Fair enough. It's hard to know how you'd even define the population to sample from, anyway, let alone get answers from a statistically valid cross-section of them -- polls like this are pretty much the best anyone's ever going to get, even if the numbers have a big and indeterminate margin of error.

I must admit, I was surprised that the numbers were as high as they were, although on reflection not astonished given my own flist! I suspect that the actual percentages would be much lower if you really could do a wide-ranging survey, but it's still a substantial total. And considered as polls of something like "people very actively involved in slash fandom" rather than "all slash fans" the self-selecting nature probably makes a lot less difference.