Divorces per marriage may be approximately 50% (not sure).
This does not imply that (people who got divorced) / (people who got married) is 50%, and this is what the study measured.
Simple example: 3 guys have a happy first marriage. 1 guy has 3 failed marriages. Divorces/marriages = 3/6 = 50%. People who got divorced/people who got married =1/4= 25%.
Isn't overall the divorce rate is 50%? Something fishy is happening here.