Imputation Method Of Missing Data In Cognitive Diagnosis Evaluation

With the continuous progress of education reform,the diagnosis and evaluation of student performance at home and abroad has gradually increased.Cognitive diagnosis is a meticulous evaluation of student performance.It focuses on the estimation of the strengths and weaknesses of individual knowledge structure.The completeness of student data is particularly important.However,due to the influence of various non-cognitive factors,there is always a certain degree of lack of test data obtained,if the missing data is not filled,will cause a misjudgment of the state of personality knowledge.The existing methods of filling missing data are mainly proposed for traditional tests.Due to the particularity of the preparation and scoring of cognitive diagnostic tests,most of them are not suitable for CDT,but there is no way to fill in missing data for cognitive diagnostic tests.Therefore,based on this situation,this research proposes a method of filling missing data in cognitive diagnostic evaluation,the filling method based on individual attribute mastery probability(AMPI)and the filling method based on the shortest distance between individuals(MDmin).Conducted for three studies: study one,theoretical analysis of the characteristics of missing data in cognitive diagnosis,and the thinking and rationality of MDmin and AMPI;study two,through simulation studies,to explore the filling of MDmin,AMPI and the four existing missing data The method is the filling effect on the test capacity,the missing ratio,the number of questions and the number of attributes;research three,through empirical research,using empirical data to further verify the effectiveness of the two methods in empirical.The research conclusions of this study are as follows:1.MDmin and AMPI are methods to fill in based on individual information;2.MDmin and AMPI have the same effect trend in parametric and non-parametric diagnostic methods;3.MDmin and AMPI show better filling effects under different conditions.MDmin is better than the other five methods for comparison,AMPI is slightly inferior to MI method,but in most cases,AMPI is more suitable for CDT.The two methods are not affected by the sample size.As the missing proportion increases,the filling effect decreases,as the number of attributes increases,the filling effect decreases,and as the number of questions increases;4.In the empirical results,it can be seen that MDmin Both AMPI and AMPI can be used in empirical research,and are consistent with the trend of various indicators without missing data;5.Overall,the filling effect of MDmin is better than AMPI.