A Psychometric Evaluation of the Swedish Translation of the Perceived Stress Scale: A Rasch Analysis

Abstract

Abstract Background: Stress reflects physical and psychological reactions to imposing demands and is often measured using self-reports. A widely-used instrument is the Perceived Stress Scale (PSS), intended to capture more general aspects of stress. A Swedish translation of the PSS is available but has not previously been examined using modern test theory approaches, such as Rasch analysis. Methods: The current study used data from 793 university students to investigate the dimensionality of different version of the PSS (14, 10, and 4 items) as well as potential response patterns among the participants. Results: The current study demonstrates that the PSS-14 has two separate factors, divided between negatively worded items (stress) and positively worded items (self-efficacy), although with only the negative subscale exhibiting good reliability. Response patterns were analyzed using Differential Item Functioning analysis, which did not find an influence of gender on the perceived difficulty of the items, but for age regarding the positive subscale (items 6 and 9). The PSS-10 also demonstrated adequate reliability for the negative subscale, but the PSS-4 was not deemed suitable as a unidimensional scale. Conclusions: Based on the results, none of the versions of the PSS should be used by sum-scoring all of the items. Only the negative items from the PSS-14 or PSS-10 can be used as unidimensional scales to measure general aspects of stress. Content validity is discussed, questioning the relevance of anger and being upset when measuring more general aspects of stress. Finally, a table to convert the PSS-7 ordinal sum scores to interval level scores is provided.