An Application Of MFRM In Independent Admissions Interviewing

As China's college enrollment system reform progresses, college admissions system is increasingly becoming an important practice of many colleges and universities to recruit modes. College admissions system has become Reform in Higher Education is an important influencing factors, of which a large part of the reason is relative to the traditional college entrance examination system, the independent entrance exam has increased in the interview part of this assessment.Self-enrollment system will be written with a combination of recruitment interviews way more in line with modern educational philosophy, but also can improve the efficiency of unity between fairness and efficiency, is conducive to the further realization of educational equity. But in many judges for a candidate or multiple groups of candidates for judges interview, interview reliability and validity instability, the result is not the exact problem has always been there, the judges and the judges of the consistency between the score is always a key issue.In this paper, D college admissions interview in 2012, for example, the use of multi-faceted Rasch model 581 candidates interview results were analyzed. Reference to the results of previous studies, this study attempts to use based on multi-faceted Rasch model (Many Facet Rasch Model, referred MFRM FACETS software developed to interview candidates scoring results to conduct investigations. Mainly for raters, raters with the candidates deviations and rating scale analysis of the three surfaces. During the study, we used MFRM severe raters for each and reliability for a specific estimate and compare, analyze each raters scoring candidates in particular produced during the specific reasons for the deviation, and extracted out of the abnormal scores. Meanwhile, the results of this study show that D college admissions interview scoring system used to fit very well with MFRM.We hope this study for college admissions interview ratings in selection and training of raters provide specific statistical basis promote MFRM in the actual score monitoring application.