The Construction For Cognitive Diagnosis Tests Of Multiple-choice Items And The Development Of Multiple-Choice Cognitive Diagnosis Model For Multiple Strategies

Cognitive diagnostic assessment(CDA) is designed to measure specific knowledge structures and processing skills of students, so as to the informations about their cognitive strengths and weaknesses are provided. To achieve the objective of CDA, both the cognitive diagnostic tests to extract the internal thinking mechanism of the examnees and the cognitive diagnostic models(CDMs) to map the relationship between unobservable artribute mastery and the observable item response are essensial.Multiple-Choice(MC) item format is commonly used in cognitive diagnosis assessment. In the traditional cognitive diagnosis modeling, the most straightforward and common way of analyzing MC responses is to treat them as dichotomous data. However, such an approach is suboptimal because it does not take into account the diagnostic insights about student difficulties and alternative conceptions that can be found in the distractors. With recent psychometric and technological advances, researchers are paying renewed attention to ways in which MC distractors can be used to provide diagnostically useful information through statistical modeling.To maximize the diagnostic capacity of MC items and to explore the extent to which using additional information from item distractors improves student profile estimation accuracy, it is essensial to develop diagnostically rich distractors and MC cognitive diagnosis tests. Matching the skill patterns of the distractors with the latent classes is an effective way to make the distractors with diagnostic capacity. The purpose of the the current study summarized and induced principles, requires and steps to construct multiple-choice cognitive diagnostic tests. Monte Carlo method was employed here to explore the rationality and the feasibility of the principles and requires. The findings were presented: under the four based attribute hierarchies, multiple-choice cognitive diagnostic tests had high average attribute match ration(AAMR) and pattern match ration(PMR). All in all, the principles and requires were reasonable and acceptable.Most of task solvings in CDA assume all examinees use the same one processing strategy. Factually, many psychological studies have shown that many cognitive tasks(such as syllogistic reasoning, figure reasoning, etc.) can be solved by more than one strategy. With the development of research in the field of CDA, the purpose of CDA has extended to know the strategies and its corresponding arttribute pattern the students used. To diagnose the multiple strategies in the MC tests, a new CDM which is called Multiple-Choice and Multiple Strategies CDM(MS-MC-DINA model) has been constructed. MS-MC-DINA model is defined in detail and model identification is achieved by JMLE. JMLE of MS-MC-DINA model were examined according several criterion, such as accuracy of item parameter, AAMR, PMR. The results showed that the accuracy of item parameter, AAMR, PMR performs well.For MC items, the information of single option or multiple options can be taken into account. In the process of task solving, single strategy or multiple strategies can be taken into account. If we combine the information of options and the strategies, there are three kinds of test conditions. They are single option information and multiple strategies, multiple options information and single strategy, multiple options information andmultiple strategies. To analye the data from the different test conditions, suitable CDMs are needed, because matching the CDM with the underlying cognitive process of task solving is the prerequisite of a valid and accurate of CDA. MS-DINA, MC-DINA and MS-MC-DINA models were used to analye the datas from the three test congditions respectively. The results indicated that the model and the test data fit better, the accuracy of item parameter, AAMR and PMR are high if the hypothesis of the model relatively consistent with the test condition. By comparing the results of MS-DINA, MC-DINA and MS-MC-DINA models used in different three test congditions, the performance of MS-MC-DINA model is proved.