Bernstein I H, Teng G, Grannemann B D, Garbin C P
Department of Psychology, University of Texas, Arlington 76019.
J Pers Assess. 1987 Winter;51(4):522-31. doi: 10.1207/s15327752jpa5104_4.
Bernstein and Garbin (1985b) suggested that the Minnesota Multiphasic Personality Inventory's major clinical scales (excluding Scales ? and 5) can be approximated by an oblique three-component structure: (a) Profile Elevation, (b) Test-taking Attitudes, and (c) Optimism-Pessimism, collectively termed the salient weight model. In this study, we found that both this model and the MMPI's principal component structure remain invariant across race, sex, and, as previously noted, context of testing (job applicants vs. inmates in correctional institutions). We further noted that several alternative definitions of Profile Elevation provide equally satisfactory representation of the relations among the scales. This factor invariance is necessary, but not sufficient, for the MMPI to be viewed as unbiased.
伯恩斯坦和加宾（1985b）认为，明尼苏达多相人格调查表的主要临床量表（不包括量表?和5）可以由一个斜交的三因素结构近似表示：（a）剖面图高度，（b）应试态度，以及（c）乐观-悲观，统称为显著权重模型。在本研究中，我们发现，该模型和明尼苏达多相人格调查表的主成分结构在种族、性别以及如前所述的测试背景（求职者与惩教机构中的囚犯）方面均保持不变。我们还注意到，剖面图高度的几种替代定义对各量表之间的关系提供了同样令人满意的表示。这种因素不变性对于明尼苏达多相人格调查表被视为无偏差是必要的，但不是充分的。
