An Empirical Analysis of the Association between Social Interaction and Self-Rated Health

Abstract

To empirically investigate the relationship of social interactions with self-rated health status. Cross-sectional study. The United States, 1972-2008. 53,043 respondents interviewed. In-person interviews were conducted by the National Opinion Research Center. Health is measured via self-assessment, and a binary indicator defines whether the respondent reports excellent or good health. Impersonal social interaction is binary, reflecting membership in at least one organization; personal interaction is binary, reflecting whether the respondent visits relatives, neighbors, friends, or bars at least once a month. Demographic, region, and time variables are standard controls. Probit models were estimated at aggregated and disaggregated (race, gender, age, and education) levels. Impersonal interaction fell 5% between 1974 and 1994, with deeper declines (12%) in personal interaction. Neighbor visits fell most (24%), but friends gained (5%). Marginal estimates indicate both personal and impersonal interactions are associated with better health. However, friends show the strongest effect. A 1-SD increase in the probability of meeting friends versus joining a club or association is associated with a higher (.019 vs. .014) probability of being in very good or excellent health. Social interaction, impersonal or personal, is significantly associated with health. Public policy facilitating greater interactions of the type that benefits different demographic groups may be health promoting. Results are suggestive of potential network effects at the individual and community levels, which have implications for designing and targeting more effective health interventions.