目的 评估分析2014年～2016年北京市核医学住院医师规范化培训笔试考核数据,提出切实可行的试卷命题改进方案。方法 统计试卷难度、区分度、信度和效度,并对各年数据进行差异分析。结果 3年试卷难度比例均较好,2014年～2016年整卷难度系数分别为0.87、0.83和0.78,按照预定及格线评价试卷难度偏易;3年试卷不存在无区分度试题,但存在少数消极区分度试题,区分度较好的试题数量逐年增加;3年试卷分半信度均在0.90左右,内部一致性信度均高于0.80;3年试卷亚专科内容效度均高于90.0%,结构效度或偏高或偏低。结论 2014年～2016年北京市核医学住院医师规范化培训笔试试卷难度与考核组设定的最低标准不尽相符,需要重新设定最低标准或试卷难度。试卷区分度、信度和效度较好,能够有效、可靠地达到考核目的。

Objective To evaluate the data of graduation examination of the nuclear medicine residencies standardized training in Beijing, and to provide the feasible exam programs.Methods SPSS software was used to deal with the examination paper design and test scores, and analyze test difficulty, discrimination, reliability, and validity.Results The whole difficulties are 0.87,0.83,0.78 from 2014-2016, respectively, and higher than the planned values. There are some questions that may be deleted in three-year papers though discrimination analyses. To be satisfied, the reliabilities and content validities in three years are all considerably good. While the construct vadilities are not as decent as expected.Conclusions The course examination should pay attention to improve the test difficulties in consistent with the planned values or reset the qualifying standard. The graduation examination with satisfactory discrimination, reliability, and validity is capable to achieve the test purposes.