The study drew on responses from 1,323 university students, a population for whom constant connectivity is not a lifestyle choice but a baseline condition of social and academic life. The researchers were interested in more than whether the scales produced tidy numbers. They wanted to know whether the psychological story built on top of those numbers — that pressure from one’s social circle leads to a sense of overload, which in turn produces exhaustion with social media — holds up in the same way everywhere. To answer that, they combined classical psychometrics with modern item-level diagnostics in a sequence that few cross-cultural studies attempt.

The technical core of the paper is a chain of analyses that each interrogate the data from a different angle. First, Horn’s Parallel Analysis was used to establish dimensionality — essentially, whether each scale really measures a single underlying construct or a bundle of loosely related ones. Both instruments emerged as dominantly unidimensional, which is the ideal outcome for a scale intended to yield one summary score. Next came Rasch Partial Credit Modeling, an item-response-theory approach that models the probability of each response category as a function of the person’s level of the trait and the item’s difficulty. Rasch analysis produces fit statistics that flag items behaving erratically relative to the rest of the scale, and it yields an Expected A Posteriori reliability estimate, a precision metric suited to ordinal data.

On those fronts the news was mostly good. The SNS Exhaustion Scale achieved an EAP reliability of 0.88, and the Social Overload Scale 0.82 — both comfortably within the range considered adequate for research use. Rasch item-fit statistics were acceptable for every SNS Exhaustion item and for five of the six Social Overload items. The exception was SOS6, which showed localized misfit, meaning it did not cohere with the rest of its scale in the way a well-behaved item should. At this point, a conventional study might have declared victory: two reliable, essentially unidimensional scales, ready for cross-cultural comparison. The team instead pressed on to the question that most studies skip.

That question is differential item functioning, or DIF. An item shows DIF when people from different groups — here, different countries — with exactly the same underlying level of the construct give systematically different answers to it. DIF is the statistical fingerprint of cultural sensitivity: a question about feeling overwhelmed by friends’ requests may simply not carry the same social meaning in Accra as it does in Hong Kong or Gaziantep, even when translated carefully. The researchers ran a primary cumulative-logit ordinal DIF analysis, then verified their findings with an independent polytomous Mantel DIF procedure, and finally performed a category-collapsing sensitivity analysis to check whether the signals were artifacts of sparse response categories rather than genuine cultural differences.

The results were sobering. Cross-method DIF evidence was strongest for item SNS1 and for Social Overload items SOS3 through SOS6, each flagged in at least one country comparison. Collapsing response categories reduced but did not eliminate these item-level signals, suggesting they reflect substantive cultural differences in how the questions are understood rather than mere statistical noise. In other words, scales that looked reliable and well-fitting still contained items that functioned differently across cultures. Reliability, the authors emphasize, is a statement about internal consistency within a sample — it says nothing about whether the same score means the same thing in Lagos, Hong Kong or Istanbul.

Why does this matter beyond the technical literature? Because cross-cultural comparisons of social media harm are increasingly used to shape policy, design interventions and support sweeping claims about a global epidemic of digital fatigue. If an item functions differently across countries, mean comparisons built on it can be distorted in ways that no amount of sample size will fix. A finding that Ghanaian students report less social overload than Hong Kong students, for example, might partly reflect differences in how the items are interpreted rather than genuine differences in experienced pressure. The study does not claim that cross-cultural research is impossible — it claims that the diagnostic work must come first.

The team then tested whether the structural theory itself survived measurement scrutiny. Using ordinal measurement invariance analysis, they found approximate threshold and loading invariance across the three countries — enough, they argue, to justify cautious multi-group structural equation modeling. The model they tested followed a stressor-strain-outcome logic: social digital pressure from one’s network acts as a stressor, social overload is the strain, and SNS exhaustion is the outcome, with overload mediating the pathway from pressure to exhaustion. The striking result was that this mediation held in Hong Kong and Türkiye but not in Ghana. The pathway was country-conditioned, and the pattern remained stable even after the researchers removed the culturally sensitive SOS item and after jointly adjusting for age, gender, daily social media use and daily smartphone use.

That stability is important, because it rules out the most convenient alternative explanations. The Ghana-Hong Kong-Türkiye difference in the mediation pathway is not an artifact of one problematic item, nor of demographic composition, nor of how much time students spend on their phones each day. Something about the way social digital pressure translates into overload and then into exhaustion genuinely differs across these contexts — perhaps reflecting differences in collectivist norms, in the density of offline and online social obligations, or in how university students in each society experience the demands of their networks. The study cannot say which mechanism is responsible, but it demonstrates convincingly that the difference is real enough to survive a battery of sensitivity checks.

The broader lesson the authors draw is that scale validity and structural theory cannot be separated in cross-cultural social media research. A scale can be reliable, unidimensional and well-fitting by every conventional criterion and still contain items that are culturally sensitive, and those sensitivities can change how structural relationships should be interpreted. Their prescription is procedural: item-level diagnostics such as Rasch analysis and DIF testing, ordinal measurement invariance checks, and measurement-informed multi-group modeling should precede any claim about universal psychological mechanisms. For a field racing to document the psychological costs of constant connectivity around the world, the study is a reminder that the instruments of measurement are themselves cultural artifacts — and that treating them as neutral can quietly bend the science built upon them.

Subject of Research: Cross-cultural psychometric validation of social overload and social networking sites exhaustion measures

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

Article References: Oppong, D. K., Üztemur, S., Gökalp, A., Ahorsu, D. K., & Lin, C.-Y. (2026). Cross-cultural measurement of social overload and social networking sites exhaustion: Rasch and DIF evidence from Ghana, Hong Kong, Türkiye. BMC Psychology. https://doi.org/10.1186/s40359-026-05751-4

Image Credits: AI Generated

DOI: 10.1186/s40359-026-05751-4

Keywords: social overload, SNS exhaustion, Rasch model, differential item functioning, measurement invariance, cross-cultural psychology, psychometrics, social media burnout, multi-group SEM, social digital pressure, Ghana, Hong Kong