Campbell D T, O'Connell E J
Multivariate Behav Res. 1967 Oct 1;2(4):409-26. doi: 10.1207/s15327906mbr0204_1.
When traits are measured by the same method (or at the same time) the intertrait correlations are higher than when the intertrait correlation is across methods. An empirical investigation of the nature of such method factors is reported, in which the method factors seem to operate in a multiplicative rather than additive way, or in which larger method loadings are associated with larger trait loadings. An implicit-theory-of-personality explanation is offered for the case of ratings as methods. An attenuation model has more general applicability to the oases illustrated. The possibility is raised that the finding, if confirmed in other domains, demonstrates a fundamental innappropriateness of factor analysis for the componential analysis of indi6dual differences data.
当特质通过相同方法（或在同一时间）进行测量时，特质间的相关性高于跨方法的特质间相关性。本文报告了对这类方法因素性质的实证研究，其中方法因素似乎以乘法而非加法的方式起作用，或者较大的方法载荷与较大的特质载荷相关联。针对以评级作为方法的情况，提供了一种人格内隐理论解释。衰减模型对所说明的情况具有更广泛的适用性。文中提出一种可能性，即如果该发现在其他领域得到证实，那么它表明因素分析对于个体差异数据的成分分析存在根本的不适用性。
