ERIC Number: ED178583
Record Type: Non-Journal
Publication Date: 1979
Reference Count: N/A
The Problem of Item Bias: A Comparison of Techniques. Final Report.
Rudner, Lawrence M.; And Others
The psychometric rationales of six techniques for quantifying estimated item bias were reviewed: analysis of variance, transformed item difficulties, chi square, item characteristic curve theory, factor analysis, and distractor response analysis. The limitations and advantages of the approaches were described in terms of their underlying assumptions and psychometric soundness. By using Monte Carlo generated item response data, theis research also sought to determine the effectiveness, sufficiency and similarity of select biased item detection techniques. Experimental variables included: different proportions of bias for items in the item pool, different types of generated item bias, and different test lengths. The three parameter item characteristic curve (ICC) theory and the five interval chi-square techniques were found to produce satisfactory results under all test conditions. The transformed item difficulties technique was found to be sensitive to bias in item difficulty but insensitive to bias in item discrimination. The one parameter ICC theory technique was unsatisfactory under all conditions. (Computer program subroutines and bibliographic references are appended). (Author/CTM)
Publication Type: Reports - Research
Education Level: N/A
Sponsor: National Inst. of Education (DHEW), Washington, DC.
Authoring Institution: N/A