r语言软件GDINA_认知诊断分析系统(flexCDMs)设计及其实现

【Abstract】 Cognitive diagnostic assessment arose in 80 s last century as a new testing mode which combines cognitive psychology theory and modern measurement methods. It intended to explore cognitive processes and cognitive structure of human in some specific areas, evaluate knowledge structure and processing skills of individuals. In recent years, cognitive diagnostic assessment had been used in educational assessment and clinical diagnosis of psychological diseases. Cognitive diagnostic assessment rely on estimation of cognitive diagnostic models’ parameters which could be achieved with CDM package and GDINA package based on R language. But it require the knowledge of R language which is difficult for unprofessional Researchers and users. And there are no software or platforms for cognitive diagnostic assessment yet in China. Based on such situations, this paper developed a new cognitive diagnosis analysis system(flex CDMs), this system was online used and didn’t require users to grasp any programming language, so this system would greatly lower the threshold of cognitive diagnostic assessment. The flexCDMs system was consisted of 7 function modules. The first function module was construction of Q-matrix, the reachability matrix、the ideal measure pattern and the ideal master pattern could be calculated by flexCDMs system, also, the system could estimate and validate Q-matrix. The second function module was parameter estimation, the system could estimate parameters of dichotomous CDMs(DINA、DINO、ACDM、LLM、r RUM、GDM、LCDM、GDINA Mixed) and polytomous CDMs(Sequential G-DINA). The third function module was test quality analysis, the system could calculate the item discrimination and the testing reliability. The forth function module was goodness-of-fit test including global fit、item-model fit and person fit. The fifth function module was DIF detection, the flexCDMs used Wald method or M-H method to detect differential item function. The sixth function module was graphical output, the item characteristic curve、the bar graph of examinees’ score and the scatter graph of the probability of attributes Mastery could be plotted by flexCDMs system. And the last function module was data simulation. The system could simulate item parameters、master pattern of examinees and response matrix, and some real data are also provided in this system. The flexCDMs system could be used online, the language of system was Chinese, and it only required users to provide the response data of examines and the Q-matrix for complex cognitive diagnosis. So this simple, practical and easy operative system will greatly promote the popularization and application of cognitive diagnostic assessment.