Assessing the dimensionality of the sense of coherence scale (SOC-L9) using Ghanaian university students: Guarding against the method effect

Abstract

Background: The dimensionality of the sense of coherence (SOC-L9) scale has been in contention due to the varied factor structure revealed in the literature. In this study, we assessed the dimensionality of the SOC-L9 scale using Ghanaian university students while guarding against the method effect. The study also examined the gender measurement invariance of the scale. Methods: This research conveniently sampled 1062 students who responded to the SOC-L9 scale with negative items reversed to positive items. A larger proportion of the participants were male students (n = 769, 72.4 %), with 293(27.6 %) being female students. The youngest participant was 18 years old, whereas the oldest was 42 years old. Following all validation studies protocols, four distinct confirmatory factor analysis models were fitted and compared (i.e., unidimensional, three-factor first-order, three-factor second-order and bifactor models). Results: The initial model comparison revealed that the bifactor CFA model [CFI = 0.958, SRMR = 0.036, AIC = 21231.35, BIC = 21370.45] was superior to the unidimensional [CFI = 0.914, SRMR = 0.046, AIC = 26280.67, BIC = 26414.8] and 3-factor models [CFI = 0.932, SRMR = 0.040, AIC = 26221.67, BIC = 26370.71]. Upon further probing, it was discovered that SOC-L9 functions best as a unidimensional scale for the university student population. Gender measurement invariance was established for configural invariance [CFI = 0.986, SRMR = 0.044], metric invariance [CFI = 0.894, SRMR = 0.051] and scalar invariance [CFI = 0.983, SRMR = 0.047]. Conclusion: The SOC-L9 scale has a nested structure with the various sub-scales interacting to produce a summary total observed score. The structure of the SOC-L9 requires scholars to treat the scale as a unidimensional scale rather than a multidimensional one. This latent structure was found to be consistent with male and female university students.

Keywords

• Sense of coherence
• Method effect
• Bifactor model
• Confirmatory factor analysis
• Psychometrics