Popular Internet Addiction Questionnaire Shows Cracks in Chinese Young Adults

A widely used questionnaire for measuring problematic internet use may not work as cleanly in Chinese university students as its original design assumed, according to a new psychometric study published in Frontiers in Public Health. The research, led by Sai-fu Fung and Cyrus Ngo Tin Lai of City University of Hong Kong with colleagues at The Chinese University of Hong Kong and Université du Québec à Trois-Rivières, put the 18-item Problematic Internet Use Questionnaire, or PIUQ-18, through a rigorous statistical evaluation in a sample of 511 Chinese emerging adults, 85.5 percent of whom were female. Problematic internet use is a growing concern among university students, and the PIUQ-18 is one of the instruments most often used to measure it.

Yet the questionnaire's underlying factor structure, meaning the way its items group together into distinct dimensions, has remained debated across cultures. The original version proposes three subscales: Obsession, Neglect and Control Disorder. Whether those three dimensions hold up as separate constructs outside the populations where the scale was first developed has been an open question.

To address it, the team used a split-sample design, dividing their participants into two groups: 260 for exploratory factor analysis and 251 for confirmatory factor analysis. The full 18-item questionnaire showed good internal consistency, with a Cronbach's alpha of .87 and McDonald's omega of .87, indicating that the items hang together reliably as a whole. But when the researchers tested the original three-factor model, it ran into a serious statistical problem.

The latent correlations between subscales exceeded 1.00, with the Obsession-Control correlation at 1.05 and the Neglect-Control correlation at 1.09. Correlations above 1.00 are mathematically inadmissible in standard factor models and suggest that the supposed dimensions are not empirically distinguishable in this sample. Searching for a better-fitting alternative, the team derived a new three-factor structure through exploratory factor analysis.

This alternative model was the best-fitting admissible version of the PIUQ-18, with a comparative fit index of .91, a Tucker-Lewis index of .90, a root mean square error of approximation of .07 and a standardized root mean square residual of .07. Even so, its incremental fit remained below conventional benchmarks, meaning the improvement over simpler models was not strong enough to fully satisfy accepted standards for structural models. The researchers also examined how distinct the dimensions really were using the heterotrait-monotrait ratio of correlations, or HTMT, a modern diagnostic for discriminant validity.

The HTMT values fell below conventional thresholds, which on its face suggests acceptable separation between factors. However, the confidence intervals for the latent correlations indicated that the dimensions remained substantially related, tempering any conclusion that the three subscales measure clearly different things in this population. Here the six-item short form emerged as the stronger performer.

It showed acceptable fit for a unidimensional structure, with a CFI of .97, a TLI of .95, an RMSEA of .08 and an SRMR of .05, along with theoretically consistent associations with well-being indicators. The nine-item version showed comparable but weaker fit, with a CFI of .91, a TLI of .88, an RMSEA of .11 and an SRMR of .07. Not every part of the scale fared equally well.

The Control Disorder subscale exhibited weaker reliability and validity than the other dimensions, and the authors note that it requires refinement before it can be considered a dependable component of the questionnaire. The authors are careful about the limits of what their results support. They describe the findings for the PIUQ-SF-6 as preliminary evidence for a unidimensional factor structure, not as proof that the short form is ready for use as a screening tool.

The sample, drawn from Chinese emerging adults and heavily weighted toward female participants, may not generalize to other populations, and the split-sample design, while methodologically sound, reflects a single dataset. For researchers and clinicians who rely on the PIUQ family of instruments, the practical message is one of caution rather than abandonment. The full questionnaire remains internally consistent as a global measure of problematic internet use, and the six-item short form shows early promise as a brief unidimensional index in this population.

But the original three-dimensional scoring of the PIUQ-18 appears statistically untenable among Chinese emerging adults, and anyone using the scale in similar groups should treat its subscale structure, particularly the Control Disorder dimension, as a work in progress rather than a settled measurement standard.

Subject of Research: Psychometric evaluation of the Problematic Internet Use Questionnaire and its short forms in Chinese emerging adults

Article Title: Psychometric evaluation of the problematic internet use questionnaire in Chinese emerging adults: an exploratory alternative three-factor structure and unidimensional short forms

Article References: Fung, S.-F., Lai, C. N. T., Garneau, J., & Parent-Lamarche, A. (2026). Psychometric evaluation of the problematic internet use questionnaire in Chinese emerging adults: an exploratory alternative three-factor structure and unidimensional short forms. Frontiers in Public Health, 14, Article 1980618. https://doi.org/10.3389/fpubh.2026.1980618

Image Credits: AI Generated

DOI: 10.3389/fpubh.2026.1980618

Keywords: problematic internet use, psychometrics, PIUQ-18, factor analysis, university students, Chinese emerging adults, internet addiction, short forms, construct validity, behavioral addiction, questionnaire validation, Frontiers in Public Health