The trade-off between accuracy and precision in latent variable models of mediation processes.
The trade-off between accuracy and precision in latent variable models of mediation processes.
复制标题
DOI：
10.1037/a0024776
复制
发表时间：
2011-12
影响因子：
7.6
通讯作者：
Shrout PE
中科院分区：
文献类型：
--
作者：
Ledgerwood A;Shrout PE
社会心理学家非常重视理解机制，并经常使用中介分析来阐明潜在的影响过程。这种分析可以使用观察变量(例如，典型的回归方法)或潜在变量(例如，扫描电子显微镜方法)进行，在这些方法之间进行选择可能是一个比研究人员通常意识到的更复杂和更重要的决定。本文通过检验潜变量模型与观测变量模型中精确度和精确度之间的相对权衡，增加了关于调节的文献。虽然过去的工作表明，潜变量模型往往会产生更准确的估计，但我们证明，观察变量模型往往会产生更准确的估计，并在典型的三变量中介模型中，通过不同水平的影响大小和可靠性，从理论和经验上检验这种相对权衡。我们讨论了社会心理学家寻求发现中介变量的含义，并推荐了在中介分析中最大化准确性和精确度的实用方法。
Social psychologists place high importance on understanding mechanisms, and frequently employ mediation analyses to shed light on the process underlying an effect. Such analyses can be conducted using observed variables (e.g., a typical regression approach) or latent variables (e.g., a SEM approach), and choosing between these methods can be a more complex and consequential decision than researchers often realize. The present paper adds to the literature on mediation by examining the relative tradeoff between accuracy and precision in latent versus observed variable modeling. Whereas past work has shown that latent variable models tend to produce more accurate estimates, we demonstrate that observed variable models tend to produce more precise estimates, and examine this relative tradeoff both theoretically and empirically in a typical three-variable mediation model across varying levels of effect size and reliability. We discuss implications for social psychologists seeking to uncover mediating variables, and recommend practical approaches for maximizing both accuracy and precision in mediation analyses.