Quote:
Originally Posted by PsYcNeT
Why is the general public so bad at understanding polling? Math literacy is just...awful.
|
Well, I don't think the statisticians have helped with this. I heard Nate Silver on CBC leading up to this, and he kept repeating this BS that "a 90% chance of a Biden win means if you did the election 10 times, he would win 9 times out of 10."
That's not correct. Even if there is a multi-verse, we only live in one dimension. There is only one election result. The population for that event is the people who actually voted. When polling, you are taking a sample of the voting population, and that results in sample error. The 9/10 for Biden actually means there's a 10% chance that your sampling didn't capture the true population.
Now in this case, where in back-to-back Presidential elections the sampling error reared its ugly head, you have to ask if maybe they are systematically under-estimating their sampling error.
My guess is that they are. And if they're under-estimating their sampling error, I don't think it's wrong to say they are wrong.