Detecting Differential Item Functioning Using Cognitive Diagnosis Models: Applications of the Wald Test and Likelihood Ratio Test in a University Entrance Examination

2021 ◽  
pp. 1-23
Author(s):  
Roghayeh Mehrazmay ◽  
Behzad Ghonsooly ◽  
Jimmy de la Torre
Keyword(s):  
Differential Item Functioning ◽  
Likelihood Ratio ◽  
Likelihood Ratio Test ◽  
Wald Test ◽  
Cognitive Diagnosis ◽  
Entrance Examination ◽  
Ratio Test ◽  
Cognitive Diagnosis Models ◽  
Item Functioning ◽  
University Entrance Examination

The Effects of Purification and the Evaluation of Differential Item Functioning With the Likelihood Ratio Test

Methodology ◽  
2012 ◽  
Vol 8 (4) ◽  
pp. 134-145 ◽  
Author(s):  
Fabiola González-Betanzos ◽  
Francisco J. Abad
Keyword(s):  
Differential Item Functioning ◽  
Likelihood Ratio ◽  
Likelihood Ratio Test ◽  
Type I Error ◽  
Error Rates ◽  
Ratio Test ◽  
Type I ◽  
Two Stage ◽  
Item Functioning ◽  
Size Type

The current research compares the effects of several strategies to establish the anchor subtest when detecting for differential item functioning (DIF) using the IRT likelihood ratio test in one- and two-stage procedures. Two one-stage strategies were examined: (1) “One item” and (2) “All other items” used as anchor. Additionally, two two-stage strategies were tested: (3) “One anchor item with posterior anchor test augmentation” and (4) “All other items with purification.” The strategies were compared in a simulation study, where sample sizes, DIF size, type of DIF, and software implementation (MULTILOG vs. IRTLRDIF) were manipulated. Results indicated that Procedure (1) was more efficient than (2). Purification was found to improve Type I error rates substantially with the “all other items” strategy, while “posterior anchor test augmentation” did not yield a significant improvement. In relation to the effect of the software used, we found that MULTILOG generally offers better results than IRTLRDIF.

Sign in / Sign up

Export Citation Format

Share Document