[usaar333]

> The probability of a marriage surviving also decreases in a relatively linear fashion.

How did you arrive at this conclusion? Looking at the rates, it is definitely far greater than linear, indeed the 5 year survival point from point X is constantly increasing (81% survival first 5 years; 90% years 15-20). Also bear in mind that "survival" includes not dying, which might throw inferring divorce rates off by a few percent at 20+ years.

Extrapolating to 25 years, you'd get a 49.5% divorce rate.

Again, demographics must be accounted for. Almost the entire audience here is in the higher education demographic which has radically lower divorce rates; you're looking at sub-30% divorce rates.

[john_b]

I should have been more specific. I didn't mean that the probability of a marriage surviving some fixed time interval decreases linearly with the duration of the marriage. Everyone knows that marriages that have lasted longer are less likely to end than newer marriages. What I meant is that the cumulative probability of a marriage surviving decreases roughly linearly from the 5 year point onward, as Tables 5 & 6 show (and Fig. 4 shows graphically).

As an unmarried person, it's the latter that is of concern to me. If I had been married for 15-20 years already, the probability you cite would be more pertinent.