05-31-2020, 07:14 PM
(05-31-2020, 06:10 PM)HarleyDog Wrote: Ok, so I followed your advice about the data being available and googled. I seen this, read it, and I guess I may be off by 1% on the high side and even worse on the low side. However, I think it says what I was trying to say. I don't know the website, nor do I know if it leans right or left. I just clicked on it because the title and here it is.
https://www.pewresearch.org/fact-tank/2016/11/09/why-2016-election-polls-missed-their-mark/
Pew notes that they ranged from 1-7%, which is true. What would be incorrect is saying that they were between 5-8%.
You can go to RCP and look at the final average, which was 3%. That takes into account all of the available polling. Of the last 20 polls, 6 were within the 5-7% range, which means 70% of the polls ranged from -2% to +4%, with the final actual percent being +2%. The majority of the polls were not in the range you stated.
Nate Silver, the biggest figure in modern statistics, breaks it down:
https://fivethirtyeight.com/features/the-polls-are-all-right/
What makes people believe the polls were inaccurate is the fact that the election was decided by 4 states with a total of 114k votes between the 4. When you have state polls giving a candidate a 1-2% edge and they lose by 0.1% to 1.2%, that falls within the margin of error. It indicates that the poll was accurate, but anyone interpreting it as guaranteeing a win would have been wrong. As Silver breaks down, the 2016 polls were historically accurate.