When you consider the birthday paradox, and consider demographics, it probably doesn't. If you assume that DOB's are evenly and randomly distributed over the l

When you consider the birthday paradox, and consider demographics, it probably doesn't. If you assume that DOB's are evenly and randomly distributed over the last 100 years (1 of ~36500 values) then the probability of none of 100 people sharing a DOB is only ~87%. If you tuned for demographics the true stats would be much worse. That said, you probably have anti-clustering aspects - parents obviously can't share birth dates with their children, and siblings can't either (unless twins). But! couples tend to be of a similar age...so, tricky.