You’re right it’s a necessary assumption.
And since it’s actively ridiculous that hard lefties are more likely to be shooting or vandalizing hard righties than vice versa, it’s obvious that the result is false because one of the premises is false.
There was some evidence of a shy-Trump voter in 2016, in that telephone polls consistently gave Trump a slightly lower result than robo-polls. The theory was that people were less willing to admit that they supported Trump when talking to a real-live pollster, but were willing to admit it in an automated poll. The difference was about 1%, but in certain states…
By contrast, there doesn’t seem to be that difference this time, as mentioned by electoral-vote.com in this post from last week:
Hahahahaha, I love this. Look out! The liberals are gonna throw free healthcare at you! AHHHHH!
2019 was an outlier precisely because it wasn’t really a general election. It was a general election in name but in practice a once in a generation (minimum) “general election” based around a single topic that was tearing the country apart: Brexit. Many, many people voted for their not usual party (on both sides, many traditional left wingers voting Tory as they were pro Brexit and a lot of tactical voting on the anti Brexit side to push that agenda) because Brexit was what they really cared about. A single issue is a lot easier to gauge. Not that that helped much with the recent referendums … (I’d argue that the years of arguing about Brexit had polarised so many people and so many had become much more vocal that it was easier to gauge).
The 2017 election pretty much was wrong until the last few days. 2015 was famously wildly out. The Brexit and Scottish Independence referendum polls were quite wrong. You can take this back quite far to the 1992 General Election where everyone predicted either a Labour victory or a hung parliament. Instead the Tories won with 65 more seats than Labour.
That was true in 2016 (replace Clinton with Biden), but less so, probably a lot less, in 2020.
What sort of risk? How do you quantify that there is less risk in that scenario?
I could be persuaded that some are afraid of reputational risk… they know they’re making a garbage choice and want to hide their shame. I could likewise be persuaded that some people are just obstreperous and want to tweak the media because fuck the media. (watch one of these rubes with his media degree from Fox News University try to “spar” with journalists, it’s hilarious and sad).
But… actual, physical risk? We’re told endlessly (and gleefully) that liberals will definitely lose the coming civil war because the conservative side has all the gun-toting ex-military ex-police he-man badasses. Either that story is wrong, or there’s no real risk to conservatives. Both can’t be true.
Yes, the article was about Robert Cahaly and his company, the Trafalgar Group. But I chose to keep it vague because I wanted to discuss the general issue of whether pollsters can objectively identify shy voters rather than have this thread be a discussion of Cahaly.
Assuming we read the same article, I feel Cahaly didn’t really make his case. He just said other polling companies are wrong and he’s right without offering any evidence to support this beyond the elections in which he made correct predictions. But the reporter brought up examples of elections in which Cahaly’s predictions were wrong.
Cahaly did mention a proprietary model at one point, so arguably he has a method which he is choosing to keep to himself so other companies can’t duplicate it.
But I feel it’s also possible that Cahaly is just going with his gut. It’s noteworthy that Cahaly, a Republican, always predicts the Republicans will get more votes than other polling companies are predicting. Cahaly may be projecting his beliefs of what should happen on to his predictions of what will happen.
He did produce a good prediction for the outcome of the 2016 presidential election. But hundreds of pollsters predicted the outcome; some of them were bound to be right even if they were just guessing at random. I feel it remains to be seen if Cahaly has a method for predicting elections outcomes consistently.
This is the kind of thing I’m wondering. How do we know there is a significant amount of shy voters planning on voting for one candidate but not a significant amount for the other? Is there some factor we can look at that let’s us be confident in predicting a candidate has shy voter support?
The logic of a shy 2020 Trump voter escapes me. If he’s so darned great why wouldn’t a Trump supporter proudly proclaim that?
On the one hand, the proud in-your-face MAGA redneck driving a pickup truck with six Trump flags might well lie to a pollster, because fuck pollsters.
On the other hand there were surely many Trump voters in 2016 who weren’t so dumb as to think he’s so darned great, but voted for him anyway, and weren’t proud their vote.
What about voters who strategically lie to pollsters? I’ve never been contacted by a pollster, but I’ve often thought that if I was, maybe I should (falsely) tell them I am voting for Trump. Nothing depresses voter turnout like overconfidence. I think if the polls had been closer in 2016, Clinton probably would have won.
