Psychometric Evaluation of the Insomnia Severity Index in 570,295 Chinese Adolescents: A Bifactor Item Response Theory Analysis

Insomnia is a prevalent and significant issue among adolescents, yet its accurate measurement remains challenging. The Insomnia Severity Index (ISI), a widely used self-report measure for assessing insomnia symptoms, lacks well-established factor structure and psychometric properties in Chinese adolescents. To address this gap, our study employed bifactor modeling and item response theory (IRT) to explore the dimensionality and reliability of the ISI in a large sample of 572,095 Chinese adolescents.

1. Factor Structure Investigation:
2. • We tested five alternative factor models using confirmatory factor analysis (CFA).
   • The best-fitting model was a three-factor structure with three correlated factors.
   • However, bifactor CFA revealed that a bifactor two-factor model, excluding item 4, exhibited superior fit indices and accounted for more common variance in the ISI items.
3. Reliability and Discrimination:
4. • Based on the bifactor two-factor model, we employed bifactor multidimensional IRT (MIRT) to assess item parameters and the test information function of the ISI.
   • The results indicated that the ISI demonstrated high reliability and discrimination for measuring individuals with average to high levels of insomnia severity.
   • However, its reliability and discrimination were lower when assessing individuals with very low or very high levels of insomnia severity.
5. Implications:
6. • These findings have important implications for the measurement and diagnosis of insomnia in Chinese adolescents.
   • Researchers and clinicians should consider the specific context and severity levels when utilizing the ISI in this population.

In summary, the ISI proves to be a reliable and valid measure for assessing insomnia severity in Chinese adolescents, but its performance varies across different severity levels. Future research should explore additional factors influencing insomnia and refine measurement tools accordingly.