measurement served as an useful tool for research with small sample size to identify items rated inconsistently by raters.

並列關鍵字

Bias ； Inter-rater Agreement ； Badminton

主題瀏覽