Reliability and factor structure in an adolescent sample of the Dutch 20-item Toronto Alexithymia Scale.

This study investigated the psychometric properties of the 20-item Toronto Alexithymia Scale (TAS-20) in an adolescent sample (N = 406, ages 12 to 17). This is rarely done even though the TAS-20 is used in adolescent research. Five published factor models were tested. For good fitting models, a second-order model with alexithymia as a higher-order factor and metric invariance across sex and age groups was tested. Confirmatory factor analyses showed that the original three-factor model and a four-factor model provided acceptable fit. Both models were invariant across sex, but not across age. Second-order models did not provide good fit. Reliability was good for the "Difficulty identifying feelings" subscale and acceptable for the "Difficulty describing feelings" subscale, but not for the "Externally oriented thinking" subscale. Measuring alexithymia with the TAS-20 in adolescents thus seems problematic, especially in younger age groups.