Using contemporary psychometric methods to construct and validate scores on a short-form version of the Psychological Strain Scales

Abstract

Background: The 40-item Psychological Strain Scales (PSS) was developed to measure four dimensions of psychological strains based on the Strain Theory of Suicide. Although the PSS has been increasingly used in recent years, the dimensionality and scoring of the PSS remained to be thoroughly examined using the Item Response Theory (IRT) approach. Methods: Data came from a large sample of 11,412 Chinese undergraduate students. A bifactor-IRT model, specifying one general strain factor and four specific strains factors, was examined for fit to the sample data. A detailed item analysis, with analysis of the differential item functioning (DIF) of the items across gender, was undertaken to evaluate the dimensionality of the PSS. The associations among the PSS scale scores with scores on the concurrent measures, assessing psychache and suicidal behaviors, were examined. Results: IRT-derived specific bifactor indices showed that the PSS was unidimensional, and thus the PSS total scores should be reported. Unidimensional subset of 5 items identified (Item 9, Item 12, Item 14, Item 16, and Item 20), using bifactor-IRT modeling and incremental validation, were selected to construct a potential short form of the PSS (PSS-SF). The PSS-SF scale scores demonstrated strong psychometric properties and associations with scores on the concurrent measures assessing relevant constructs. Limitations: This study used cross-sectional data from a non-clinical sample of Chinese undergraduate students. Conclusions: The PSS-SF should be considered as a unidimensional instrument with potential in enhancing our understanding and measurement of psychological strains with reduced response burden.