Construct Validity and Measurement Invariance of the eHEALS in a Diverse US High School Sample

• Published in: Journal of media psychology 2024-05, Vol.36 (3), p.188-198
• Main Authors: Joseph, Patrece L., Fleary, Sasha A.
• Format: Article
• Language: English
• Subjects: Adolescent Development; Classical Test Theory; Construct Validity; Ethnic Identity; Female; Health Literacy; High Schools; Human; Internet; Male; Measurement Invariance
• ISSN: 1864-1105; 2151-2388
• DOI: 10.1027/1864-1105/a000399

Although the eHealth Literacy Scale (eHEALS) is a widely used measure of eHealth literacy (eHL), it has not been validated in a diverse sample of US adolescents. This study assessed the construct validity and item and measurement invariance of the eHEALS using both Rasch analyses and classical test theory (CTT). Data were collected from adolescents at a US Northeastern high school (N = 355, 64.9% were 16-18-years old, 54.8% male, 76.4% from a racially minoritized group, and 50.5% Latinx). Adolescents completed a survey that included demographic variables (i.e., age in years, gender, and ethnicity) and the eHEALS. Using Rasch and CTT methodologies, the dimensionality of the eHEALS and item and measurement invariance across gender, age, and ethnicity were assessed. The unidimensional structure of the eHEALS as well as item and measurement invariance across gender, age, and ethnicity were supported. Rasch analyses also suggested that responses to eHEALS items can be summed to determine eHL abilities, although there was a ceiling effect. The eHEALS is a unidimensional measure of eHL that may provide information in health education settings about adolescents' eHL skills. It performs similarly across male and female as well as older and younger adolescents. However, given the ceiling effect and changes in adolescents' Internet use since its development, future studies should revise the eHEALS to better capture adolescents' eHL abilities. Future studies should also examine item and measurement invariance by age (in years), race, family income, and longitudinally to determine the usefulness of the measure across groups and time.