The size of the sample doesn't matter, it's how inclusive and random it is. It's how its conducted, not how many people are in it. Though, it is true that the bigger the sample the more accurate.