ABSTRACT
It is a pleasure to compliment the authors on their excellent reviews of the methodology for detecting and estimating differential item functioning (DIF). They give a clear account of the three major techniques for studying DIF— standardization, the Mantel-Haenszel statistic, and comparisons of item response functions (IRT). Although the papers divide along the lines of nonparametric versus parametric approaches to data analysis, the authors are entirely fair in their respective positions. And in the context of the Scholastic Aptitude Test, which Dorans is assuming, these positions are not all that far apart. For a test containing such a large number of items as the SAT, the standardization and IRT approaches are essentially equivalent.
