中文摘要：

本文以上海某大学攻读硕士学位研究生英语复试为个案，在细化考试构念为源语阅读能力、目标语写作能力、翻译知识、语际能力和认知能力的前提下，分析2696名考生应答数据，并对该试题进行了初步的效度验证。统计分析发现，分数信度较好、汉译英试题的区分度较好；考生成绩因报考院系、报考专业、考生本科毕业院校所在省份不同而呈现显著差异，差异来源主要可归因于汉译英成绩。考试的卷面效度、内容效度、准则关联效度较好。复试更有利于检测考生的英语应用能力，但题量偏小，适合辅助遴选的考试目的。

英文摘要：

The present research reported, under the construct framework of the test, the score structure, the reliability, the difficulty coefficient, and the discrimination of an entrance test of English for master-degree candidates. ANOVA and regression analyses of data from 2696 subjects indicated that region, university type, research area applied by the test-takers influenced their average test scores, and that Chinese-English translation ability mainly accounted for the difference. The test scores are accountable in terms of face validity, content validity, and criterion-referenced validity. This test fulfills its purpose of providing additional evidence for screening master candidates by examining their English competence.