Design And Application Of Item Selection Strategy In Adaptive Testing Of Modern Educational Technology

With the increasing importance of technology in all aspects of social life,education informatization is more and more concerned by the education sector.Integrating technology into education evaluation makes personalized evaluation more popular.Computerized Adaptive Testing is widely used in education evaluation because of its advantages of ―thousands of people,thousands of papers‖ and high efficiency and accuracy.Modern Educational Technology is a common course which aims to improve normal of information technology ability students.The content domain and target level of its teaching content are clearly divided.Its evaluation method also needs to consider the balance of content and target level while considering personalization.Obviously,it is not enough to only apply the Maximum Fisher Information to the Computerized Adaptive Testing of this course.Therefore,it is necessary to design appropriate item selection strategy for the course.Firstly,this paper analyzed the existing literature on computer adaptive test,and summarized the item selection strategies of Computerized Adaptive Testing.At present,the existing item selection strategies mainly include four categories: improving the accuracy of test results,controlling the exposure rate of test questions,controlling the balance of content and comprehensive item selection strategies.The new item selection strategies often improve the common and basic item selection strategies by adding parameter settings to meet different test requirements.However,there is no item selection strategy that takes the goal level balance into account and takes it as a constraint condition with the content balance control at the same time.Based on the analysis and summary of the existing item selection strategies,combined with the test content and target level distribution of the Modern Educational Technology,an item selection strategy based on content balance and target level balance is designed.According to the question distribution in the existing question bank,the questions answered in each test are chosen to ensure that each content domain or target level in each Adaptive Testing needs to answer the questions.The number distribution is consistent with that in the question bank,that is,the proportion of the number of answer questions in each content field or target level to the length of the test is consistent with the proportion of the number of questions in the content field or target level to the total number of questions in the question bank.In addition,you can also set the number of questions for each content area and target level according to the test needs.In order to verify the effectiveness of the new item selection strategy,Monte Carlo simulation experiment was carried out by using the commonly used maximum information method and new item selection strategy.There are two question banks in this simulation test,one is the parameter value of the question simulated by Win Gen software,the other is the parameter value of the question in the existing question bank.The test adopts the termination condition of fixed length method.Under the two strategies,30,45,60 and 75 fixed length adaptive tests are conducted respectively.Each test length corresponding to each strategy is tested three times,and a total of24 simulation experiments are required.After analysis,the new item selection strategy based on content balance and target level balance is slightly inferior to the maximum information method in ability estimation,but the difference between them will decrease with the increase of test length.In terms of test efficiency,test question call uniformity,test overlap rate,the new item selection strategy and the maximum information method have their own advantages and disadvantages,but the difference is not significant.For the control of content and goal level balance,the new item selection strategy has more advantages,which is completely consistent with the expectation.On the basis of the simulation test,the Computerized Adaptive Testing system for the Modern Educational Technology is developed by using the new item selection strategy.140 subjects are randomly selected for the test,and each subject needs to answer 75 questions.After the test,through the analysis of the response time,the ability level estimation,the content of the answer questions and the distribution of the target level,we can see that the item selection strategy based on the content balance and the target level balance can effectively achieve the balance of the test content in both the content domain and the target level,and will not have too much impact on the accuracy of the ability estimation of the subjects.This strategy can make the number of questions in each content domain and target level consistent with the expectation,and because the selection strategy takes into account the amount of information of the remaining questions in the question bank,the subjects answer the questions that match their ability level,which can get a more accurate judgment of the knowledge points in each content domain and target level.