Mõttus René, Realo Anu, Allik Jüri, Esko Tõnu, Metspalu Andres, Johnson Wendy
Centre for Cognitive Ageing and Cognitive Epidemiology, Department of Psychology, University of Edinburgh, Edinburgh, United Kingdom; Department of Psychology, University of Tartu, Tartu, Estonia.
Department of Psychology, University of Tartu, Tartu, Estonia.
PLoS One. 2015 Mar 9;10(3):e0119667. doi: 10.1371/journal.pone.0119667. eCollection 2015.
The study investigated differences in the Five-Factor Model (FFM) domains and facets across adulthood. The main questions were whether personality scales reflected coherent units of trait development and thereby coherent personality traits more generally. These questions were addressed by testing if the components of the trait scales (items for facet scales and facets for domain scales) showed consistent age group differences. For this, measurement invariance (MI) framework was used. In a sample of 2,711 Estonians who had completed the NEO Personality Inventory 3 (NEO PI-3), more than half of the facet scales and one domain scale did not meet the criterion for weak MI (factor loading equality) across 12 age groups spanning ages from 18 to 91 years. Furthermore, none of the facet and domain scales met the criterion for strong MI (intercept equality), suggesting that items of the same facets and facets of the same domains varied in age group differences. When items were residualized for their respective facets, 46% of them had significant (p < 0.0002) residual age-correlations. When facets were residualized for their domain scores, a majority had significant (p < 0.002) residual age-correlations. For each domain, a series of latent factors were specified using random quarters of their items: scores of such latent factors varied notably (within domains) in correlations with age. We argue that manifestations of aetiologically coherent traits should show similar age group differences. Given this, the FFM domains and facets as embodied in the NEO PI-3 do not reflect aetiologically coherent traits.
该研究调查了成年期五因素模型（FFM）各领域和层面的差异。主要问题是人格量表是否反映了特质发展的连贯单元，从而更普遍地反映了连贯的人格特质。通过测试特质量表的组成部分（层面量表的项目和领域量表的层面）是否显示出一致的年龄组差异来解决这些问题。为此，使用了测量不变性（MI）框架。在一个由2711名完成了《大五人格量表第三版》（NEO PI-3）的爱沙尼亚人组成的样本中，超过一半的层面量表和一个领域量表在18至91岁的12个年龄组中未达到弱测量不变性（因子载荷相等）的标准。此外，没有一个层面量表和领域量表达到强测量不变性（截距相等）的标准，这表明相同层面的项目和相同领域的层面在年龄组差异方面存在变化。当项目针对其各自的层面进行残差分析时，其中46%具有显著的（p < 0.0002）残差年龄相关性。当层面针对其领域得分进行残差分析时，大多数具有显著的（p < 0.002）残差年龄相关性。对于每个领域，使用其四分之一的随机项目指定了一系列潜在因子：这些潜在因子的得分在与年龄的相关性方面（在各领域内）差异显著。我们认为，病因学上连贯的特质表现应显示出相似的年龄组差异。鉴于此，NEO PI-3所体现的FFM领域和层面并未反映出病因学上连贯的特质。
