欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
认知诊断、项目自动生成是现代心理测量领域的重要发展领域，二者的结合更是心理测量领域亟待开展的重要课题。本研究以小学数学问题解决认知诊断项目自动生成为例，探讨认知诊断领域的项目生成技术及算法。研究发现：（1）计算机自生成的项目参数与原模板参数具有较高的一致性。（2）同一项目模板下生成的不同试题的测量学特征基本不变。（3）同一批被试在自动生成的两份试卷的前、后测的能力（仍）值高度相关（r=．811），前、后两次对被试诊断结果的一致性高达86．5％。这表明本文所设计的认知诊断测验项目的自动生成技术及其算法基本可行，小学数学问题解决认知诊断项目的自动生成效果较好。这也为其它认知诊断领域的项目自动生成提供了技术借鉴和支持。
英文摘要：
Cognitive diagnosis （CD） is supposed to find out the advantages and deficiencies of individuals about their cognitive status, which is helpful to detect the underlying routines of psychological behavior and to understand their internal progressing mechanism. Automated item generation （AIG） is a technology designed to generate items with computers given existing information. For example, the characteristics of items are calibrated on the cognitive model, thus equation testing is not required. Cognitive diagnosis and automated item generation are both important in modem psychological measurement. How to combine both in application is a major topic to explore. Illustrating with ordinary mathematic problem solving, this paper investigated the methodology and algorithm of automated item generation in cognitive diagnosis. In this study, two stage test designs were used. First, twenty seven item templates were used to generate test items and two tests were formed. In pretest, 852 participations were examined and the reliability were 0. 764, 0. 737 respectively ; then the characteristics of these items were calibrated. Second, two tests were generated with the item templates and ten items were used as the same templates, which were used to detect the validation of automated item generation. Findings showed ： （ 1 ） Item parameters generated by the computer were highly consistent with the parameters in the model; （2） The tested characteristics of different items generated by the same item model were almost the same; （ 3 ） The abilities consequences obtained by two different auto generated tests were highly correlated （ r = 0.811 ） , and cognitive diagnosis obtained by the same tests were highly consistent. The above results suggest that the methods and algorithm of automated item generation in cognitive diagnosis are feasible, and they could work efficiently, which provides a mechanic reference and support of the automated item generation in cognitive diagnosis area.
同期刊论文项目
同项目期刊论文
期刊信息
位置：