The first possible reason for sex differences could be differences in health selection
In Model 3 of Table 3 , I test the hypothesis that socioeconomic status affects the mortality differentials by the age gap to the spouse. This model includes measures of the target person’s highest educational degree and wealth as well as the variables already included in Model 2. The results show that both socioeconomic variables are important predictors of survival differences. For men, adding these measures to the model reduces the relative risk of dying when they are younger than their wives, but it increases the survival advantage when they are older than their wives. For women, adding measures of socioeconomic status has virtually no impact when they are younger than their husbands but slightly increases the chance of dying when they are older than their husbands. In results not shown here, I tested another model that included an interaction between the socioeconomic variables and the age gap to the spouse.