The plausibility of cross-national comparisons of internet use types

In studies that compare Internet use between groups, factor analysis is often used to create broader categories of use. Composite variables are constructed using a factor structure that fits the overall data. This approach overlooks tests that show whether the constructs are valid and whether items relate similarly to the general constructs in different countries. In this study, World Internet Project survey data were submitted to multi-group latent variable analysis and tested for configural and measurement invariance to illustrate why these tests might be important in cross-national comparisons. It was found that while the general Internet use constructs were applicable across national contexts, the way in which the items contributed was different.