In this paper we will review the research we have been doing over the last few years attempting to differentiate among the various facets of the social support construct. We will cite both correlational questionnaire studies as well as behavioral studies, and will conclude with the opinion that the effects of support are much more complicated than the original proponents of the concept had imagined. In all probability, we will raise more questions than we answer, and will offer the hypothesis that the effectiveness of support may not reside in any particular behavior of significant others, but in how that behavior and ensuing relationships are perceived. We also will suggest that individuals are not simply passive recipients of support, but in part, determine the nature of the supportive behavior they receive from others.

The original epidemiological research that sparked the current wave of interest in social support demonstrated that psychosocial assets, broadly defined to include both personal and environmental characteristics, had a role in positive health outcomes (de Araujo, van Arsdel, Holmes & Dudley, 1973; Nuckolls, Cassel, & Kaplan, 1972). This message was important in order to counter the then dominant medical belief that disease and symptomatology were exclusively programmed by internal biological variables (Cassel, 1976; Cobb, 1976). Unfortunately, the field did not go beyond the demonstration that environmental and life style variables influence health outcomes. Little attention was given to the fact that the predictors used were extremely heterogeneous and were confounded with other psychosocial variables, such as personal competence (Heller, 1979). Researchers worked toward repeated demonstrations of predictive validity (that support influenced health) and paid little attention to construct validity (Heller,