Crosscultural Measurement Invariance of the Internet Addiction Test-Revised (IAT-R) Among Japanese and Filipino University Students

A Cross-Continent Analysis of the Invariance of Product Information in Cross-Border Electronic Commerce

Cross-border electronic-commerce (CBEC) is growing. However, due to differences in culture, habits, history and language among other factors, consumers in different regions may have different perception towards the same product information on CBEC platforms, which may lead to differences in their cognition of the product with implications for purchase intentions. Presently, little research has attempted to understand whether there are such differences between global consumers through the examination of measurement invariance (MI) in CBEC environments. By using multiple-group confirmatory factor analysis (MG-CFA), this study explored the invariance of two product information cognitions on CBEC platforms, namely product description and product awareness, among consumers in North America, Europe, Latin America and Oceania. Data was collected from users of a popular CBEC platform in China. We find no significant differences in understandings and levels of awareness of product information across the four groups of consumers.

Development and Differential Item Functioning of the Internet Addiction Test-Revised (IAT-R): An Item Response Theory Approach

This study developed and investigated the differential item functioning (DIF) of the Internet Addiction Test-Revised (IAT-R) with an item response theory approach. In the Asian College Health Assessment (ACHA), 1,072 university students completed a survey in 2016-2017 school year. Confirmatory factor analysis models with robust maximum likelihood and diagonal weighted least square estimation methods were used to evaluate the construct validity of the 20-item IAT-R. Graded response model was used to produce categorical characteristic curves (CCCs), test characteristic curves (TCCs), item information function (IIF) curves, and test information function (TIF) curves for detecting DIF of the polytomous responses. Furthermore, DIF between genders was examined by ordinal logistic regression and Monte Carlo simulations. A first-order three-factor model was the most parsimonious model with normed fit index (NFI) of 0.915, non-normed fit index (NNFI) of 0.927, comparative fit index (CFI) of 0.937, and root mean square error of approximation (RMSEA) of 0.050. The emerged factors included Excessive Use and Neglect Work, Anticipation and Lack of Control, as well as Neglect Social Life and Salience. CCCs, TCCs, IIFs, and TIFs showed that all items were sensitive at moderate-to-high trait values. No nonuniform scale-level DIF relating to gender was determined. Under no DIF, the thresholds for proportional beta change exhibited a fairly steady trend (below 0.10) across items. In conclusion, IAT-R is a valid measurement scale of Internet addiction with measurement equivalence between genders being established.