A network perspective to the measurement of sense of coherence (SOC): an exploratory graph analysis approach

Abstract

The measurement of sense of coherence (SOC) has received attention for more than three decades. Despite the extensive use of SOC-13, there is still a long debate regarding its dimensionality structure. Recently, there has been an increasing use of network modeling as a valid alternative to latent-variable modeling. This study proposes an exploratory approach to the structure of SOC-13 by adopting a network perspective. The network structure was estimated with a Gaussian Graphical Model, and Exploratory Graph Analysis (EGA) was used to inspect network dimensionality. We fit and compared the unidimensional, first- and second-order confirmatory factor analysis (CFA), bifactor-CFA, and structure derived from EGA. Our results showed unacceptable fit values for the CFA models, suggesting that SOC-13 is not unidimensional. Inspection of the estimated network suggested that the SOC-13 items emerged as a dynamic system of mutually interacting nodes that formed three distinct clusters of items (communities) that are not those defined in the literature. EGA identified three communities of items: the first community was characterized by comprehensibility and manageability items, the second community was characterized by comprehensibility and manageability items, and the third dimension was characterized by all meaningfulness items and one comprehensibility item. Our study presented a novel perspective in investigating the structure of SOC-13 that strengthens the assumption that SOC should be conceptualized as a complex system of cognitive (comprehensibility), behavioral (manageability), and motivational dimensions (meaningfulness) that are deeply linked and not necessarily distinct.