Factorial validity of the Theory of Mind Inventory-2 in children with autism spectrum disorder.

The Theory of Mind Inventory-2 appears to be a promising measure assessing children's theory of mind. However, a lack of evidence on its factorial structure interferes with score interpretation. To examine whether the currently-available structures of the Theory of Mind Inventory-2 are robust and replicable, confirmatory factor analysis was conducted in 242 children with autism spectrum disorder aged 3-12 years. Two possible structures were examined: the development-based structure, which reflects children's developmental stages of theory of mind (early, basic, and advanced), and the content-based structure, which represents children's social-related functions (emotion recognition, mental state term comprehension, and pragmatics). Four fit indices were adopted simultaneously to examine the model fit of both structures. We found that the development-based structure had a better model fit and was further modified. After modification, the development-based structure showed an improved model fit, but it was not completely acceptable in all fit indices. These findings suggest that the scores of the Theory of Mind Inventory-2 are more appropriate for reflecting children's developmental stages of theory of mind than the social-related functions in children with autism spectrum disorder. However, the domain scores should be cautiously interpreted because the model fits were not completely acceptable. For further revisions, ambiguous terms (e.g., "want" and "need") and confusing concepts assessed by the items (e.g., "recognize" and "understand") could be specified for a better representation of children's developmental stages of theory of mind. Moreover, further validations are needed. LAY SUMMARY: In children with autism spectrum disorder, the scores of the Theory of Mind Inventory-2 are more appropriate for indicating the developmental stages of theory of mind (early, basic, and advanced) than for indicating the social-related functions (emotion recognition, mental state term comprehension, and pragmatics). However, since the factorial validity was not completely acceptable, the domain scores should be interpreted cautiously. Moreover, further revisions and validations are warranted.