Quality and quantity: the role of social interactions in self-reported individual health.

The public health literature on the detrimental effects of social isolation has shown that the quantity of social connections is positively correlated with individual health. Drawing on pooled cross-sectional data, we test this hypothesis on a representative sample of the Italian population. Our findings show that, in addition to the quantity of interactions, it is their quality--as measured by subjective satisfaction derived from relationships with friends--that works as the best predictor of self-reported health. The frequency of meetings with friends is significantly and positively correlated with good health in all regressions. However, when we add our measure of the quality of relationships to the probit equations, the statistical significance of "quantitative" measures is scaled down. Satisfaction with relationships with friends exhibits a positive and highly significant coefficient. Results of the multivariate probit analysis point out the potential role of unobservable variables suggesting the existence of endogeneity problems which require further investigation. We point out the existence of health disparities based on socio-economic status. There is a higher probability that poorer and less educated individuals report poor health conditions. The risk is even higher for unemployed and retired workers. This paper contributes to the literature in two substantive dimensions. This is the first empirical study of the relationship between social interactions and health in Italy. Second, we add to previous empirical studies by taking into account not only the frequency of various kinds of meetings but also indicators of their "quality", as measured by agents' subjective satisfaction with their social participation. The reliability of the analysis also benefits from the uniqueness and comprehensiveness of our dataset, which tries to overcome a structural deficiency in Italian data by merging information on agents' behaviours and perceptions with data on household income.