Life Stress, Vulnerability, and Depression: A Methodological Critique of Brown et al.

Publisher Website
Google Scholar
Abstract
The conceptualization and measurement underlying the `vulnerability' model of depression in women proposed by George Brown and his colleagues are criticized and an alternative, additive `strain' model is proposed. It is also argued that Brown et al.'s statistical methods are inappropriate. An alternative approach utilizing logit and linear probability estimation is proposed. With these techniques, the `vulnerability' and `strain' models are tested, using Brown and Harris' data. The `vulnerability' model is confirmed with the linear probability estimation but not with the logit estimation, while the `strain' model is confirmed with the logit estimation but not with the linear probability estimation. The significance of these results for an assessment of the work of Brown et al. and their implications for other research involving highly skewed dichotomous dependent variables are discussed.