Social support and disorder: toward an untangling of cause and effect.

Social support has become an increasingly popular concept and research tool for studying psychosocial factors in relation to many forms of disorder. The concept has promise from a research perspective and has considerable appeal due to its potential implications for treatment and prevention. However, the vast majority of studies examining support and disorder associations have contained major methodological problems. Using multiple regression procedures, the present study directly contrasts findings from retrospective analyses with results from three different types of prospective analyses (without controlling for prior symptoms, controlling for prior symptoms, and testing important interaction effects), both for psychological and for physical symptoms. The results demonstrate that support-disorder associations vary as a function of the design employed, the control variables included, and the type of disorder studied. The implications of these findings are discussed for evaluating the existing literature on support-disorder associations and for current concepts of social environmental predictors of disorder.