【摘要】 大规模教育测验数据包含丰富信息，采用认知诊断方法对其进行分析，可以更好地揭示个体的认知过程、发挥测验的诊断功能。本研究以我国某项大型科学素养测验为例，通过验证Q矩阵和选择拟合最佳的认知诊断模型，对学生作答数据进行了认知诊断分析。结果表明，基于科学内容领域和科学能力所标定的Q矩阵拟合最佳；饱和的G-DINA模型拟合最好；学生在科学能力属性上掌握更好，不同掌握模式差异较大。研究应用认知诊断方法对大规模测验进行分析，为后续研究提供了路径和方法上的参考。更多还原

【Abstract】 Large-scale educational assessment data contains abundant information. Applying cognitive diagnosis methods to analyze such data can more effectively uncover individual cognitive processes and enhance the diagnostic function of assessments. Taking a large-scale science assessment in China as a case study, this study applied cognitive diagnosis analysis to students’ response data by verifying the Q-matrix and selecting an appropriate cognitive diagnosis model. The results indicate that the Q-matrix, including both science content and science competence, is the most suitable model. Moreover, the saturated G-DINA model fits the large-scale assessment data better than other unsaturated models. In addition, students demonstrated a stronger mastery of science competence attribute, with significant differences observed across various mastery modes. This application of cognitive diagnosis methods to analyze large-scale assessments provided valuable insights for future research.更多还原