This study analyzed polytomous testing data of an internet game addiction scale for adolescents
developed by Korea Agency for Digital Opportunity and Promotion using generalized partial credit model
that hypothesizes cumulative mechanism and generalized graded unfolding model for unfolding mechanism
to explore how these two models interpret items differently and what test experts or practitioners should
consider when they choose models for item analysis. Item response theory uses dominance process
model(DPM) or cumulative mechanism traditionally that examinee high on the trait assessed are assumed
to answer positively with high probability. But recently, ideal point model that hypothesized that the
probability of endorsement is assumed to be directly related to the proximity of the statement to the
examinee’s location on the assessed trait could be applied to affective testing data. By analyzing testing
data with IPM, we could consider not only cumulative item characteristic curve but also unfolding item
characteristic curve, then deeper understanding could be obtained about adolescents’ response patterns to
each item.