Validation of the Korean Version of the Epworth Sleepiness Scale Using Rasch Analysis

초록

The Epworth Sleepiness Scale is widely used to assess daytime sleepiness, but validation in Korean populations has been limited. Prior studies targeted specific groups and relied on classical test theory, which is sample-dependent and provides limited item-level insights. To address these limitations, this study applied Rasch analysis to evaluate the psychometric properties of the Korean version of the Epworth Sleepiness Scale. This study applied Rasch analysis to evaluate the Korean version of the scale. Data from 751 adults (mean age = 46.2 years; 376 men, 375 women) were analyzed using the Rasch rating scale model. Model - data fit was acceptable (mean Infit = 1.03 +/- 0.14; Outfit = 0.98 +/- 0.10), unidimensionality was supported (first contrast = 1.68), and local independence was confirmed. Item difficulties ranged from -1.86 to 2.28 logits, with high item separation (17.33) and reliability (1.00). Person ability averaged -1.9 logits (SD = 1.52), but separation (1.69) and reliability (0.74) were below recommended thresholds, indicating limited discriminative power. Rating scale categories were properly ordered (thresholds = -2.48, 0.39, 2.09). Differential item functioning analyses revealed no item bias by gender; however, several items showed age-relateddifferences across the age groups. The Korean version demonstrated strong structural validity and stable measurement performance at the item level, although person-level reliability was modest. The presence of age-related differences and reduced measurement sensitivity for individuals with low levels of sleepiness highlight areas for refinement. Further validation in clinical and age-specific groups is recommended.

키워드