Well, in general, the presence of bias in voluntary surveys is well known and well documented. I think any statistics society anywhere would support that, including the American Statistical Association and the Royal Statistical Society. The real question is how much bias is present in each of the responses. It can be quite pernicious in some cases; it can be less severe in others. So the correct answer to the question is that it's a complete unknown.
On July 27th, 2010. See this statement in context.