Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

A growing body of research linking marriage to health suggests that marriage confers health benefits, particularly to men. It is still unclear, however, whether an observed health advantage is due primarily to marriage per se or to the selection of psychologically and physically healthier individuals into marriage. The most recent research in this area uses longitudinal data to net out this selection effect. However, most studies on this topic focus on one or two measures of health at a time, and often neglect cohabitation, an increasingly common form of union status. In this research, we improve on past studies by using data from six waves of the longitudinal Canadian National Public Health Survey (NPHS) to estimate the effect of marital status (married, cohabiting, never married, separated, divorced, widowed) on a variety of health and behavior outcomes. An advantage of using data from Canada is that, in the U.S. correlations between marriage and improved health may be associated with the greater probability of health insurance among the married. The presence of the Canadian health system reduces differences in health coverage as a possible explanation for health differentials. Overall, we aim to inform policymakers by contributing to the public discussion about the role of relationship status in the well-being of individuals.

With respect to behavioral outcomes, we examine obesity and smoking. Duncan et al. (2006) using U.S. data find no effect of marriage and cohabitation on smoking while Averett et al. (2007) find that partnered individuals are thinner, but after controlling for time-invariant selection marriage is often associated with weight gain. Our work is particularly timely because the proportion of Canadian adults who are obese has grown from 5.6% in 1985 to 15.5% in 2005. Recent increases in the incidence of obesity in the U.S. and Canada have garnered considerable attention from policymakers in both countries because these drastic changes in the distribution of weight have significant health consequences and so, in addition to these behaviors, we consider measures of health status. These include a self-reported health measure, a measure of the number of chronic conditions, measures of stress and indicators variables for heart disease and diabetes.

We describe five theoretical explanations for a link between relationship status and health and behavior outcomes (selection, protection, social obligation, physical activity, and marriage market) and evaluate each as data allow. The selection hypothesis indicates that healthier individuals are more likely to be selected into marriage and is the only one of our explanations which indicates that marriage itself is not the cause of better health outcomes. Although we cannot directly test all of these hypotheses, the NPHS does have information on exercise and diet which we incorporate into our analyses.

Taking advantage of the longitudinal aspect of the NPHS, we estimate individual fixed effects models which net out the time-invariant selection effect leaving us with the potentially causal effects of union status controlling for the effects of aging and other respondent characteristics. We run our models separately for men and women.