The impact of measurement noninvariance across time and group in longitudinal item response modeling

• Published in: Asia Pacific education review Vol. 25; no. 4; pp. 911 - 924
• Main Author: Choi, In-Hee
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01.10.2024; Springer Nature B.V; 교육연구소
• Subjects: Data Analysis; Education; Educational Research; Educational tests & measurements; Factor Analysis; Feedback (Response); Growth Models; Item response theory; Longitudinal Studies; Measurement Techniques; Time; 교육학; Explanatory item response modeling; Growth model; Measurement invariance; Longitudinal item response model
• ISSN: 1598-1037; 1876-407X
• DOI: 10.1007/s12564-023-09907-4

Longitudinal item response data often exhibit two types of measurement noninvariance: the noninvariance of item parameters between subject groups and that of item parameters across multiple time points. This study proposes a comprehensive approach to the simultaneous modeling of both types of measurement noninvariance in terms of longitudinal item response modeling while incorporating a latent growth curve model. The growth curve model represents the growth trajectory using initial status and (non)linear change over multiple time points. In addition, this study investigates the impact of measurement noninvariance across time points and subject groups in the examination of group differences in growth over multiple time points. To demonstrate the efficacy of the proposed model, an empirical application to large-scale longitudinal survey data is provided, along with the results of a simulation study.