NotesFAQContact Us
Search Tips
ERIC Number: ED462397
Record Type: Non-Journal
Publication Date: 1996-Jan
Pages: 33
Abstractor: N/A
Reference Count: N/A
A Review of Methods for Detection of Test and Item Bias.
Breunig, Nancy A.
Few issues have provoked as much controversy as the methods for detecting item and test bias. A recent illustration of the controversy surrounding this issue could be seen in the emotional reactions to the publication of "The Bell Curve." This paper reviews methods of evaluating both item and test bias. Small heuristic data sets are provided to illustrate required calculations. Test bias is presented in the context of bias in selection, and models based on regression and decision-theoretic approaches are discussed. Item bias is discussed by reviewing latent trait, chi-square, and item difficulty techniques. There is no definitive method for fair selection or for detection of biased items. The test user needs to consider the purpose of the test and the inferences that will be made from the test and decide based on these considerations the methods or combination of methods that would be best to use. In particular, with regard to item bias, it may be best to use a combination of methods and see if results from the different approaches are in agreement. (Contains 3 tables, 4 figures, and 15 references.) (SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A