A note on the odds ratio DIF index

Publication Date

2021

Document Type

Article

Abstract

Jin, Chen, and Wang (2018) proposed using the log of the odds ratio (OR) as a DIF index for item responses that follow the Rasch model. The OR is relatively simple to calculate and does not require grouping examinees by ability. This index was more robust to unbalanced DIF than the Mantel-Haenszel (MH) or logistic regression (LR) procedure. The authors cautioned against using this index with data that does not follow the Rasch model. This brief commentary explains one problem with using the OR method with 2PL or 3PL item responses: If the response functions are identical for both groups, when the group mean difference is large the log of the OR will vary with the value of the item parameters.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Share

COinS