Accuracy of DIF detection procedures in cognitive diagnostic modeling

• Main Authors: Pei-Ming Chiang, 江培銘
• Other Authors: Huey-Ing Tzou
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/88643903837256668788

博士 === 國立臺南大學 === 教育學系測驗統計碩博士班 === 103 === The purpose of this study was to investigate the efficacy of different DIF detection procedures in the cognitive diagnostic model. The study manipulated different attribute mastery profile distributions, cognitive diagnostic models, the ratios of sample size, and the numbers of attribute to examine the Type I error and statistical power of three DIF detection procedures. Comparing with other two Mantel-Haenszel methods, Wald test resulted higher statistical power but also inflated Type I error in many simulation conditions. Using attribute mastery profile as matching criterion could enhance the effectiveness of Mantel-Haenszel method, but it was also affected by model estimation error. The Mantel-Haenszel method matching on total scores yielded lower Type I error by adding iteration and purification procedures, but the results would be unstable when the attribute mastery profile distributions of focal and reference group were different.