It also seems to be based on a statistical misinterpretation. The divorce rate and marriage rate per 1000 people have both been declining for decades, and the divorce rate hovers around half of the marriage rate, but this doesn't equate to 50% of marriages ending in divorce. Since divorces generally occur a significant time after the marriage, and the marriage rate is also declining, comparing the present rate of divorce to the present rate of marriage results in a misleading figure.
Unfortunately, I haven't been able to locate any raw statistics which would allow me to discern the true rate. This writer paints a much rosier picture, but he also doesn't link to the actual statistics he used or what process he used to determine the rates he gives, so I'd be wary about taking it at face value. I haven't found any other resource so far that attempts to put numbers to the true rate, although I have found some others which note the problems with the conventional statistics.
My number is from from Stevenson-Wolfers (alt alt), particularly Figure 2, page 37. It is not a misinterpretation. To be more precise, about 50% of first marriages from the 70s made it to 25 years. First marriages from 80s appear to a do a little better, heading for maybe 55%, but it is too soon to tell [this previously said that it was too soon only because of binning decisions, but that's wrong; the problem is that the data was collected in 2001].
Dalrock writes an interesting article related to Dr. Helen Smith's book the Marriage Strike. I really have to bump it up on my too rapidly growing reading list. (^_^)