News | July 1, 2015


© MPIDR / Peter Wilhelm

On July 1, MPIDR doctoral candidate Oliver Wisser successfully defended his doctoral dissertation. In his work, he refined a mathematical model that can be used by demographers to describe mortality. The goal was to reflect better the gender-specific differences in mortality.

As people age, the likelihood that they will die increases. The change in this likelihood over the life course is usually described using a mathematical model called the Gompertz law of mortality. Because this law does not apply to middle age, the demographer William Makeham later introduced a constant that corrected for this mistake by accounting for the fact that mortality can be influenced by the environment.

However, accounting for environmental influences on mortality is insufficient and simple to fully identify the underlying factors may even generate erroneous results when the goal is to describe sex-specific differences in mortality. The model may, for example, calculate that the rate of aging is higher among elderly women than among elderly men—even though in industrialized countries women on average live longer than men. To help correct this flaw, Oliver Wisser introduced behavioral risk factors into the model, as behavior can play an important role in mortality among young and middle-aged adults.

The evidence indicates that men tend to die earlier than women not because of biological differences, but because men are more likely than women to engage in high-risk behavior. An example is the gender gap in behavior on the road. Young men in particular tend to engage in risky driving behavior, and are thus more likely than young women to die in a car accident. Another example is the high level of alcohol and tobacco consumption among men. Because of these factors, mortality in the young adult years is lower among women than among men. But according to the mathematical Gompertz-Makeham model, mortality among women appears to increase faster with age.

Very heterogenous: Because mortality varies considerably between the sexes and between different countries, it is difficult to describe mortality using a generally applicable mathematical model. In his doctoral dissertation, Oliver Wisser worked on refining such a model so that it better reflects gender-specific differences.

© Oliver Wisser / MPIDR

“To arrive at a good model that reflects these differences, it is necessary to develop a concept that distinguishes between the external behavioral causes and the biological causes of aging,” Oliver Wisser explains. And this is exactly what he did in his dissertation: He introduced into the Gompertz model a factor he calls ‘middle mortality.’ To find out whether this new model—called the κ-γ-Gompertz model—works, he tested it using data from the Mortality Database. This database, which is maintained by the MPIDR, contains data on mortality developments in 37 countries.

The new model makes more sense from a statistical perspective and is an extension of the classical approach. The results also show that male behavior is mainly responsible for the divergence in life expectancy between the sexes in the last century, and thus confirms the findings of other empirical studies. One advantage of the model is that it describes behavior-related mortality, using two simple variables and thus allowing for simple comparisons between countries with very different life expectancies.

However, this model is also flawed, as it does not fully reflect the fact that for some causes of death, such as heart and lung diseases among smokers, the effects of behavior are subject to temporal delays. This problem is especially evident among the male population of Europe and other countries in the second half of the 20th century. In the future, the model could be improved through the inclusion of, for example, cause-of-death statistics. Using these kinds of statistics would also be helpful in developing more realistic mathematical mortality models.