Survey is the better way but not the best. Sometimes the participants of any survey dont represent the whole of people or society, although they have chosen by a random way. The best way imo is counting the participant of any bitcoin forum like this forum, any exchangers site, gambling site, etc
Maybe but traffic counts can be deceiving too. How many trolls are here that don't use Bitcoin? How many Something Awful goons? How many law enforcement officers? How many people don't give two shits about Bitcoin but came here because they heard it's great for signature campaigns and the minute that money dries up their gone? You're going to scoop up a lot of people that you don't want with traffic counts.
Edit: Check out this thread -
https://bitcointalk.org/index.php?topic=1139726.0Further to this, you shoudl check out the thread asking people who has 1btc or more. Im gobsmacked that 90% of the people there dont own a whole coin. If you made assumptions that 50% of the accounts on ths forum don't own a whole btc then you can see how any sorts of surveys, traffic counts etc can be deceiving.