The Regression BIAS Method: A Unified Approach for Detecting Item BIAS and Selection BIAS

Abstract
Although several different procedures are available for detecting item bias and selection bias, there does not appear to be a single procedure for detecting both types of bias. It is proposed that the Regression Model, which is popular in selection bias research, be adopted for item bias detection, thereby providing a common framework for studying both types of bias. In keeping with the commonality, the new method will be referred to as the Regression Bias method. An empirical test of the Regression Bias method as well as a comparison of the Regression Bias method with the other commonly used item bias detection methods are presented.

This publication has 21 references indexed in Scilit: