@Alodar: I don't see how your quote changes negates the content of my post. I specifically mention the issues with (lack of) representativeness due to a small sample size and the fact that we're selecting from forum users only.
You can always lower sample size if you're willing to accept a smaller confidence level or larger margin of error. The numbers you quote seem to be for 95% confidence interval and a 5% margin-of-error.
Also, polls are regularly collected that use ratios MUCH smaller than 10%. US election polls don't survey 15 million people after all. Other sources I'm seeing suggest that "A good maximum sample size is usually 10% as long as it does not exceed 1000."
