Let me start off by stating that I am looking for a factual answer not debate or opinion, I firmly believe that there must be one, and that is why I put this in this forum instead of IMHO or Elections or where ever.
I’ve seen here and there the statement that the polls preceding the presidential election were off, in part, because of “voter embarrassment” on the part of some of those who voted for Trump. Where did this come from, how did the idea that voters were embarrassed to be voting for Trump and so lied to the pollsters as opposed to the thought that maybe they just figured it was none of anyone’s business instead?
Not a political scientist, but two variations of this (alleged) phenomenon that have been talked about for decades are the “Shy Tory Factor” and the “Bradley Effect.”
Sure, a voter can think that it’s none of anyone else’s business how they plan to vote. But the difficulty with the “shy Trump voters” is that one of their biggest complaints, and a major reason they voted Trump, was the belief that everyone was ignoring their problems. Well, nobody’s going to pay attention to your problems if you’re not willing to tell us what they are.
The other issue is-why privilege this particular group of voters? There could be plenty of other voters in other demographics who voted differently than they polled, such as wives of Trump supporters who didn’t want their spouses knowing that they were going for Hillary.
Well, we have empirical evidence, now. We know that the polls favored Clinton, and that the election went to Trump. “Shy Hillary voters” would not account for that (they’d make the discrepancy even worse), but “shy Trump voters” just might.
Or, if you meant before the election, people were talking about both possibilities, and nobody really knew which one was correct, or to what degree.
I’m ignorant of the poll processes (aside from exit polling); but 1) since there were 2.5 million more votes for Clinton is it possible the people polled were telling the truth but were NOT voting in a state w/ more electoral votes; 2) if people are shy to say who they’re voting for, maybe they’re also shy to say WHERE they’re voting, since location’s self-reported when a poll’s being done via call to a cell phones?
There were (more or less) leftist riots at some of the Trump rallies during the primaries, folks getting beat up, and a 24/7 media onslaught claiming that only xenophobic racist sexist homophobic knuckle dragging cro-magnon slugs in hick flyover Jesusland were voting for Trump, so it wasn’t surprising to me at all.
There is a big difference in recent years with respect to yard signs and bumper stickers. In populated areas nobody dares put a conservative bumper sticker on their car, they are worried about vandalism and other nutters.
As a political agnostic, it was obvious to me going back decades that a candidate that told the media to get bent would do very, very well, the wonder is it took this long. Once the voters saw this, they had their guy and didn’t give a fuck what the pollsters said or pushed, the “news” etc. A huge swath sees the democrats and republicans as a “uniparty”; Trump had to beat the Democrats, sure, but first he had to beat the Republicans and the media outlets. An amazing accomplishment every step of the way.
Your factual response to the OP of this General Question would be which part of this?
PS, I live in heavily populated Northern Utah and there were plenty of Trump/Carson/McMullin/Cruz stickers on cars starting about a year ago and I heard no reports of cars bearing those stickers in Utah being vandalized.
The way polling works is you try to get a representative sample by calling around the country so this would not rely on people’s own report of their location.
What I believe happened was that since the fewer people you have for a poll the greater the margin of error, the polls were correct for the country but wrong for the states. To take an extreme hypothetical. If you polled twenty people in each state you would have 1000 people total. The margin of error in the poll for the country would be large but the poll would be usable. However, the margin of error for each state would be so large that the poll would not be usable. Most polls are national polls, most of the state by state polls were done by the campaigns and not shared with the press.
That would make sense if the polls were national. I know from reading the Votemaster every morning that all the polls (all the ones he used anyway) were by states, often carried out by a local university (poly sci dept., presumably). Pennsylvania, for example, consistently showed at least a 5% margin for Clinton, which obviously didn’t materialize.
It is also possible that things changed dramatically in the final week on account of Comey, and the polls never caught up. We will never know.
Emphasis added. I’m not a political scientist, but I used to live with one - the first 18 years of my life - and one thing I learned from him is that **social science is so-so science. **It’s not all bullshit, but the vagaries of human behavior (and, when talking about elections, the problem of being unable to replicate tests) means that conclusions should rightly be tentative.
And conditions change much more rapidly than social science research can! To some degree, I believe political forecasting will always be fighting the last election - until we all have implanted politics-meters beaming our RD scores directly to Nate Silver’s laptop or whatever.