Abstract
Idiographic measures, such as the Top Problems Assessment (TPA), offer great utility over nomothetic measures as they can be used to characterize participants in a person-centered manner. Despite such utility, idiographic measures rely on heuristic coding systems, which limit their applicability as a research tool. To address such concerns, the present study utilized Latent Dirichlet Allocation (LDA) to model the latent topics in youth- and parent-reported concerns identified by the TPA in a sample of 172 treatment-seeking youth. Clinical and demographic correlates associated with the identified topics were assessed using general linear models. Topic modeling identified four latent topics in youth-reported concerns that were characterized by Emotional Reactivity, Worry, Avoidance, and Contexts and six latent topics in the parent-reported concerns, Worry, Avoidance, Emotional Reactivity, Negative-Valenced Emotions, Parent-Child Interactions, and Other Maladaptive Behavioral Response. Certain latent topics were significantly associated with standardized measures of anxiety, avoidance, age and reported sex. Results serve as a proof of concept for applying topic modeling to the TPA and further contribute to efforts that characterize domains of impairment in youth with psychopathology. Results of the present study and subsequent discussion of important methodological and conceptual considerations may guide the use of similar research in leveraging data-driven modeling to elucidate latent topics underlying qualitative clinical data.
