Snowden L R, Campbell D R, Nelson L S
J Pers Assess. 1984 Jun;48(3):271-8. doi: 10.1207/s15327752jpa4803_8.
Relationships between sociodemographic variables and the MMPI, on the one hand, and several dimensions of problem drinking, on the other, were examined from a multiple regression strategy that enhanced control and clarified sizes of effects. To investigate a sample of 385 persons convicted on drunk driving, a plan was implemented calling for entry of five sociodemographic variables first in equations, to assess and control for their effects, before entering 14 MMPI validity and clinical scales, and entering finally product terms representing sex-by-MMPI interactions. These factors were used to predict six problem drinking criteria, all continuous, based on factors and total scores from the Michigan Alcoholism Screening Test (MAST). Results indicate substantial variance attributable to personality factors in comparison with sociodemographic ones, and a pattern of prediction confirming generalizations reported in the literature, but also qualifying them by indicating confounding and interactions.
一方面研究了社会人口统计学变量与明尼苏达多相人格调查表（MMPI）之间的关系，另一方面研究了问题饮酒的几个维度之间的关系，采用多元回归策略，增强了控制并明确了效应大小。为了调查385名因酒后驾车被定罪的人员样本，实施了一项计划，要求先在方程中输入五个社会人口统计学变量，以评估和控制其效应，然后再输入14个MMPI效度和临床量表，最后输入代表性别与MMPI交互作用的乘积项。这些因素被用于预测六个基于密歇根酒精中毒筛选测试（MAST）的因素和总分的问题饮酒标准，所有这些标准都是连续的。结果表明，与社会人口统计学因素相比，人格因素导致的方差很大，预测模式证实了文献中报道的一般情况，但也通过指出混杂因素和交互作用对这些情况进行了限定。
