This is one of good example that polls is not reliable source to see if the candidate would secure a win. Since it didn't cover up the whole number of voters so there's really changes that happens when election happen and we see that happen before.
This happens in general with statistics. When based on a sample and making an extrapolation there is always a margin of error, which does not have to be the one calculated by the same statistics.And I am talking about those who want to do it in a neutral, honest way. If you want to elaborate statistics to influence public opinion in one way or another, it can be done. In this case, it serves to keep us entertained and to speculate on possibilities until we have the result of the elections, but I am not sure of anything, especially when the key is in a few tens of thousands of swing state voters.
