Item Bias Elimination Models for Test Reliability and Validity

Controversy over testing practices has occurred at various times since the inception of widescale testing. The issue of bias in measurement has come to the forefront. Questions of test bias are closely related to questions of test validity. A test possesses validity if it measures what it purports to measure and invalidity if it does not. Bias or systematic error is a kind of invalidity that arises relative to groups. In general, a test is biased against or for a particular group if it under-predicts or over-predicts, respectively their performance on the criterion of interest relative to some other group(s). Bias typically arises when a given identifiable group scores low or high on a test relative to some other group(s), other things being equal. Bias is a major factor for tests considered unfair, inconstant, and contaminated by extraneous factors.

Autorentext

Jose Q. Pedrajita is a faculty of the Division of Educational Leadership and Professional Services of the University of the Philippnes College of Education. He is an Associate Professor of Educatioal Research and Evaluation.