目的：以Rasch模式分析舞獅運動裁判判決的信度與效度。方法：資料取自2006年全民運動會舞獅（臺灣獅）比賽項目七縣市隊伍與九位裁判的判決結果，以Facets軟體進行多層面Rasch模式分析，利用適合度考驗、難度、估計誤差和非期望反應建立對照量尺，根據非期望反應值和標準殘差探討選手與裁判的表現。結果：Rasch模式能夠對裁判的判決表現進行整體的描述與分類，並提出客觀的數據（標準化殘差值）追蹤不良的判決。結論：Rasch模式是建立在資料與模式適合的基礎上，當出現適合度不佳的試題或是個人（裁判）時，即針對非期望反應進行分析，作爲最後評斷測驗成敗的依據。本研究利用Rasch模式能精確而深入地分析裁判的判決行爲。

Purpose: To analyze the judge's rating of Lion's Dance by using many-Facet Rasch model. Methods: The data of the rating of Lion's Dance judges in the final of The 2006 National Games was analyzed. The data was analyzed by Facets Rasch software. The scale of performances was created, using infit, outfit, measure, error of estimate, and unexpected responses. The performances of players and judges were based on unexpected responses and standardized residual. Results: Rasch model could describe and classify the performance of the judge. Furthermore, the judge's unexpected response was also tested and overtook. Conclusion: Rasch model could better analyze judge's rating precisely.