The Nedelsky Model For Multiple Choice Items

Traditionally, multiple choice (MC) items are scored binary; one point is earned when the correct answer is chosen and none when any of the incorrect options ( called "distractors") is chosen. This facilitates data analysis but it also entails loss of information (e.g., Levine and Drasgow, 1983). There have been various attempts to include the incorrect options into an IRT model (e.g., Bock, 1972; Thissen and Steinberg, 1984). Here, we discuss a novel model called the Nedelsky model (NM). We focus on the theoretical properties of the NM. Estimation and testing are described briefly. An application to real data can be found in Verstralen (1997) and Verstralen and Verhelst (1998), who invented the model.