I think the point PhantOS is making isn't that a sample of 63,000 is statistically insufficient, but rather the fact that people are using these surveys to treat other people badly. Even though they're using the views of one person to judge like 28,571 other people. That is a stuffty thing to do from a human-morality standpoint. If a similar study showed something bad about conservatives, I wouldn't be using that to stuff on my conservative friends.