Abstract
Objective: Non-Parametric IRT methods were applied to 127 clinicians' ratings of their patients using the Dutch Impact Message Inventory–Circumplex (IMI–C) to develop the IMI-C Short. Method: Four items from each octant subscale of the IMI-C were selected to maximally differentiate individuals along the continuum of impact messages. Results: Using larger samples (patients: N = 700, 812; raters: N = 42, 85, respectively), IRT-based reliability was generally comparable between the brief and parent subscales. Classical Test Theory-based reliability was adequate for the brief subscales and they converged with their parent subscales. Since the IMI-C is purported to represent the circular arrangement of impact messages, the fits of the parent and abbreviated scales were assessed and found to be good and not differ significantly. Conclusion: The new IMI-C Short offers advantages over the full-length IMI-C.

This publication has 28 references indexed in Scilit: