Rough Structural Modeling and Its Applications in Educational Measurement

• Main Authors: Chen, Tzu-Liang, 陳姿良
• Other Authors: Sheu, Tian-Wei
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/38755247548488378906

博士 === 國立臺中教育大學 === 教育測驗統計研究所 === 102 === The purposes of this study are to propose Rough Structural Modeling (RSM) to improve structure and visualization of Rough Set Theory in educational measurement, to propose program of testing decision attribute for solving shortcoming of decision attribute that is subjective judged by experts in Rough Set Theory. Through three empirical examples (analyses of Misconceptions’ Domain, sensitivity, and curriculum’s structure), the objective and innovative method with verified feasibility and reliability is proposed in nonparametric statistical. RSM is a method of structural analysis that combined Rough Set Theory and Formal Concept Analysis (FCA) with operation of reachable matrix. It generates Re-configuration Matrix to establish hierarchical structural graph based on operation of binary concept lattice. The graph of different decision attributes can be applied in analyses of Misconceptions’ Domain, sensitivity, and curriculum’s structure. Program of testing decision attribute is applicable regardless of the amount of participants that combines Receiver Operating Characteristic (ROC) and Nonparametric Chi-square Test. Applying the program in analysis of Misconceptions’ Domain with different Gamma value for improving related method. For sensitivity analysis, shortcoming of subjective judged by experts in ROC can be improved by evaluating cut-off points of Gamma value and sorting. Because of objective factors, limitations of the study are as follows: (a) The sampling methods are purposive sampling. (b) The data are both binary data in formal context and operation of binary concept lattice.