Previous work investigating the dimensionality of psychological constructs has assumed a fixed-effects model, in which one true correlation describes the relationship between two given dimensions. We challenge this assumption by showing how a random-effects model may aid in representing individual perceptions of multidimensional constructs. Using the Pay Satisfaction Questionnaire (PSQ), we demonstrate that the relationships that individuals hold between the (purported) dimensions of the PSQ are predictable by cognitive complexity, pay level satisfaction, pay level, and interactions; we also show how between-dimension variation assists in establishing criterion-related validity.
Sturman, M. C., & Carraher, S. M. (2007). Using a random-effects model to test differing conceptualizations of multidimensional constructs [Electronic version]. Retrieved [insert date], from Cornell University, School of Hospitality Administration site: http://scholarship.sha.cornell.edu/articles/100