Abstract: Objective To evaluate the mastery of scoring standards by raters in 2018 clinical fundamental skills test of National Medical Licensing Examination phased the first stage,to explore the consistency between standardized patients (SP) and examiners' scores, and to provide more information for relevant research. Methods In 2018, based on the scores of communication capacity and humanistic care from 77 candidates in clinical fundamental skills test of the National Medical Licensing Examination phased the first stage in a randomly selected medical college, use the many facets Rasch model to calculate estimated ability of 77 candidates, analyze the internal consistency of raters and the leniency and strictness of evaluation. Results The results showed that the average estimated capacity of 77 candidates was 2.75 logits, and the most examinees' infit was less than 1.5.The average severity of the raters was -0.55, the severity of the examiners was -0.45, of the SP was -0.70,The difference was not statistically significant. Conclusions Raters have a good command of scoring standards, the overall standard is relatively loose.The scores of SP and examiners were consistent.

Key words: National medical licensing examination, Rater error, Rasch theory, Many facets Rasch model, Standardized patients