The structure of schizophrenia symptoms has a substantial impact on the development of pharmacological and psychosocial interventions. Typically, reflective latent variable models (eg, confirmatory factor analysis) or formative latent variable models (eg, principal component analysis) have been used to examine the structure of schizophrenia symptoms. More recently, network analysis is appearing as a method to examine symptom structure. However, latent variable modeling and network analysis results can lead to different inferences about the nature of symptoms. Given the critical role of correctly identifying symptom structure in schizophrenia treatment and research, we present an introduction to latent variable modeling and network analysis, along with their distinctions and implications for examining the structure of schizophrenia symptoms. We also provide a simulation demonstration highlighting the statistical equivalence between these models and the subsequent importance of an a priori rationale that should help guide model selection.

中文翻译：

---

精神分裂症的症状结构：潜在变量建模与网络分析的含义

精神分裂症症状的结构对药理学和社会心理干预的发展具有重大影响。通常，反射性潜变量模型（例如，验证性因素分析）或形成性潜变量模型（例如，主成分分析）已被用于检查精神分裂症症状的结构。最近，网络分析作为一种检查症状结构的方法出现了。然而，潜在变量建模和网络分析结果可能导致对症状性质的不同推断。鉴于正确识别症状结构在精神分裂症治疗和研究中的关键作用，我们介绍了潜变量建模和网络分析，以及它们的区别和对检查精神分裂症症状结构的影响。