Samples of thousands can significantly reflect the billions on earth. Samples have nothing to do with their relative size in comparison to the population. Even if you're just taking a convenience sample, which is inherently biased, it can be statistically significant as long as it is still random. Sure, people who come to the forum don't represent the entire population of gamers (which includes people who don't go to the forums), but when you have 90% of the 11,000 people who took that poll agreeing that they want more than one raid, it's saying something. It's like asking a group of 11,000 men if they enjoy summer or winter more. if 90% of them say summer, sure, you're limited by the fact that you're missing the female vote, but does that make the poll useless? No. Otherwise, we would never have any data on anything that goes on in the world or a nation, because we can't poll hundreds of millions of people. We poll samples. Samples can be biased, but biases don't determine significance.