Investigating the Sources of Differential Item Functioning by Using Cross-Classification Multilevel Item Response Model

• Main Authors: Guo-wei Sun, 孫國瑋
• Other Authors: Ching-Lin Shih
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/sfwkg9

博士 === 國立中山大學 === 教育研究所 === 106 === Differential item functioning (DIF) analyses are important in terms of test fairness and test validaity. As Zumbo (2007) states, “Third Generation DIF” is best characterized by a subtle but extremely important change in how we think of DIF. The matter of wanting to know why DIF occurs is an early sign of the third generation of DIF. To detect differential facet functioning (DFF) in such a multilevel setting, the cross-classification multilevel item response model (CCMIRT) can be adapted. When the group main effect and item-by-group interaction effects are included in the CCMIRT, the random effects of group over items represent the DIF residual. The CCMIRT can be further extended by adding item characteristics predictors to explain the DIF. The purpose of this study is to investigate the sources of DIF by using the CCMIRT combining with DFF. In the simulation study, the variable about DIF effect were manipulated to better understand the performance of the CCMIRT. For fitting real test situation, the model including multiple item properties was suggested in this study.