(10-06-2020, 02:27 PM)p_rushing Wrote:

(10-06-2020, 10:17 AM)The Real Marty Wrote: No one is paying the pollsters to make fake polls.   Their credibility is all they have.  They're not going to publish a fake or incorrect poll on purpose.  

But that leaves the question, can the polls be wrong?   I think the pollsters are doing their very best to get it right this time.  

A good poll has to be constructed pretty carefully.  They have to get the correct sampling of democrats, republicans, old people, young people, rich people, poor people, men, women, different races, college degree or high school degree, cell phones, land lines, and so on.  They have to try and predict who is going to vote.   It's not easy.   Just getting people to answer the phone is getting harder and harder, with cell phones screening out people not on a person's contact list.  Last time, they didn't realize that there would be a split based on educational levels, with voters without a college education going heavily for Trump.

Polls are paid for my the parties and others. They may try to sample correctly, but generally it is harder to sample conservatives because they don't usually talk to unknown people on the phone or like to give their voting info out. When you already oversample, then add that some people don't want to announce they are voting for Trump, it skews the numbers begin time.

Of course, that is the big question: are there hidden Trump voters because for some reason, Trump voters don't like to answer polls and Trump haters like to answer polls?   It's a tricky question.  People with cell phones can screen calls.   I personally don't answer a call if my phone doesn't name the person from my contact list.  So I have probably screened out a pollster or two.  But do Trump voters do that more than Biden voters?  What is the evidence of that?   In 2016, the polls right before the election showed about a 3% lead for Clinton, nationwide.  Clinton wound up winning the popular vote by about 2%.  So either the polls were off by 1%, or there was a late shift towards Trump that was too late for the polls to pick up.  But it was just 1%.  Trump won the election by very narrowly winning a few swing states where the polling was off for some reason.  

Again, the last pre-election polls in 2016 had Clinton winning the national popular vote by 3%.  Right now, the average of all the polls shows Biden with a lead of between 7% and 9%.  And some of those polls were taken before it was revealed that Trump doesn't pay income taxes, and before he had the disastrous debate, and before he was diagnosed with Covid-19.  So it's hard to see the polls swinging back towards him at this point.  Another point to keep in mind is that, compared to 2016, there are very few undecided voters.  

Again, the big question is, are Trump voters refusing the answer the phone or refusing to answer polls more than Biden voters are, and if so, is it enough to make the polls not just slightly wrong, but very wrong?  I'm sure a lot of Biden supporters are sweating out that question.  But I'd sure rather be Biden than Trump at this point.

---