The State-Trait Anxiety Inventory: Shadows and Lights on its Construct Validity

Past studies on the factor validity of the Trait subscale of the Spielberger's State-Trait Anxiety Inventory (STAI-T) do unanimously agree on its structure. In fact, researchers are still debating whether the STAI-T is unidimensional or multidimensional. Our aim was to clarify what the STAI-T measures. The STAI-T, the Beck Depression Inventory-II, the Teate Depression Inventory, and the Beck Anxiety Inventory were administered to 1124 psychiatric outpatients and to 877 healthy subjects. A confirmatory factor analysis was performed in order to compare various models in the literature. The internal consistency and convergent and discriminant validity of the STAI-T as well as its factorial subscales were assessed. The one-construct two-method (i.e., the STAI-T measures one substantive anxiety construct plus artifacts due to negative-positive item polarity) and the bifactor (i.e., the STAI-T comprises two first-order specific factors ["Anxiety" and "Depression"] and one first-order general factor) models were the best-fitting solutions for the STAI-T in both the clinical and nonclinical samples. The STAI-T total score correlated more strongly with measures of depression than with a concurrent measure of anxiety. The STAI-T should be considered a measure of general negative affect, including specific aspects of cognitive anxiety and depression together.