Actually anyone who had taken a statistics course would have predicted a Clinton victory in 2016. My use of statistics enabled me to very accurately predict the 2018 mid-term elections. More accurately than anyone else on this forum -- including you.
Take a course in Market Research 101 and get back to me. Yes, polls are based on the OPINIONS of those individuals polled. But the results are based on a random sample of people.
When the information from a poll is presented to the public, there is very little interpretation going on. A question is asked: "Which Democratic candidate do you prefer in the 2020 presidential election?". A list of all known or predicted candidates is given. "Other" is a valid answer. The results of the poll are printed. No interpretation needed.
The reason why there is a "statistical significance" factor and a "margin of error" has NOTHING to do with the interpretation of the polling results.
"Most surveys report margin of error in a manner such as: “the results of this survey are accurate at the 95% confidence level plus or minus 3 percentage points.” That is the error that can result from the process of selecting the sample. It suggests what the upper and lower bounds of the results are."
https://ropercenter.cornell.edu/poll...g-fundamentals
Originally Posted by SpeedRacerXXX
There was no great magic in 2018. No special use of statistics. It played out almost exactly the way the polling data showed and stayed largely within the margins of error. I predicted exactly what happened as well, but certainly don't tout any special use of statistics.
House to Dems by no record setting margin. Senate remains fairly static with some Republican gains.
2016 was also within the margin of error for most of the polls out there too, just went to the underdog by the couple points. It was entirely pundits that predicted Clinton's massive win, not so much the pollsters. And the pundits got left with egg on their faces.
But to the question of polling questions, there is an art to the writing of questions and then advertising the answers. While you choose a very straightforward one as an example, a better one is.
Question : Do you support universal healthcare?
Answer : Yes - 70% plus now support it.
Question : Do you support universal healthcare if it means giving up your current plan and paying a larger share?
Answer : No - Only 13% or so support that.
Easy to play games with polls just by questions alone. And in the political world it also easy to manipulate samples as not really being random.