I had originally posted a poll with Often/Sometimes/Never answers, but I realized, like you point out, that this poll would have heavy selection bias. So I changed the poll to gather reasons why people select. That's probably more interesting data than Yes/No anyways. :)
The ratio of people that does this is something that you could accurately measure if you have a somewhat popular website. (The reason why not so much, of course. I doubt many people have a conscious reason for this behaviour anyway. I don't.)
