Sexual abuse, family environment, and psychological symptoms: on the validity of statistical control

M. R. Nash, T. L. Hulsey, M. C. Sexton, T. L. Harralson, and W. Lambert (1993) reported on the effects of controlling for family environment when studying sexual abuse sequelae. Sexual abuse history was associated with elevated Minnesota Multiphasic Personality Inventory and Rorschach scores in a sample of 105 women, but many of the reported differences disappeared when a Family Functioning Scale score was used as a covariate. The present article considers the findings of Nash et al. in terms of the theoretical and statistical constraints placed on analysis of covariance and other partializing procedures. Because family dysfunction is not always causally antecedent to sexual abuse, and given the quasi-experimental quality of most abuse research, the use of covariate techniques to test hypotheses about the causal role of family environment in the impacts of sexual abuse may be ill advised. Analyses of a 2,964-subject data set illustrate these concerns.