Does It Work for Everyone? The Influence of Demographic Variables on Statistical Reliability

Recent developments have highlighted the importance of tailored health education efforts. However, little research has explored differential functioning of survey items for diverse populations. This work explores differences in statistical reliability for multiple scales across demographic groups. U...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Andrew C. Pickett, Danny Valdez, Adam E. Barry
Formato: article
Lenguaje:EN
Publicado: New Prairie Press 2020
Materias:
Acceso en línea:https://doaj.org/article/b4876be366b2400bbe4c96994160c8f9
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
Descripción
Sumario:Recent developments have highlighted the importance of tailored health education efforts. However, little research has explored differential functioning of survey items for diverse populations. This work explores differences in statistical reliability for multiple scales across demographic groups. Understanding such differences is important in health research, given the rapid shifts occurring in global demographics. Study data were collected from eight years of the National College Health Assessment (n = 885,084), a large-scale annual survey of U.S. university students. Meta-analytic reliability generalization was used to compare reliability of two scale measures for multiple demographic groups. In nearly all cases, there were statistically significant differences in reliability across demographic groups. Researchers should consider relative functioning of any scale employed in their work. For certain demographic groups, various scales may not be sufficiently reliable. However, this may be obfuscated in larger samples, containing large numbers of individuals for whom the scale is sufficiently reliable. We suggest a thorough subsets analysis of data to ensure uniform functioning of items prior to use. Just as health interventions should be tailored to populations of interest, so too must research methods and tools.