In most marriages, men are older than their wives
Thus, the age groups are not only wide but also overlapping
Given my results, this composition favors men. Thus, the age gap between the spouses may in part explain why marriage is more beneficial for men than for women. My results also suggest that the possible selection bias caused by an insufficient length of partnership is of no importance in explaining the effects of the survival differences by the age gap to the spouse.
Previous research has suggested that selection and discrepancy from the social norm cause the mortality difference by the age gap to the spouse. This explanation was proposed in the 1970s, when social norms for mating behavior in general and especially for the age difference between partners were probably much stronger than today. My investigation supports this explanation for men but not for women. If social norms for the age gap to the spouse were the driving force of the observed mortality differentials, female mortality could be assumed to be lowest at ages where women are a few years younger than their spouses. Here, I find that mortality in women is lowest when a woman is the same age as her husband and increases with increasing age discrepancy.
I extend previous research of this area in several aspects. First, I apply a longitudinal approach. By using the Danish registers, it is possible to track all individuals from the date of their marriage until their date of death and to incorporate all life events-such as the death of the spouse, a divorce, or a remarriage-into the analysis of events within the observed period. The longitudinal approach avoids some of the drawbacks of earlier studies.
In a next step, it might be of interest to know whether the effects of the age gap to the partner can also be observed in longtime cohabiting couples or other kinds of partnerships, especially in same-sex couples
Another limitation of previous research that I overcame in this study is the age grouping into five-year age groups. Because of the age grouping in earlier studies, each of the spouse-age-difference intervals covered an eight-year period. Spouses who were stated as being in the same age group could differ plus or minus four years, while the difference for an individual who is married to a spouse in the neighboring age group varies from one to nine years. In my data set, the exact date of birth is known for every individual; thus, age and the age gap to the spouse are measured in days.
A further extension of previous research is also related to the data set. My study uses population data/register data, not samples as were used in previous research, to test these hypotheses. I was thus able to avoid many problems related to sampling methods while substantially increasing the statistical power.
It can be concluded that the driving force of the observed mortality differences by the age gap to the spouse remain unclear. Further research is necessary using models that test for additional multiplicative effects as well as for unobserved heterogeneity. A short coming of this study is that it does not include any behavioral or psychological aspects of the e from administrative registers. Future research should point in this direction as it is assumed to be of importance to account better for social values and norms as well as certain behavioral aspects.
Further research directions are of possible interest. In general, the age gap to the partner should affect the survival chances of members in all kinds of longtime partnerships between two individuals. Due to data limitations, studies, including the present one, have had to focus on married couples exclusively. The Danish data that are now available permit such analyses.