Paul Ginns, Andrew J. Martin, Gregory A. D. Liem, Brad Papworth
Structural validity studies of Big Five measures using confirmatory factor analysis often yield mixed levels of fit. However, substantial evidence of criterion-related validity also exists for such measures. We investigated the structural and concurrent validity of a recently developed, short, trait-descriptor-based Big Five measure with a large sample of Australian adolescents. Using confirmatory factor analysis (CFA) and the more recently developed exploratory structural equation modeling (ESEM), we investigated impacts of a range of modeling decisions, including item parceling and item valence models. The most substantial improvements in fit over CFA-based models resulted from use of ESEM and fitting of correlated uniquenesses based on item valence; effects of modeling decisions on criterion validity were modest. Implications for substantive research are discussed.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados