ABSTRACT
This chapter focuses on the end product of the scoring process—the scores reported to the test takers. Dorans and Holland suggested examining the subpopulation linking functions versus the total population linking function. Brennan argued that the differences of linking functions should be examined between pairs of subpopulations (e.g., males versus females). Measurement invariance refers to the degree to which an instrument measures an entity in the same way across different subpopulations. It is closely related to the same construct requirement of score equating, as well as the equity requirement. Score equity assessment, like DIF, is straightforward given the correct data collection design and sufficient data to conduct score linkings in subgroups. While a single group taking both tests is preferable, large equivalent groups taking either test is often adequate. The non-target part of the population may include test takers who are not proficient enough in the language of the test.
