RESUMO
The State-Trait Inventory of Cognitive and Somatic Anxiety (STICSA) is a commonly administered self-report instrument of state-trait cognitive and somatic anxiety. Extant research has consistently supported the intended oblique two-factor scoring structure for the STICSA. However, this model assumes that population-level data have (or approximate) a simple structure and that item-level variance is unidimensional. These assumptions may not be tenable and have unintended consequences for STICSA subscore interpretation. Consequently, we tested these assumptions by fitting confirmatory and exploratory structural equation models to STICSA scores for a diverse sample of college students enrolled at a large Southwestern university in the United States (n = 635). Results indicated that cognitive and somatic factors are not equally robust and that STICSA items appear to measure a nonnegligible mixture of both latent cognitive and somatic anxiety. It is recommended that future research use exploratory structural equation model in tandem with CFA to directly model data complexity.