The emotional and cognitive components of trait anxiety: A latent state-trait model

Abstract

A cross-validation of the latent state-trait anxiety (LSTA) model proposed by Steyer, Majcen, Schwenkmezger & Buchner (Anxiety Research, 1, 281–299, 1989) for test halves created from items of the German version of the State-Trait Anxiety Inventory (Laux, Glanzmann, Schaffner & Spielberger, 1981) fails. It is argued that the reasons for this failure are the heterogeneity of the items of the trait anxiety inventory and the fact that the trait inventory contains more 'cognitive' items than the state inventory. These arguments are supported by further analyses of the cross-validation data. If the trait test halves are replaced by new test halves which are constructed only from 'emotional' items, the LSTA model does not have to be rejected. If, however, the trait test halves are constructed only from 'cognitive' items, the model again has to be rejected. Theoretical parsimony is suggested as a general methodological principle. This principle is realized in the latent state-trait anxiety model because, instead of assuming different latent traits for the state and trait test halves, this model postulates only one single latent trait which conjoins the two different kinds of manifest variables.