Introduction and Purpose:
This study examined how the choice of item response model—dominance versus ideal point—affects the identification of latent profiles of character strengths and their associations with well-being. While dominance models are commonly applied, recent evidence suggests that ideal point models may be more suitable for personality-type constructs such as character strengths. Using an inappropriate model may lead to biased trait estimates and distort subsequent latent profile classifications.
Method:
Participants were 872 university students who completed the IPIP-VIA Character Strengths Inventory and the Mental Health Continuum-Short Form. Trait estimates were derived using a dominance model (Generalised Partial Credit Model) and an ideal point model (Generalised Graded Unfolding Model). Separate latent profile analyses (LPA) were conducted for the two sets of trait measures.
Results:
Three distinct character strength profiles (low, medium, and high strengths) with acceptable fit emerged under both approaches. However, profile membership differed across models. The ideal point model yielded clearer differentiation between the high and medium profiles. ANOVAs revealed significant differences in well-being across profiles, with stronger effects under the ideal point model.
Conclusions:
Results demonstrate that the choice of measurement model affects latent profile identification and the relationships between the profiles and well-being. Ideal point models appear more sensitive in distinguishing individuals at higher trait levels, offering greater precision for identifying profiles, evaluating positive psychology interventions, and examining the role of character strengths in well-being.