Psychometric properties of the Epworth Sleepiness Scale: A factor analysis and item-response theory approach

ABSTRACT

The purpose of this study is to examine the psychometric properties of the Epworth Sleepiness Scale (ESS) in two languages, German and English. Students from a university in Austria (N = 292; 55 males; mean age = 18.71 ± 1.71 years; 237 females; mean age = 18.24 ± 0.88 years) and a university in the US (N = 329; 128 males; mean age = 18.71 ± 0.88 years; 201 females; mean age = 21.59 ± 2.27 years) completed the ESS. An exploratory-factor analysis was completed to examine dimensionality of the ESS. Item response theory (IRT) analyses were used to provide information about the response rates on the items on the ESS and provide differential item functioning (DIF) analyses to examine whether the items were interpreted differently between the two languages. The factor analyses suggest that the ESS measures two distinct sleepiness constructs. These constructs indicate that the ESS is probing sleepiness in settings requiring active versus passive responding. The IRT analyses found that overall, the items on the ESS perform well as a measure of sleepiness. However, Item 8 and to a lesser extent Item 6 were being interpreted differently by respondents in comparison to the other items. In addition, the DIF analyses showed that the responses between German and English were very similar indicating that there are only minor measurement differences between the two language versions of the ESS. These findings suggest that the ESS provides a reliable measure of propensity to sleepiness; however, it does convey a two-factor approach to sleepiness. Researchers and clinicians can use the German and English versions of the ESS but may wish to exclude Item 8 when calculating a total sleepiness score.

KEYWORDS:

• Subjective sleepiness
• psychometric analysis
• Epworth Sleepiness Scale
• factor analysis
• IRT analysis

Acknowledgements

We thank the research team at Clemson University for their assistance in data management and initial data analyses including: S. A. Bryant, E. E. Howard, D. M. Morris, A. L. Old, T. A. Roberson, E. G. Rummel, K. L. Sullivan, E. C. Szubski and J. H. Walker.

This research was partially completed as part of a Fulbright Scholar award to JJP at the University of Vienna, Vienna, Austria.

Declaration of interest

The authors declare that they have no conflict of interest.