Data_Sheet_1_Integrating multi-informant reports of youth mental health: A construct validation test of Kraemer and colleagues’ (2003) Satellite Model.PDF

Accurately assessing youth mental health involves obtaining reports from multiple informants who typically display low levels of correspondence. This low correspondence may reflect situational specificity. That is, youth vary as to where they display mental health concerns and informants vary as to where and from what perspective they observe youth. Despite the frequent need to understand and interpret these informant discrepancies, no consensus guidelines exist for integrating informants’ reports. The path to building these guidelines starts with identifying factors that reliably predict the level and form of these informant discrepancies, and do so for theoretically and empirically relevant reasons. Yet, despite the knowledge of situational specificity, few approaches to integrating multi-informant data are well-equipped to account for these factors in measurement, and those that claim to be well-positioned to do so have undergone little empirical scrutiny. One promising approach was developed roughly 20 years ago by Kraemer and colleagues (2003). Their Satellite Model leverages principal components analysis (PCA) and strategic selection of informants to instantiate situational specificity in measurement, namely components reflecting variance attributable to the context in which informants observe behavior (e.g., home/non-home), the perspective from which they observe behavior (e.g., self/other), and behavior that manifests across contexts and perspectives (i.e., trait). The current study represents the first construct validation test of the Satellite Model. A mixed-clinical/community sample of 134 adolescents and their parents completed six parallel surveys of adolescent mental health. Adolescents also participated in a series of simulated social interactions with research personnel trained to act as same-age, unfamiliar peers. A third informant (unfamiliar untrained observer) viewed these interactions and completed the same surveys as parents and adolescents. We applied the Satellite Model to each set of surveys and observed high internal consistency estimates for each of the six-item trait (α = 0.90), context (α = 0.84), and perspective (α = 0.83) components. Scores reflecting the trait, context, and perspective components displayed distinct patterns of relations to a battery of criterion variables that varied in the context, perspective, and source of measurement. The Satellite Model instantiates situational specificity in measurement and facilitates unifying conceptual and measurement models of youth mental health.

精准评估青少年心理健康，需收集多名知情人的报告，而这类报告间通常一致性较低。这种低一致性可能源于情境特异性。换言之，青少年的心理健康问题表现场景存在个体差异，而知情人观察青少年的场景与视角也各不相同。尽管学界常需理解并解释这类知情人报告间的差异，但目前尚无整合多知情人报告的统一指南。构建此类指南的第一步，是识别能够可靠预测这类知情人报告差异的程度与形式的因素，且这种预测需具备理论与实证层面的合理性。然而，尽管学界已认识到情境特异性，但目前鲜有整合多知情人数据的方法能够在测量层面妥善考量上述因素；即便有方法声称可做到这一点，相关实证检验也极为匮乏。约20年前，Kraemer及其同事（2003）提出了一种颇具前景的方法。该卫星模型（Satellite Model）借助主成分分析（Principal Components Analysis, PCA）与知情人的战略性筛选，在测量层面实现情境特异性的量化，具体可拆解为三类成分：反映知情人观察行为时所处情境（如家庭/非家庭场景）的方差成分、反映其观察行为时所持视角（如自评/他评）的方差成分，以及跨场景与视角均存在的行为方差成分（即特质成分）。本研究是首次对卫星模型开展结构效度检验的研究。本研究招募了134名青少年及其家长，采用临床-社区混合样本，让其完成6套平行的青少年心理健康测评问卷。此外，青少年还参与了一系列模拟社交互动任务，互动对象为经过培训、可扮演同龄陌生同伴的研究人员。第三名知情人——未经培训的陌生观察者——观看了全部互动过程，并完成了与青少年和家长相同的测评问卷。我们将卫星模型应用于每套测评问卷，结果显示：由6个条目构成的特质成分（克朗巴赫α=0.90）、情境成分（α=0.84）与视角成分（α=0.83）均具有极高的内部一致性。代表特质、情境与视角成分的得分，与一系列在测量场景、视角与来源上各不相同的效标变量呈现出各异的关联模式。卫星模型可在测量层面实现情境特异性的量化，同时为青少年心理健康的统一概念模型与测量模型提供支撑。