The margins are everything. A poll that shows it's +1 Trump instead of +1 Biden is more accurate than a poll that shows +8 Biden.
They showed him like +8% nationally. It's going to be like +3 or +4. Think about the millions of dollars they spend on this. They're pretty bad at sampling.
Might be asking to much for perfect margins.Predicating the overall winner is good enough for me. National polls are externally accurate in predicting the winner and approval ratings are undefeated in picking a winner.
You'd be happy if the polls showed a nail biter and the margin was +30? This election proves how unreliable they are unless they're calling an absolute landslide.
It’s important to remember is that a poll that is within the margin of error is actually a good poll. States with major errors this cycle are basically all of the important ones: Florida Wisconsin Michigan Pennsylvania Nevada To be fair to aggregators, they certainly analyzed the election with a major polling error as a possibility. My take is basically the same as I had for a while. The country is sharply divided and with a high turnout there was a close election. So, the pollsters may have gotten a little too creative with their electorate models when all they basically needed was to model based on registration.
Im happy if they tell me who the winner will be.Approval ratings polls have predicted that since the 1940's and aggregate national polls did it in 4 of the last 5 elections and was only off by 78,000 votes in the one EC loss.
Two days before the election -- many individual polls were showing a blue wave. In the actual election the results were outside the margin of error in the majority of these individual polls in nearly all the states -- especially swing states. Why?
Polling clearly isn't perfect,but national polling its nearly perfect in picking the overall presidential race winner.