| Summary: | This dissertation uses cognitive fit to present a novel approach to choosing whether to analyze clinical data assuming a categorical or dimensional latent variable structure. Cross-sectional and longitudinal models assuming a categorical (i.e., latent class analysis model and latent transition analysis model) and dimensional (i.e., graded response model and longitudinal graded response model) latent variable structure were fit to clinical data from clinically referred youth (N = 204). Results found a large overlap in resulting youth predictions across models. Information regarding whether youth improved, deteriorated, or remained the same in terms of symptom severity was similar between latent class and graded response models. However, the presentation and interpretation of information for use in feedback differed: categorical models yielded information about group (qualitative) differences whereas dimensional models yielded information about individual differences in degree (quantitative). Based on cognitive fit, the most effective and efficient clinical-decisions occur when the specific clinical task matches with the information presentation. This match allows for a deliberate and clinically useful choice in model selection for the creation of clinical feedback. Additionally, results from an informal clinical survey investigating information presentation preferences for specific clinical tasks are presented. Clinician preference may indicate the natural fit between task and information presentation. Implications for clinical feedback are discussed and future studies are proposed to explore further the effect of information presentation on clinical decision-making.
|
|---|
