Investigation of the factor structure of the Turkish version of the State-Trait Anxiety Inventory

Abstract

Aim: Confirmatory factor analysis (CFA) and Rasch Analysis are commonly used methods to examine the structure of the psychological scales. In this study, it is aimed to evaluate the factor structure Turkish version of the State-Trait Anxiety Inventory by using statistics based on the Rasch model and CFA.Methods: The State-Trait Anxiety Inventory (STAI) was used for the analysis. Of the study group, 186 (46.5%) were male and 214 (53.5%) were female. Unidimensionality was investigated using a Rasch-based principal component analysis (PCA) of the residuals, chi-square tests, item fit statistics, and other statistics. CFA has also been applied to test the hypothesis of a one-factor solution.Results: The item-trait interaction chi-square statistic was 342.344 for the state scale (p<0.001) and 381.247 for the trait scale (p<0.001). For the state scale, 16.00% of the t-tests for the PCA weresignificant at the 5% level, while 19.50% were significant for the trait scale. The fit residuals of items 4, 8, and 18 on the state scale were over the +2.5 threshold, while the fit residuals of items 23, 24,and 34 on the trait scale were above the +2.5 threshold. Similarly, the scale structure evaluated by CFA was conditioned to be inadequate goodness-of-fit.Conclusion: This study found that neither the trait nor the state scale of the STAI met the unidimensionality assumption. Consequently, both the Rasch analysis and CFA have been verified as succeeding tools in assessing the scale sub-dimensions and determining whether the response items can be utilized for a total scale score.