ABSTRACT
This chapter looks at methods from measurement modeling that have been developed to address sources of unfairness that enter in the inferential step of synthesizing evidence across tasks. The data may represent common tasks and procedures for all examinees or they may include principled variations for some or all tasks, as discussed in Chapter 9. Several approaches are described, illustrated, and discussed from a sociocognitive perspective: person-fit analysis, with and without additional person and task information; differential item functioning (DIF); and item response theory mixture modeling.
