Marriage, Health and Later-Life Mortality: Results from a Sample of Union Army Veterans

Sven E. Wilson, Brigham Young University

This study exploits a rich historical collection of data drawn from the pension, military and census records of nearly 10,000 Union Army veterans who were living in 1900. The purpose is to investigate the role that marital status has on mortality. Simple descriptive statistics show that married veterans have a 10-year survival rate about 5 percentage points higher than non-married veterans. Proportional hazard models reveal that these marital status effects are robust to the inclusion of both demographic variables and health status at baseline. The strongest effects are for widowers, though the divorced/separated and never married also have higher mortality than the married. Finally, estimates are also obtained separately for those with both “healthy” and “unhealthy” BMI levels. These estimates show that the marital status effects exist primarily among less healthy veterans, which implies that marriage is an important buffer against mortality among those whose mortality risks are high.