They won't answer most other sex related questions either.
The pollsters almost invariably count such answers as being non-answers and they drop the respondents from the statistical base. That has the effect of doubling the homosexual answer.
The Gallup poll erred in making their practice known ~ and they got the sort of response you'd expect if a small group ALWAYS answered the phone (or did a pollster call back) and the larger group only responded at 9% of the time.
There's also a bias where wealthier Americans have more phone numbers and are more likely to be selected by a pollster ~ if a random number dialer is in use. That's the same sort of problem that had pollsters telling us Dewey had beaten Truman.
People who are issued cellphone service by their employers are also easily overrepresented in any telephone poll.
Again, pollsters substitute for these deficiencies in modern random sample polling with older methods ~ see AUGERIES ~ and simple rationalization.
I watched primary numbers. Where no one was much interested, they didn't show up to vote for the guy in the Fall either. Sure, there's not a lot of precision in that sort of analysis, but it gives you a trendline.
-PJ