本研究目的在於探查評分者的系統性偏誤。研究資料是標槍投擲技術測驗，共有18名選手，每人試擲三次，由五位評分者針對十三項投擲技術，給予三個等級的評定。研究方法採用多層面Rasch測量（Many-Facet Rasch Measurement, MFRM），利用Facets程式交叉分析個人和評分者、試題和評分者的測量差異，探查評分者系統性偏誤。研究結果：一、個人層面和評分者層面、試題層面和評分者層面的交叉分析，發現都是隨機性偏誤。二、評分者在個人的系統性偏誤：（一）A評分者對8號個人的評分高估。（二）A評分者對11號和17號個人的評分低估。三、評分者在試題的系統性偏誤為A評分者在第5題的評分高估。結論：MFRM能夠探查競技運動主觀評分之評分者系統性偏誤。

The purpose of this study was to detect the systemic bias of raters. The data was a Javelin throwing skills testing. The subjects were 18 javelin players that each player threw the javelin three times. Then five raters gave each trial a three-grade assessment in the thirteen skills of Javelin throwing. The Many-Facet Rasch Measurement (MFRM) was applied to the research method. The method was to use Facets program to cross analyze the measured differences between the persons and raters, items and raters. It detected the systemic bias of rater in sports. The Results were: 1. the cross-analysis between the persons and raters, items and raters of inter facets existed randomness bias. 2. The systemic bias of the raters for each person: (1) the rater A rated higher bias measure for No. 8 person. (2) The rater A rated lower bias measure for No. 11 and No. 17 person. 3. The systemic bias of the raters for each item was that rater A was rating higher bias measure for the 5th item. The conclusion: using the MFRM can be detected the rater's systemic bias in the subjective rate of sports.