The purpose of this study was to investigate the relationship of a constellation of family variables (family stress, family resources, family coping, sibling position, family of origin history of symptoms, and extended family social support) to selected symptoms: anxiety, depression, and somatization. The Bowen Theory considers the family as an interactive system experiencing stress in varying degrees. Bowen's conceptualizations of family emotional system, chronic anxiety and symptom formation, multigenerational transmission process, sibling position, and extended family were applied to this study.

Two hundred ten (n = 210) volunteer graduate nursing students comprised the convenience sample. Each subject completed: a demographic background sheet; Family Inventory of Life Events (FILE); Family Inventory of Resources for Management (FIRM); Family Crisis Oriented Personal Scales (F-Copes); and Symptom Check List-90 Revised (SCL-90R).

Factor analysis was done on the basis of multicollinearity among the variables of interest and resulted in computations of composite factors for anxiety, depression, and somatization (renamed "symptom experience") and for extended family social support.

Regression analysis revealed that family stress, family resources, and family coping combined to explain 17% of the variability in symptom experience (R('2) = .17, p < .01) with family stress being the single best predictor. Extended family social support correlated weakly with symptom experience (r = -.15, p < .05) for those subjects whose parents were alive (n = 154). Family of origin history of symptoms proved to be a more important variable than hypothesized. This variable correlated moderately with symptom experience (r = .33, p < .001), being the best predictor of symptom experience in a regression analysis when compared with sibling position and extended family social support (R('2) = .11, p < .01). Extended family social support proved to be a disappointing variable, perhaps due to the use of non-Bowen tools. T-tests performed on symptom experience according to sibling position failed to show any significant differences. Additional findings included: (1) significant differences on t-testing between subjects whose parents were both alive versus subjects with one or both parents deceased on symptom experience (t = 2.91, p < .005) and family of origin history of symptoms (t = 3.94, p < .001); (2) significant correlations of subject symptom experience with all categories of family of origin history of symptoms (grandparents, r = .34, p < .001; father, r = .23, p < .001; mother, r = .21, p < .005; and siblings, r = .16, p < .05).

These findings were discussed in light of the Bowen Theory as well as other research on family systems, with suggestions for replicative studies and tool development.