Development of a theory of mind assessment for children using multidimensional Rasch modelling.

Abstract

Existing theory of mind (ToM) measures for children meet challenges from three perspectives. Developmentally, they lack items covering the entire spectrum of ToM abilities, namely, the early, basic and advanced levels. Dimensionally, most measures are unidimensional, not distinguishing between cognitive and affective ToM. Practically, most measures rely heavily on verbal abilities or lack engaging formats. This study aimed to address these critical issues by developing a Theory of Mind Assessment (ToMA). The items were generated based on classical scenarios spanning all developmental levels. The responses from 574 neurotypical children aged 37 to 194 months were analysed using the multidimensional Rasch model. Ten items showed satisfactory model fits when classified into cognitive (4 items) and affective (6 items) domains, with 16 misfit items excluded. Two items seemed easier for girls and two for boys, and the item difficulties were adjusted. The mean Rasch person reliabilities were 0.78 and 0.81. The scores exhibited small and high correlations with subjective and objective criteria. The newly developed measure may offer reliable, valid and sex-unbiased assessments while satisfying developmental, multidimensional and practical requirements. It seems promising for application in clinical and research settings and is worthy of future refinement and validation to provide high-quality ToM assessment.