I don't think anyone ever claimed this represented the overall population. It's clearly the data set for OkCupid users, which may be a decent proxy for online dating users. Unless there is any reason to doubt the accuracy of the data presented, I don't see the problem. Black women on OkCupid are not as desirable as white women. We may not like the results, but doesn't make it invalid.

To use your analogy, if 95% of people going through burger king drive through prefers Coke vs Pepsi, that's relevant data for that purpose.