Stress, alcohol use and alcohol-related problems: the influence of negative and positive affect in two cohorts of young adults.

Abstract

Past research has suggested that trait-based positive and negative affect mediate the relationships between self-report measures of stress and health. We extended these assumptions to the study of health-related behaviors, specifically the much studied stress-alcohol relationship. In addition, we sought to replicate previous findings showing that stress is more strongly associated with alcohol problems and not alcohol use as many theories (e.g., tension-reduction hypothesis) have assumed. Structural equation modeling was used to test a mediating variable model in two cohorts (574 and 239 subjects) of randomly selected young adults between 18 and 35 years of age. Participants were from the 1996 data wave of the Niagara Young Adult Health Study, a community-based study incorporating a modified cohort-sequential design. The larger (older) cohort consisted of subjects initially contacted in 1990. For both cohorts, the measurement model proved to be very reliable. The final structural models did not match the predicted model. The correlations between the Stress and Alcohol Use latent variables (necessary to test a mediational model) were not significant in either cohort. However, in the older cohort, trait-based positive affect and hostility acted as intervening variables, allowing for an indirect relationship between stress and alcohol use. In the younger cohort, this indirect pattern was not present because none of the latent affect variables led to alcohol use. In both cohorts, stress predicted alcohol problems, even after controlling for the relationships between stress, affect and alcohol use. Significant cohort differences in the structural models were observed. The assumptions regarding the mediation of affect on the relationship between stress and health cannot be extended completely to the stress-alcohol relationship. Two reasons for this are: (1) the differences between the two age cohorts in the relationships between stress, affect and alcohol use and (2) the finding that stress is more predictive of alcohol-related problems than it is of alcohol use.