PURPOSE The study investigates the efficacy of new features introduced to the selection process for medical school at the University of New South Wales, Australia: (1) considering the relative ranks rather than scores of the Undergraduate Medicine and Health Sciences Admission Test and Australian Tertiary Admission Rank; (2) structured interview focusing on interpersonal interaction and concerns should the applicants become students; and (3) embracing interviewers’ diverse perspectives. METHODS Data from 5 cohorts of students were analyzed, comparing outcomes of the second year in the medicine program of 4 cohorts of the old selection process and 1 of the new process. The main analysis comprised multiple linear regression models for predicting academic, clinical, and professional outcomes, by section tools and demographic variables. RESULTS Selection interview marks from the new interview (512 applicants, 2 interviewers each) were analyzed for inter-rater reliability, which identified a high level of agreement (kappa=0.639). No such analysis was possible for the old interview since it required interviewers to reach a consensus. Multivariate linear regression models utilizing outcomes for 5 cohorts (N=905) revealed that the new selection process was much more effective in predicting academic and clinical achievement in the program (R2=9.4%–17.8% vs. R2=1.5%–8.4%). CONCLUSION The results suggest that the medical student selection process can be significantly enhanced by employing a non-compensatory selection algorithm; and using a structured interview focusing on interpersonal interaction and concerns should the applicants become students; as well as embracing interviewers’ diverse perspectives.

中文翻译：

---

通过改进选拔算法和改变澳大利亚面试焦点来加强医学生选拔过程：一项描述性研究

目的本研究调查了澳大利亚新南威尔士大学医学院选拔过程中引入的新特征的有效性：(1) 考虑本科医学与健康科学入学考试和澳大利亚高等教育入学考试的相对排名而不是分数秩; (2) 结构化面试，重点关注人际互动以及申请者成为学生时关心的问题；(3) 接受采访者的不同观点。方法 对 5 组学生的数据进行了分析，比较了旧选拔过程中的 4 组学生和新过程中的 1 组学生的医学课程第二年的结果。主要分析包括通过部分工具和人口变量预测学术、临床和专业结果的多重线性回归模型。结果 对新面试（512 名申请者，每人 2 名面试官）的选拔面试分数进行了评估者间的可靠性分析，结果表明高度一致（kappa=0.639）。旧的采访不可能进行这样的分析，因为它需要采访者达成共识。利用 5 个队列 (N=905) 结果的多元线性回归模型显示，新的选择过程在预测项目中的学术和临床成就方面更加有效（R2=9.4%–17.8% vs. R2=1.5%–8.4%） ）。结论 结果表明，通过采用非补偿性选择算法可以显着增强医学生的选择过程；并采用结构化面试，重点关注人际互动以及申请人成为学生时所关心的问题；以及接受采访者的不同观点。