ABSTRACT

Much of the discussion about the effects of social support on health has revolved around the issue of stress-buffering versus direct (main) effects on health. (For overviews see Alloway & Bebbington, 1986; Cohen & Wills, 1985.) When social support is beneficial for a person’s health in some proportion to his or her exposure to stress, it is called a buffer effect. A beneficial effect of support that is independent of concurrent stress levels is called a main effect. These two kinds of effects are not mutually exclusive. The presence of a stress-buffering effect is usually inferred when stress and support interact statistically in their association with health outcomes (i.e., when the association between stress and ill health is stronger at low levels of support than at high levels). There are a number of different approaches to measure this association (e.g., linear, nonlinear), and also to measure the differences in its strength (e.g., hierarchical and nonhierarchical). The first issue was the topic of a major controversy in the field (e.g., Tennant & Bebbington, 1978; Brown & Harris, 1978a), whereas the second has received comparably little attention.