Finding Optimal Number of Rating Scale Categories in Multiple Mini-Interviews Using the Many-Facet Rasch Measurement Model

• In Event: Use of Various Rasch Measurement Models

Abstract

The medical, pharmacy and optometry programmes of the University of Auckland first used a common MMI in 2014 as a tool to select suitable candidates for these programmes.
Data were fitted to the Rasch model using Facets and initial results reveal good fit.
The MMI used a 10-point rating scale. In this study, alternative scoring models were formed by combining adjacent categories. Fit statistics, person separation and reliability statistics, average category measures and thresholds were compared using the guidelines proposed by Linacre (2002) and Lopez (1996).
Results suggest that the 4-category model will result in best data fit. However, the 5-point scoring model might be a better alternative to the original scoring model as its categories have clearly stated definitions.

Authors

• Vernon Mogol, University of Auckland

• Warwick Bagg, University of Auckland

• John Shaw, University of Auckland

• Phillippa Poole, University of Auckland

• John Monigatti