Dimensionality and measurement invariance of the smartphone addiction scale among pharmacy students using general, correlated, second-order, and bifactor models; association with the bergen social media addiction scale: a multi-university study

The Smartphone Addiction Scale Short Version (SAS-SV) is widely used to assess problematic smartphone use, yet evidence on its underlying construct structure remains inconsistent across sociocultural contexts. This study evaluated the dimensionality, measurement invariance, and convergent validity of the SAS-SV by competing factor models—general, correlated, second-order, and bifactor in a multicenter Nigerian sample. A cross-sectional online survey was conducted among 688 undergraduate pharmacy students (354 males, 334 females; mean age = 22.33, SD = 3.31 years) at three public universities in Nigeria (University of Ibadan, Obafemi Awolowo University, and University of Benin). The participants completed the SAS-SV and the Bergen Social Media Addiction Scale (BSMAS). The dataset was split into two subsamples for exploratory factor analysis (EFA; n1 = 344) and confirmatory factor analysis (CFA; n2 = 344). We evaluated the structural architecture using competing CFA models, ancillary bifactor indices, multigroup measurement invariance, and bivariate correlations with BSMAS scores. Among the competing confirmatory models, the bifactor model showed superior fit (χ2/df = 3.67, CFI = 0.949, TLI = 0.911, SRMR = 0.043), although RMSEA was elevated (0.088, 90% C.I. [0.070, 0.108]). Ancillary bifactor indices, including Explained Common Variance (ECV) = 0.697, Hierarchical Omega (ωH) = 0.752, Construct Replicability (H) = 0.862, and Factor Determinacy (FD) = 0.891, strongly supported the predominance of a general smartphone addiction factor. Furthermore, the SAS-SV demonstrated configural, metric, and scalar invariance across genders and institutions, as well as strong convergent validity with the BSMAS ( r = 0.781, p < 0.001, 95% C.I. [0.743, 0.813]). Although multidimensional subthemes exist, the SAS-SV functions primarily as a robust, unidimensional instrument in this population. Researchers and clinicians can confidently use the total composite score to assess problematic smartphone use and should interpret it with awareness of underlying sub-scale variance.