I got the Shaunti Feldhahn divorce data book much sooner than expected. I haven’t had a chance to read it all the way through yet, but she is using census stats, so isn’t just making up stuff. That said, the 25% number is an estimate derived from taking widows out of the data on first marriages where the person is still married to their first spouse. Otherwise, the number is 72% of first marriages with first spouse.
The 50% number was a projection based on trends at the time it was formulated, and even then it was 40-50%.
Anyone saying likelihood of marriage ending in divorce is 50% is not looking at how many ever-married people have divorced.
What did happen, and she notes this, is that before the 1970s divorce spikes, marriages remained intact 85% of the time. That dropped to 70-72% (remember, this includes intact marriages where death ended the marriage, otherwise it’s closer to 75%) by 1985 and stayed there. Interested parties might look at that stability and contrast it with fertility declines over the same period of time.
The interesting thing to me is that a 25% divorce rate is miserably bad, but there is enough data to show it’s remained constant over several marriage cohorts. And it’s, well, it’s half of 50%. I haven’t gotten to the part where she compares by age bracket, but that should be interesting.