(PS4-16) Read the Room: An Innovative Application of Kraemer and Colleagues’ (2003) Satellite Model to Integrate Anxiety Data Within a Social Context

Abstract Text: Anxiety researchers often use data sources that yield discrepant results about a person’s anxiety (De Los Reyes, Wang et al., 2023). Using principal components analysis (PCA), Kraemer et al.’s (2003) Satellite Model posits that discrepancies among observer- (e.g., parent, teacher) and self-reports about a single person’s behavior can be meaningfully integrated to produce indices of three sources of variability: Trait (behavior that manifests across the contexts and perspectives reflected by three informants’ reports), Perspective (whether the informant provides reports as an observer or self-rater), and Context (environment in which the informant observes the person being rated). Recent work demonstrates the Satellite Model’s validity when integrating reports about an adolescent’s social anxiety, using reports from parents and adolescents completed as part of a standard survey battery as well as reports from unfamiliar untrained observers (UUOs) who rated adolescents based on their interactions with same-age unfamiliar peers (Charamut et al., 2022). Here, the adolescent’s social anxiety is a response to the social context. Their anxiety is influenced by not only their behavior but also the interaction partner (i.e., unfamiliar peer). We tested the value of integrating social anxiety data about the social context, namely the anxiety levels of an adolescent and an unfamiliar peer when interacting with each other.
A mixed clinical/community sample of 134 adolescents (M=14.5 years, 66.4% female) participated in the Unfamiliar Peer Paradigm (Cannon et al., 2020), a set of social interactions with unfamiliar peers. Two trained raters coded adolescents’ anxiety within these tasks (_MICC=.75) using a well-established coding scheme (Glenn et al., 2019). Adolescents self-rated their arousal within these tasks via the Self-Assessment Manikin (SAM; Lang, 1980). We constructed two PCA models. The PCA model of adolescent social anxiety included parents’, adolescents’, and UUOs’ reports on the Social Phobia Scale (SPS; Mattick & Clarke, 1998). The PCA model of the social context included UUOs’ reports about the unfamiliar peer using the Social Interaction Anxiety Scale (SIAS; Mattick & Clarke, 1998), unfamiliar peers’ reports about the adolescent using the SIAS, and adolescents’ self-reported arousal using the SAM.
Both PCAs yielded interpretable Trait, Perspective, and Context components. A hierarchical regression revealed that SPS Trait scores of adolescent social anxiety explained variance in observed adolescent anxiety (ΔR₂=.23, p < .001), and SIAS Trait scores of the social context incrementally explained variance in observed adolescent anxiety (ΔR₂=.11, p < .001).
Kraemer et al.’s (2003) PCA method facilitates integrating data that disagree for valid reasons, and has historically been applied to integrate data about an individual assessment target. Yet, to validly assess the social anxiety of an individual, one requires data about not only that individual, but also those with whom they interact. Our application of the Satellite Model demonstrates the value of capturing the social anxiety “released” by all involved in a social context, even if the aim of these data is to explain the anxiety of a single individual in that context.