การเปรียบเทียบประสิทธิภาพในการตรวจสอบการทำหน้าที่ต่างกันของข้อสอบ ระหว่างวิธีการวิเคราะห์องค์ประกอบจำกัด แมนเทล-แฮน์เซล และการตอบสนองข้อสอบ / นิคม กีรติวรางกูร = A comparison of the efficiency in detecting differential item functioning among restriced factor analysis, Mantel-Haenszel, and item response theory procedures / Nikhom Keeratiwarangkul
To compare results from the detection of uniform differential functioning test items among Restricted Factor Analysis (RFA), Mantel-Haenszel (MH) and Item Response Theory (IRT) two-parameter procedures. The factors manipulated in this study were : 1) two sample sizes of examinees : 300 (for small sample) and 1000 (for large sample). 2) two sizes of test length : 25 (for short form) and 75 (for long form) items, 3) three levels of item discrimination : high, moderate, and low, 4) three levels of item difficulty : high, moderate, and low, and 5) two levels of item bias : high and low. The major findings are as follows : 1. The Restricted Factor Analysis (RFA) has the highest efficiency in detecting differential item functioning (DIF) followed by the Mantel-Haenszel (MH) and the Item Response Theory (IRT) two-parameter procedures. For Type I error rate, the Item Response Theory (IRT) two-parameter procedure yields the highest followed by the Mantel-Haenszel (MH) and the Restricted Factor Analysis (RFA) procedures. 2. The Mantel-Haenszel procedure is highly efficient in detecting DIF under these conditions : low item difficulty, high item discrimination, long form, and large sample size. 3. The Item Response Theory two-parameter procedure is highly efficient in detecting DIF under these conditions : low item difficulty, all three levels of item discrimination, long form and large sample size.
