Psychol Assess. 2018 Sep;30(9):1160-1173. doi: 10.1037/pas0000566. Epub 2018 Mar 29.
The magnitude of components of variance in trait scales-true score, method variance, and error-can be estimated from information on the mono-method correlations among personality traits within a domain and on cross-observer agreement on domains and facets. Estimates of these components in NEO Inventory facet scales were compared with prior estimates that were based on a consideration of internal consistency and retest reliability (McCrae, 2015). Together, results suggested that (a) about 40% of the variance in self-reports and single informant ratings is due to method variance; (b) as with substantive traits, method biases exist on several different levels, some broad, some narrow; and (c) consequently, a large number of distinct biases affect personality scale scores. Method biases beyond acquiescence and evaluation were also found in a clinical instrument, the Personality Inventory for DSM-5. Because many biases appear to be idiosyncratic, it is unlikely that validity scales could be created to assess or control all of them. These findings underscore the value of utilizing multiple informants in research and individual assessment. To the extent that they can be distinguished from valid variance, method biases are themselves of clinical interest as potentially important elements of the self-concept. (PsycINFO Database Record
特质量表各分量方差的大小——真分数、方法方差和误差——可以根据特质领域内单方法相关信息和领域和方面的跨观察者一致性来估计。与先前基于内部一致性和重测信度考虑的估计相比，对 NEO 量表方面的这些分量进行了比较（McCrae，2015）。总的来说，结果表明：（a）自我报告和单一信息提供者评分的方差约有 40%归因于方法方差；（b）与实质特质一样，方法偏差存在于几个不同的层次上，有些广泛，有些狭窄；（c）因此，许多不同的偏差会影响人格量表分数。在 DSM-5 人格量表等临床仪器中也发现了除了默许和评估之外的方法偏差。由于许多偏差似乎是特质性的，因此不太可能创建有效性量表来评估或控制所有偏差。这些发现强调了在研究和个体评估中利用多个信息提供者的价值。在能够与有效方差区分开来的程度上，方法偏差本身就具有临床意义，因为它们是自我概念中潜在的重要组成部分。（PsycINFO 数据库记录）
