Data sure does lie if you interpret data selectively. If the alternative to a divorce is an unhappy marriage, then divorce can produce better outcomes than staying together. Just because people are divorced doesn't mean they're unhappy, and optimizing for marriage rather than happiness would be foolish.

Furthermore, individuals getting divorced have a disproportionate effect on the marriage rate, because, obviously, they have strictly more marriages (and (one expects) more partners) than the people who haven't been divorced.