Developing Ideal Intermediate Personality Items for the Ideal Point Model

Abstract

The importance of intermediate items has been overlooked since the emergence of dominance-based Likert-type scales. The current study aims to advance our understanding of the psychometric properties of intermediate items by showing that they can be successfully calibrated by an ideal point model. A student sample and an MTurk sample were selected to answer personality scales with intermediate items included. Results showed that personality scales with intermediate items demonstrated satisfactory model fits to the ideal point model, but not to the dominance model. When analyzed with the ideal point model, intermediate items provided more information than extreme items for respondents with high latent trait levels. Among four proposed domains of intermediate items (Frequency, Average, Condition, and Transition, “FACT”), the Average domain was consistently found to exhibit the best properties. The proportion of intermediate items in the scale also affected the model fit and the validity of the scale.