Abstract
The first wave of a longitudinal and prospective community study of minor psychiatric disorder among a cohort of socially disadvantaged women has been completed. The study hopes to extend the social model of depression, as initially proposed by Brown and Harris [1], by examining the mediating effects of personality factors and social support. Findings from previous studies have been largely restricted to associations between variables which may be at least partly attributable to methodological limitations in design and measurement techniques: specifically, the use of inherently biased sampling methods, single interview, retrospective data collection and less standardized diagnostic criteria to determine caseness. This paper discusses design strategies which were adopted to reduce the confounding between variables resulting from these limitations. The effectiveness of these strategies is evaluated in light of some methodological and ethical issues which have arisen during the recruitment of participants and the first wave of data collection.