Ramsay Hunt Experience
Footballguy
Hey, we're making progress! You're actually stumbling upon legitimate criticisms of some polls.then Hillary Clinton won the 2016 election
everyone had her penciled in, she was up double digits in many polls
they were wrong
If I wanted a 65-70% of the people don't mind confederate statues poll results, I'd do that poll in an area largely southern and in a city that has statues and parks of the Confederacy. I'd get my poll results one way or the other.
If i wanted them to be AGAINST the statues, I'd do that poll in a very liberal black neighborhood in California .... I'd get my poll results one way or the other.
I'd also word the polls differently, to get the results I wanted.
I obliquely mentioned that the 2016 presidential polls were not as inaccurate as people think. Most polls had her winning the popular election by about 3%. She won by 2%. Well within the respective margins of error. But horse race national election polls have one big problem. They aren't measuring the right thing. Because winning the popular vote doesn't make you president.
So we look to the state polling. Even though Trump pretty much swept the toss-up states and Hillary was "ahead" in the latest polls in many of those states, almost all of those polls were well within the margin of error as well. Because Hillary didn't have big leads in those polls and Trump won those states by a whisker (Wisconsin is an exception, but we didn't have that recent a poll from before the election).
Your later discussion highlights the legitimate problems with some polls that we're already talking about. If you pulled 65% of your respondents from the South, you wouldn't have a representative sample. You could have polled 2 million people and that poll would be worse than one that polled 1,000 people but which was representative.