ERIC Number: ED206650
Record Type: Non-Journal
Publication Date: 1980-Dec
Reference Count: N/A
Latent Trait Theory Applications to Test Item Bias Methodology. Research Memorandum No. 1.
Osterlind, Steven J.; Martois, John S.
This study discusses latent trait theory applications to test item bias methodology. A real data set is used in describing the rationale and application of the Rasch probabilistic model item calibrations across various ethnic group populations. A high school graduation proficiency test covering reading comprehension, writing mechanics, and mathematics was administered to 1,042 white and 11,441 black students in a large west coast school district. Using UCON estimation procedures for item difficulties, item plots for each ethnic group by the three separate subtests were prepared. The derivation of acceptable tolerance limits is described and applied to the current data set, wherein a biased item is revealed. The mathematics are given although their derivation is not described except when required for completeness. (Author/BW)
Publication Type: Reports - Research
Education Level: N/A
Authoring Institution: Oakland Unified School District, CA. Dept. of Research and Evaluation.; Los Angeles County Superintendent of Schools, CA. Div. of Program Evaluation, Research, and Pupil Services.