Social network interactions: a buffer or a stress.

Forty-four caregivers to spouses with a diagnosis of Alzheimer's disease provided a stressed subject population considered at high risk for depression. Unlike more typical unidirectional measures of perceived social support quality, subject ratings were elicited separately as to how helpful as well as how upsetting each network member was in five different support categories. Correlations between perceived network "upset" and depression (Beck Depression Inventory) were highly significant, while in no case did perceived "helpfulness" relate to depression. Using stepwise multiple regression, the set of five support category Upset ratings predicted depression better than did helpful/upset ratios, which in turn predicted depression better than the Helpfulness ratings as a group. The implications of these findings for the conceptualization of social support and its measurement are discussed.