Structural Models.

Purpose
This study aimed to understand whether incorporating wellbeing as another dimension within general factor models of mental health is (a) feasible and (b) useful.
Methods
Data from two time points (Year 7 and Year 9) for 15258 adolescents who participated in the HeadStart programme in England was used. In Stage 1, we used structural equation modelling on time point 1 data to test different latent variable models incorporating psychopathology and wellbeing dimensions. In Stage 2, we tested whether the latent factors identified in Stage 1 significantly predicted impairment at time point 2.
Results
A general factor model incorporating a shared underlying dimension between (lack of) wellbeing and psychopathology as well as unique specific factors had good fit to the data at Stage 1. Further, although both general factor models with and without wellbeing fit the data well at Stage 1, only the general factor model with wellbeing met all required fit thresholds when regressions to predict impairment were added in. The model without any general factor (correlated factors model) met pre-defined fit thresholds but had lower fit indices.
Conclusion
The incorporation of wellbeing into general factor models may help represent more nuanced mental health states and may be useful in predicting future functional states, however such a model needs further replication with comprehensive measures and comparison with alternative models to verify its validity and utility.

研究目的
本研究旨在探讨将幸福感（Wellbeing）作为另一维度纳入心理健康一般因子模型是否（a）可行，以及（b）具有应用价值。

研究方法
本研究使用了英格兰地区参与HeadStart项目的15258名青少年在两个时间点（7年级与9年级）采集的数据。在第一阶段，我们基于第一时间点的数据采用结构方程模型（Structural Equation Modelling）检验了纳入心理病理（Psychopathology）与幸福感维度的多种潜变量模型（Latent Variable Models）。在第二阶段，我们检验了第一阶段识别出的潜因子是否能显著预测第二时间点的功能损害。

研究结果
第一阶段中，纳入（缺失状态的）幸福感与心理病理间共享潜在维度，同时包含独特特异性因子的一般因子模型对数据具有良好的拟合度。此外，尽管纳入与未纳入幸福感的一般因子模型在第一阶段均能较好拟合数据，但当加入预测功能损害的回归项后，仅纳入幸福感的一般因子模型满足所有预设的拟合阈值要求。未纳入任何一般因子的模型（即相关因子模型）虽满足预设拟合阈值，但其拟合指标表现更差。

研究结论
将幸福感纳入心理健康一般因子模型，或有助于更精准地刻画复杂的心理健康状态，并可用于预测未来的功能状态；但此类模型仍需通过采用全面测评工具的重复验证研究，以及与其他备选模型的对比分析，来证实其有效性与应用价值。