Examining the structure of the Internet Addiction Test in adolescents: A bifactor approach

•Previously proposed structural models of the IAT are not optimal for adolescents.•A bifactor model with a general factor and two unique sub-dimensions fits best.•Presence of the dominant global IA factor supports the use of the total-scale score.•Scores on the two subscales add significant incremental variance to the assessment.•The two subscales are associated with unique gender and problem gambling profiles.

Young’s (1998a) Internet Addiction Test (IAT) is one of the most commonly used measures of problematic Internet use. Although the instrument is being increasingly applied in research with adolescents, its measurement structure has not been systematically evaluated with pre-adult respondents. Because the results of previous factor-analytic studies with adults have been highly inconsistent, the present study applied both the traditional (simple-structure) and novel (bifactor) modeling approaches to derive the most optimal measurement structure of the IAT for adolescents. The results of exploratory and confirmatory analyses in a large Canadian sample of high-school students (N = 1948) converged on a bifactor model with a dominant global IA factor and two distinct sub-dimensions, each associated with a unique gender and problem behaviour profile. The discussion focuses on the implications of this bifactor structure for scale scoring and substantive theory on the nature and sources of individual differences in Internet addiction.