Detection Of Differential Item Functioning Based On Item Response Theory

The detection of differential item functioning (DIF) is crucial to ensuring the fairness of tests, which is an important approach to test if the item is biased or not. IRT techniques are "gold standard" in DIF detection. This study aims to generalize the theory and the process of detection of differential item functioning under item response theory, and compares the procedures of Raju's area measures, Lord's chi-square and the likelihood ratio test in detail. It detects sexual DIF for all items in one HSK(Elementary-Intermediate) examination paper according to true data from an examination in 2003. The results from these three procedures show not very close agreement. The advantages and disadvantages of these three procedures are discussed, and the causes of DIF on sex are also mentioned.