Evaluation of Computer Simulated Baseline Statistics for Use in Item Bias Studies

Publisher Website
Google Scholar
Abstract
Though item bias statistics are widely recommended for use in test development and test analysis work, problems arise in their interpretation. The purpose of the present research was to evaluate the validity of logistic test models and computer simulation methods for providing a frame of reference for item bias statistic interpretations. Specifically, the intent was to produce simulated sampling distributions of item bias statistics under the hypothesis of no bias for use in determining cut-off points to provide guidelines for interpreting item bias statistics obtained with actual test data. The results provided support for the basic data simulation approach used in the study. Real and simulated distributions for three item bias statistics when bias was not present were very similar and the minor differences that were found between the distributions had little effect on the interpretations of item bias statistics obtained with actual test data. Seven steps for applying the method of computer-simulated baseline statistics in test development settings were outlined in the paper.

This publication has 9 references indexed in Scilit: