A DIF study and competence indicators achievement of 6th grade students’mathematics test

• Main Authors: Shuie-Lin Chuang, 莊學霖
• Other Authors: 曾建銘
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/82960346758670162501

碩士 === 國立臺中教育大學 === 教育測驗統計研究所 === 99 === We used data from TASA (Taiwan Assessment of Student Achievement) and analyzed it with two different methods: 1. We used classical test theory (CTT) and item response theory (IRT) to test the difficulty of the questions and how the students performed and achieved the correct answers. 2. We used the following methods , Mantel-Haenszel, Crossing-SIBTEST, EZDIF ETS, and EZDIF LR to verify the questions of 2006 TASA mathematics test to see if DIF exist. The test sample was sufficient, so when =.01, if more than three methods verified the question was DIF, we assumed this question was DIF. If Crossing-SIBTEST tested the question was DIF while Mantel-Haenszel and ETS did not, we verified them to see if non-uniform DIF exist. We also performed the distraction analysis to find out why DIF occurs and what methods are feasible to verify DIF. It appears that the consistency rate of the four methods for DIF is over 79.67%; among them, the consistency between LR and Mantel-Haenszel and LR and ETS are the lowest. In the end, we proposed some suggestions to school instructors, TASA database administrators based on the competency differentiations between boys and girls for future research reference. The conclusions of our study are listed below: 1. The achievement of competence indicators of 6th grade students:「N_3_10」has the lowest average achievement rate while 「N_2_8」has the highest. Boys’ correct rate is over 5% better than girls in two competence indicators while girls have only one. We applied independent sample T to examine the three indicators and verified their significance. 2. 7 questions appeared to be DIF out of 103 which is about 6.8%. In whole, we cannot decide if the 2006 TASA 6th grade math test is favorable for the boys or the girls. 3. Conclusions of option analysis: 6 DIF questions have good discrimination, the other one does not. In tense of difficulty of the questions, only one question is difficult for both girls and boys; the rest of them are ranged between easy to medium. Averagely, option distraction is at a fair level.