Selection bias, Self selection bias, reward bias, bias in the questions and how they are presented, etc. You have to really know some stuff to not have a fucked survey result and even then some of those biases will still be present. If they don’t even bother with error % boundaries it’s probably going to have some other problems too.