The problem with the "identifies as gay" statistic is that it's fundamentally flawed. Human sexuality represents a spectrum, and people who fall elsewhere on the Kinsey scale than the extreme ends tend not to be reported. Also in any culture where there's still a stigma attached to being gay some people aren't going to be honest about it.
Right. I think the survey of college students is telling. In a community or society where there's no stigma or discouragement toward same-sex behavior, a lot more people would probably be willing to explore or experiment with it. But when there is a stigma, even people who have experimented with it might think of it as an exception to their "actual" orientation.