ABSTRACT
Many-facet Rasch measurement (MFRM) refers to a class of measurement models that simultaneously consider multiple components or facets involved in rater-mediated assessments, as when raters score the performances of test takers on a set of criteria using rating scales. This chapter presents MFRM as a principled approach to the analysis and evaluation of such assessments. As an illustration, ratings of writing performances are used to elaborate on basic concepts and procedures within the context of language assessment. The key issues addressed in the chapter are: (a) measuring rater severity in the same frame of reference as test taker ability and criterion difficulty, (b) studying the consistency of ratings across test takers and criteria, and (c) providing test taker scores that compensate for differences in rater severity. The chapter also investigates the use and functioning of rating scales, briefly discusses recent advances in applied MFRM research, and touches upon some controversial methodological issues.
