The Psychometric Evaluation of the Theory of Mind Inventory

Background: Despite the tremendous activity in the field and developments in Theory of Mind (ToM) assessment procedures, the measurement of ToM has traditionally been associated with several challenges. These include a lack of content and social validity, ceiling effects when mentalizing is relatively good, and a variety of situational, motivational, cognitive, and linguistic factors that often complicate the administration of tests that rely on direct measures of child performance (Hutchins et al. 2008). In addition, while several ToM tests have been developed to assess the social cognition of individuals with HFA and Asperger Syndrome, there are no measures currently in existence that are appropriate for the measurement of ToM among those the most severe disabilities. Thus, children with ASD who are nonverbal or have limited language skills are commonly excluded from research protocols.

Objectives:  The purpose of this study was to assess the psychometric properties of a new parent-informant measure of ToM: one that would be appropriate for younger and older children and those who vary widely in their cognitive and linguistic profiles.

Methods:  Two cross-sectional studies examined the psychometric properties of the Theory of Mind Inventory (ToMI). In Study One, 135 caregivers completed the ToMI for children (ages 3 through 17) with autism spectrum disorder (ASD). In Study Two, data were collected for 124 typically developing children (2 through 12 years).  

Results:  Findings revealed excellent test–retest reliability and internal consistency for both samples (r = .86, p < 01 for both samples). A number of other measures (e.g., receptive vocabulary, a test of ToM performance using direct assessment) provided evidence of convergent validity (p < .05). A contrasting-groups method of construct validation also revealed between group differences where 1) higher ToMI scores were obtained for typically developing children compared to children with ASD (p < .05), and 2) ToMI scores were positively correlated with child age in the typically developing sample (r = .72, p < .05). Finally, a Principle Components Analysis revealed three subscales related to the complexity of ToM understanding. These appeared to reflect a general developmental progression seen in typically developing children. These were termed: early, basic, and advanced skills.

Conclusions: The ToMI performed well in all examinations of reliability and validity in both typically developing children and individuals with ASD. The advantages of the ToMI

over traditional ToM assessment procedures include the fact that responses are not scored on a dichotomous pass/fail basis but rather in terms of continuum of confidence that a child possesses a ToM understanding. A range of scores can be obtained which are construed as interval in nature and can be submitted appropriately to the most powerful statistical procedures. In addition, the ToMI is quick and easy to administer and it is not complicated by situational factors and child motivation, linguistic, or cognitive performance factors. Finally, the automated scoring and norms are now available for the ToMI.