The other flaw is that by definition people on dating sites are not currently in a relationship (well, usually), so of course they're less likely to get married in the near future. I suspect the 12.4 ratio would be similar for active OKCupid users.
Not true for OKC. Very poly friendly.
I'm not saying it didn't happen... but it might be kind of like saying "planes aren't safe, I had a friend that was in a plane crash."
I'd imagine how welcoming people are to others in poly relationships largely depends on where you are. People in El Paso, TX (where my friend and I live) are likely less open to that sort of thing than people elsewhere.
I like the corelation you can build from the site: By giving you bad matches to increase business, each person looking gets less likely to find someone because they have to send so many emails. Note that the 30% chance to get a response * 1/30 chance that an account is not active != 1/100 chance of a response. The 30% chance to get a response can already take into account that a bunch of people may not be active.
You can play with the numbers. I think the final guesses are not that accurate, but it does put things into perspective regarding the marketing numbers.