People like to throw that number out there but mostly I think they’re just innumerate or justifying their own marital failures. Sure, 50% of marriages end in divorce. But that’s just one dumb average, and when you look at the underlying factors that affect divorce rates you’ll find they’re not that surprising.
Getting married very young, getting married due to a kid, getting married after dating for a short period, working in shitty jobs, living in certain states, marrying someone who has ever been unfaithful, marrying someone who has ever experienced a divorce before.
Not trying suggest one should not necessarily avoid anyone who falls into the above. But, while I cannot find any raw datasets to try and estimate this for myself, most of this have independent effects of double digit drops. Assuming a practical amount of multicollinearity, I would wager my best predicted probability of divorce is lower than 10% without considering any personal beliefs on our relationship.
Heck just looking at single variable analyses from various studies I see reason to believe my predicted divorce rate is 30-60% lower than baseline.