Objective measurement: How rasch modeling can simplify and enhance your assessment

Abstract

Although Rasch modeling is a powerful psychometric tool, for novices its functionality is a “black box.” Some evaluators still prefer classical test theory (CTT) to Rasch modeling for conceptual clarity and procedural simplicity of CTT, while some evaluators conflate Rasch modeling and item response theory (IRT) because many texts lump both together. To rectify the situation, this non-technical, concise introduction is intended to explain how Rasch modeling can remediate the shortcomings of CTT, and the difference between Rasch modeling and item response theory. In addition, major components of Rasch modeling, including item calibration and ability estimates, item characteristic curve (ICC), item information function (IIF), test information function (TIF), item-person map, misfit detection, and item anchoring, are illustrated with concrete examples. Further, Rasch modeling can be applied into both dichotomous and polytomous data, and hence different modeling methods, including normal ogive model, partial credit model, graded response model, nominal response model, are introduced. The procedures of running these models are demonstrated with SAS and Winsteps.