On Thu, Jun 2, 2011 at 21:42, Cameron Byrne <cb.li...@gmail.com> wrote: ....> > Pure speculation here, but these stats that you refer to are not a > scientifically representative sample of the internet at large, this > sample is a self selecting group of people who have chosen to run an > ipv6 test.
Commonly called sample bias. Good statistical analysis will address (and adjust for) such bias, but that can be (very) hard work. As with all the "CNN polls", there should be a disclaimer on such sites that say "this is not a scientific poll", but that would ruin the fun..... Gary
