srw
It's a bit more complicated than that...
Again I'm going from memory, but I believe that it's a very long-tailed distribution, and the people who contribute the most to the data are the millions who ride 5 times a year, going 10 miles each time, rather than the few tens of thousands who ride every day. And the survey-based data is coming out so far adrift from the counter data that it's impossible that the methods used to collect and adjust the counter data are right. Looking at the list of locations in Central London you can see why....What sort of survey based method? If you asked a few thousand cyclists to tell you how far they had ridden in the last year, the data would probably be as reliable as asking them to self-report the size of their penises. You need a ruler, actually for some firm data.
(Oh, and we can be very sure that about half of cyclists won't mis-estimate their penis length. Breast size, though, I gather is a different matter - some horrifically large proportion are wearing the wrong size bra if you believe lingerie sellers.)