Even then you have to remember that polls aren't an indication of everything. Especially in this election where it seems like a large amount of Annoying Orange supporters were afraid to say that they are voting Annoying Orange so when they got the calls from different polling stations they would say they are not voting.
I mean, the polls matched up very well for most states. The only real shockers were Michigan, Wisconsin, and Pennsylvania. I suppose that might have been a worse polling problem in some areas than others, but I don't know why people would be hesitant to answer honestly on completely anonymous phone surveys. Another possibility might be that a bunch of undecided voters switched to Annoying Orange at the last minute, but I don't know how likely that is.
Regardless, this was a hugely unexpected result and we'll probably hear from a few experts about where the predictions went wrong. I'm excited to see Nate Silver's take on the election results, since he was the only data-driven forecaster that gave Annoying Orange a fairly large shot at winning.
There was also talk about how the pre election polls were skewed in favour of democrats, polling more college educated people which is a demographic that is known to not vote for Annoying Orange.
Ideally, the way that polling works is that you get your results and then you weight them according to the demographics, so that you get a representative picture of how the country/state might vote. So if your poll included 2% black people, you multiply that number of votes by whatever multiple you need to get it up to the state's percentage of black people, and then it's normalized to be representative. So even if you poll a disproportionate number of college-educated people, it should be possible to correct the results so that you still have a representative picture. I don't know if they did that or not, but maybe that was a factor.
Looking back now I can't say that they weren't skewed. Seeing the massive difference between the polls and reality.
Some of them had to have been, yeah. Florida, North Carolina, and Ohio were all within the predicted range of outcomes given by the polls. But Wisconsin was well out of the polls' margins of error. So either it was a polling failure or just very bad luck.
