Cross-cultural measurement of social overload and social networking sites exhaustion: Rasch and DIF evidence from Ghana, Hong Kong, Türkiye

Abstract

Social media research increasingly examines the psychological costs of constant connectivity, yet the cross-cultural comparability of widely used social overload and Social Networking Sites (SNS) exhaustion measures remains insufficiently tested. This study examined the dimensionality, factorial validity, item functioning, measurement invariance, and measurement-informed structural associations of the Social Overload Scale (SOS) and SNS Exhaustion Scale across Ghana, Hong Kong, and Türkiye. Data was collected from 1,323 university students. A sequential strategy combined Horn’s Parallel Analysis, Rasch Partial Credit Modeling, Expected A Posteriori (EAP) reliability, Confirmatory Factor Analysis (CFA), primary cumulative-logit ordinal Differential Item Functioning (DIF) analysis, an independent polytomous Mantel DIF procedure, a category-collapsing DIF sensitivity analysis, ordinal measurement invariance, and multi-group SEM. Additional structural sensitivity analyses adjusted for age, gender, daily social media use, and daily smartphone use. Both scales showed a dominant unidimensional structure and adequate measurement precision (EAP = 0.88 for SNS Exhaustion; EAP = 0.82 for SOS). Rasch item-fit statistics were acceptable for all SNS Exhaustion items and for SOS1–SOS5, whereas SOS6 showed localized misfit. However, several items, particularly SOS items, showed DIF across country pairs, indicating partial item-level comparability. Cross-method DIF evidence was strongest for SNS1 and SOS3–SOS6 in at least one country comparison, while category collapsing reduced, but did not eliminate, item-level DIF signals. Approximate threshold and loading invariance supported cautious multi-group structural modeling. The mediation pathway was country-conditioned: Social Overload mediated the association between Social Digital Pressure and SNS Exhaustion in Hong Kong and Türkiye, but not in Ghana. This pattern remained stable after removing the culturally sensitive SOS item and after jointly adjusting for age, gender, daily social media use, and daily smartphone use. The study shows that scale validity and structural theory cannot be separated in cross-cultural social media research. Reliable and well-fitting scales may still contain culturally sensitive items, and such differences can affect how structural relationships should be interpreted. Therefore, item-level diagnostics, ordinal measurement invariance, and measurement-informed multi-group modeling should precede claims about universal psychological mechanisms.