Relational Uncertainty Within Relational Turbulence Theory: The Bifactor Exploratory Structural Equation Model

Abstract

Abstract Relational uncertainty consists of self, partner, and relationship uncertainty, which are core parameters in relational turbulence theory (RTT). Advances in latent variable modeling allow researchers to examine the multidimensional construct as a bifactor model, including a general factor of relational uncertainty and residualized factors of self, partner, and relationship uncertainty. This advance is theoretically consequential because RTT maintains the importance of distinctions among the facets of relational uncertainty, even while empirical evidence demonstrates considerable overlap among them. In two data sets (college sample, N = 513; married sample, N = 354), competing measurement models specified Relational Uncertainty Scale items as a unidimensional confirmatory factor model, 3-factor independent clusters confirmatory factor model, 3-factor exploratory structural equation model (ESEM), bifactor confirmatory factor model, and bifactor-ESEM. The bifactor-ESEM provided the best fit in both samples, and bifactor statistics clarified how variance from the scale is partitioned to yield essential unidimensionality for the general factor of relational uncertainty, controlling for its residualized factors. Tests of a latent predictive model using a bifactor specification of relational uncertainty were consistent with RTT. Implications for testing communication theory using bifactor-ESEM are discussed in light of these findings.