Department of Psychology, Wake Forest University, Winston-Salem, NC 27106, USA.
J Pers Assess. 2010 Jan;92(1):1-15. doi: 10.1080/00223890903379134.
As applied to many psychological phenomena, analysis of profile similarity has an intuitive appeal that masks complex statistical issues. Psychologists have long debated the methods of indexing similarity between 2 psychological profiles, but the double-entry intraclass correlation may be emerging as a preferred approach. Unfortunately, the double-entry intraclass correlation has not been articulated clearly in terms of fundamental facets of profiles--elevation, scatter, and shape--which prevents full understanding of its meaning. In this article, I (a) articulate these effects, (b) discuss potential limitations and confusions arising from these effects, (c) present a failure to replicate previous empirical findings regarding the double-entry intraclass correlation, and (d) present alternative recommendations for analysis of profile similarity. The conceptual, mathematical, and empirical points may enhance the insights emerging from analyses of profiles and profile similarity.
应用于许多心理现象的剖析相似性分析具有一种直观的吸引力，掩盖了复杂的统计问题。心理学家长期以来一直在争论索引 2 个心理特征之间相似性的方法，但双入口的组内相关系数可能正在成为一种首选的方法。不幸的是，双入口的组内相关系数没有在特征的基本方面——升高、分散和形状——中明确表达，这阻碍了对其意义的充分理解。在本文中，我 (a) 阐明这些效应，(b) 讨论这些效应可能产生的局限性和混淆，(c) 提出对双入口组内相关系数的先前实证发现的复制失败，以及 (d) 提出分析特征相似性的替代建议。概念、数学和经验点可以增强从特征和特征相似性分析中出现的洞察力。
