In sum, previous research found that having a younger spouse is beneficial, while having an older spouse is detrimental for the survival chances of the target person. Most of the observed effects could not be explained satisfactorily until now, mainly because of methodological drawbacks and insufficiency of the data. The most common explanations refer to health selection effects, caregiving in later life, and some positive psychological and sociological effects.
RESEARCH QUESTIONS AND HYPOTHESES
In my model, exposure to risk of mortality depends on the individual’s own resources, those of their spouse, and their gender. Previous limitations are addressed by using detailed Danish register data in a time-dependent framework using hazard regression.
For men, the findings regarding the age gap to the spouse are relatively consistent: namely, that male mortality increases when the wife is older and decreases when the wife is younger. Previous research also indicated that mortality by the age gap to the spouse differs between the sexes, but none of the authors proposed reasons for this effect (Kemkes-Grottenthaler 2004; Williams and Durm 1998). The most common explanations of mortality differences by age gap to the spouse-health selection, caregiving in later life, and positive psychological effects of having a younger spouse-do not suggest large differences between the sexes. Thus, I hypothesize a similar pattern for women: namely, that the chance of dying increases when the husband is older and decreases when the husband is younger.
I also hypothesize that the duration of marriage has an impact on the mortality differentials by the age gap to the spouse. Previous studies speculated that marriages should be of sufficient duration to allow for any effects on mortality. This reasoning suggests that the mortality advantage of individuals who are younger than their spouses should not be observable in marriages of short duration.
In addition, I analyze the impact of socioeconomic status. Previous research (e.g., Kemkes-Grottenthaler 2004) indicated that the frequency of age heterogamy differs by social class. Generally, more highly educated persons and individuals with greater wealth are known to experience lower mortality, but no study has analyzed whether these socioeconomic variables might have an impact on the survival differentials by the age gap to the spouse. If the frequency of age heterogamy differs by social class, it could partially explain these survival differentials. Thus, I hypothesize that the socioeconomic characteristics of the target person and his or her spouse will change the effect of the age gap to the spouse on the target person’s mortality.
Previous research has argued that social norms and cultural background can explain the mortality differentials. Although Denmark is known to be a very https://besthookupwebsites.org/local-hookup/rochester/ homogeneous country, it is likely that social norms may differ between Danish and non-Danish as well as between rural and urban areas. Thus, I hypothesize that mortality by age gap to the spouse might differ by place of residence and by citizenship of the target person.
DATA AND METHODS
Denong the countries with the most sophisticated administration systems worldwide (Eurostat 1995). All persons living in Denmark have a personal identification number that is assigned at birth or at the time of immigration. This personal identification was a crucial part of the 1968 Population Registration Act, which introduced a computerized Central Population Register. This register serves as the source register for almost all major administrative systems in Denmark, which means that most registers can be linked by using the personal identification number. Today, many different authorities maintain about 2,800 public personal registers on almost all aspects of life. While the majority of these registers are administrative, a small proportion can be used for statistical or research purposes. Generally, the Danish registers are considered a source of detailed and exact information with a very low percentage of missing data. For this study, individual-level data from five different registers are linked with one another through the personal identification number. An overview of registers that are used for this analysis is shown in Table 1 .