There have always been all kinds of questions about how you decide who counts as homosexual. For example, in research it's important to have a logical and clear definition, and what you use can affect the outcome of your research.
If, say, you wanted to look at genetic or other physiological elements, or psychological charachteristics, whatever. Do you look at people who self-identify as gay or lesbian? Or people who have in fact had sexual relationships with people of the same sex? One encounter, 20 years ago? What if you are doing your research in a non-western country and the cultural context is different?
A lot of people today might say, obviously it's about self-identification, but that might give you a different set of people, and different results, than another measure, and it tends to favour a western model of sexuality which isn't ubiquitous by any stretch.
At the moment there is a real sense among young people that boring old hetersexuality is inherently discriminatory, so it's not that weird that stats around self-identification would be high.