Dataviz angle here... To me, it's like determining the age of a population. Saying that the population is 40 years old on average doesn't have much value. Inst

Dataviz angle here... To me, it's like determining the age of a population. Saying that the population is 40 years old on average doesn't have much value. Instead, we need a richer representation, like an age pyramid [1]. Plot on a histogram the number of divorcees for each year they got married. Of course there is going to be few divorces for those married this year, and a lot more for those married 50 years ago. Then you can compare the shapes of the histogram over time or between regions. And that can yield insight on the matter. [1] http://en.wikipedia.org/wiki/Population_pyramid http://en.wikipedia.org/wiki/Population_pyramid