机械安全技术课程目标达成度评价

Evaluation of the Goal Achievement of Mechanical Safety Technology Course

摘要: 产出导向教育(OBE)要求课程目标达成度可量化、可审计，但是已有研究大多停留在报告达成度数值上，缺少对目标–试题映射完整性、试题测量质量、认知层次分布的联合诊断。以机械安全技术课程为案例，创建可复用的达成度评价基础层，并对其进行测量可信度的方法学检验。以83名高职学生A卷期末客观题(35题)为样本，建立两层目标体系(K1~K5、A1~A4、S1~S3)、Q矩阵、评分规则的评价框架，并且用经典测量理论来对试题质量进行诊断，即KR20与Cronbach’s α信度、项目难度p、点二列区分度r、按题型编码的布鲁姆认知层次分布。知识目标平均达成度为91.2%，可测能力目标(A1~A4)平均为91.7%，整体良好。试题质量诊断显示存在显著天花板效应(平均难度p = 0.916)和普遍低区分度(23/35题，|r| < 0.20，α = 0.431)，说明达成度数值对区分学生能力的贡献很小。结论认为达成度基础层必须同时给出映射缺口和试题测量属性，才能作为TOPSIS等多准则决策的可靠数据基础层；仅依赖百分比不能用来推导出课程改进结论。

Abstract: Outcome-based education (OBE) requires that the achievement of curriculum objectives can be quantified and auditable. However, most of the existing studies only focus on the value of report achievement, and lack of joint diagnosis of the completeness of target-question mapping, the quality of question measurement and the distribution of cognitive level. Taking the course of mechanical safety technology as an example, a reusable basic layer of achievement evaluation is created, and the methodological test of measuring credibility is carried out. Taking 83 higher vocational students’ A-volume final objective questions (35 questions) as samples, the evaluation framework of two-level target system (K1~K5, A1~A4, S1~S3), Q matrix and scoring rules was established, and the classical measurement theory was used to diagnose the quality of the questions, that is, KR20 and Cronbach’s α reliability, item difficulty p, point two column discrimination r, Bloom’s cognitive level distribution encoded by question type. The average achievement degree of knowledge goal is 91.2%, and the average measurable ability goal (A1~A4) is 91.7%, which is good as a whole. The diagnosis of test quality showed that there was a significant ceiling effect (average difficulty p = 0.916) and a general low discrimination (23/35 questions, |r| < 0.20, α = 0.431), indicating that the achievement value contributed little to distinguishing students’ ability. It is concluded that the reachability base layer must give the mapping gap and test measurement attributes at the same time, so as to be a reliable data base layer for multi-criteria decision-making such as TOPSIS. Pure percentage cannot be used to deduce the conclusion of curriculum improvement.