Examining the psychometric properties of the CEFIS-AYA using item response theory
Meghan M Howe,Steven A Miller,Susan Tran,Joanna Buscemi,Lindsey Bugno,Rachel N Greenley
DOI: https://doi.org/10.1093/jpepsy/jsae084
2024-10-10
Abstract:Objective: The COVID-19 Exposure and Family Impact Scale, Adolescent and Young Adult Version (CEFIS-AYA; Schwartz, L. A., Lewis, A. M., Alderfer, M. A., Vega, G., Barakat, L. P., King-Dowling, S., Psihogios, A. M., Canter, K. S., Crosby, L., Arasteh, K., Enlow, P., Hildenbrand, A. K., Kassam-Adams, N., Pai, A., Phan, T. L., Price, J., Schultz, C. L., Sood, E., Wood, J., & Kazak, A. (2022). COVID-19 exposure and family impact scales for adolescents and young adults. Journal of Pediatric Psychology, 47, 631-640. https://doi.org/10.1093/jpepsy/jsac036) was developed to assess the pandemic's effects on adolescents and young adults (AYA). Via principal component analysis, measure developers examined the structure and reliability of the CEFIS-AYA and identified seven exposure and five impact components. This study built upon prior work through use of item response theory (IRT) models to characterize the dimensionality of the CEFIS-AYA, determine the strength of relations between items and underlying trait(s), and examine associations between trait scores and pandemic-related distress. Methods: This was a secondary analysis of data collected between July 2020 and July 2021 from three studies of emerging adults (ages 18-29; N = 834). Results: The CEFIS-AYA structure was multidimensional, with the strongest support for five traits. Trait 1 represented pandemic impact on social/emotional functioning and self-care. Trait 2 reflected other pandemic disruptions. Trait 3 represented pandemic disruptions to education and/or other milestones. Trait 4 represented pandemic impact on physical well-being. Trait 5 assessed pandemic disruptions to work/financial circumstances. Item loadings and parameters indicated variability in how consistently trait level was associated with item endorsement. Trait scores did not predict distress, except that increases in Trait 3 were associated with lower distress. Conclusions: The present study examined the psychometric properties of the CEFIS-AYA among emerging adults using a statistical framework better suited for modeling categorical data. The identified dimensional structure was relatively consistent with the initial psychometric evaluation of the CEFIS-AYA, albeit more parsimonious. However, replication is critical in light of sample demographic characteristics.