A number of issues come to mind on that.
One would be that on the proportion of women who are in marital relationships, women who are in their fifties, say, and approaching retirement, that proportion hasn't changed that much over the last 20 years. It's between 70% and 80%.
But what is particularly striking is that of the other 20% who aren't married or living common law, it's an x on the curve. It used to be that most of them were widowed. Relatively few of them were separated or divorced. Now what you find is that relatively few of them are widowed and the larger proportion of them are divorced. When we look at, for example, the incidence of low income among unattached women who are divorced, the incidence of low income is higher among that group than it is among those who are widowed or were never married.
So I think that change in family formation, with the incidence of late-life divorce, is one change that bears particular scrutiny.