That’s like asking why a Neo-Nazi in America wouldn’t want to go around proudly proclaiming his support for Hitler. He may think Hitler is great, but he’s perfectly aware that the majority of society, the media, academia, and others, don’t.
Or like asking why a lesbian in the 1950s wouldn’t want to be proudly out-and-open lesbian. She may think homosexuality is wonderful but knows perfectly well most of society doesn’t.
(Not that I’m comparing Trump to Hitler or LGBT, but you get my point)
In Taiwan, that’s exactly what the Trumpish candidate, Han Kuo-Yu, asked his voters to do in the presidential election - lie to pollsters so as to screw all polling up. Considering that he eventually lost in a landslide, it’s hard to conclude that it helped him.
I think it can go the opposite way. If people think a candidate is winning, they want to join the bandwagon. And if they think he’s losing, they abandon him as a lost cause. Sometimes faking momentum in a political campaign can create real momentum.
"11.7% of Republicans say they would not report their true opinions about their preferred presidential candidate on telephone polls.
In contrast, just 5.4% of Democrats say they’d be reluctant to share their true voting intentions — roughly half the number of Republicans reluctant to tell the truth on phone polls.
10.5% of Independents fell into the “shy voter” category"
If you’re a “shy” voter, why go to the trouble of lying? Wouldn’t it be easier to just say “no” and hang up?
Which leads to a bizarre conclusion. If the number of declared Republicans who are lying is twice the amount as the number of declared Democrats who are lying, it’s a major plus for Biden. After all if a person declares their allegiance to one party in a poll and says they don’t give true information in polls, than they must really be supporting the other party. So 11.7% of the people who said they are Republicans are actually planning on voting Democratic while 5.4% of the people who said they are Democrats are actually planning on voting Republican.
Where I live the county is clearly D majority.
The election department publishes the counts of registered voter, of mailed out ballots, of returned mail-in ballots, and of in-person voting by declared party affiliation. Declaring affiliation is optional and about 20% of voters choose not to. All the evidence is clear that at least here Trump will take a shelllacking unless vast numbers of registered Ds vote for Trump compared to the other way around: registered Rs voting for Biden.
See https://www.browardsoe.org/ for lots of actual factual numbers.
Conversely …
Anecdotally in my demographically typical part of that county …
Around here Trump yard signs outnumber Biden signs about 10 to 1. Vehicles flying Trump flags are maybe 5% of all vehicles, while vehicles with Trump stickers without flags are another 5%. That’s 10% of the cars/pickups on the road. I doubt I’ve seen 2 Biden bumper stickers & I know I’ve seen zero Biden flags on cars since Biden became a shoe-in (shoo-in?) for the nomination.
Whatever else you can say about the evident minority of Trump supporters here, you can’t say they’re hiding their enthusiasm.
Or more precisely I guess I should say that the Biden supporters, despite being in the obvious majority, are far better at hiding their support than are the far less numerous Trump supporters.
The USA is a big and diverse place. Perhaps where you are the majority Trumpers are hiding from the disapproving tutt-tutts of the local demonstrative Biden-ish minority. But it sure ain’t true here.
But again this goes back to my original question. Is there an objective way of distinguishing, before the election, which people are shy Neo-Nazis who are laying low to avoid the disapproval of the anti-Nazi majority and which people are genuine members of that anti-Nazi majority?
I have no data or cites, but my gut feeling is that this is true.
A Biden supporter in deep redneck country probably wouldn’t lie to a pollster. Sure, the Biden voter knows that he might get some ugly comments at work from the boys on the shop floor or the truck drivers he works with.But he won’t get fired. And he isn’t afraid of being outed anyway, because those co-workers aren’t going to discover what the Biden voter said in an anonymous phone call to a pollster.
Now compare this situation to:
Because, unlike the rural Biden voter I described above, the danger is real for a Trump voter in a liberal area. Suppose you’re a white-collar professional working for a large corporation in Manhattan or San Francisco: the kind of person who keeps his linked-in profile updated, and finds jobs using headhunter agencies. You’d be wise to keep your Trumpian beliefs secret. You could get fired, and your career could be ruined forever.
Your employer has forced you to sit through many training seminars on racism and white privilege, and you know that there is zero tolerance for straying from politically-correct company policy. Remember the kerfuffle a few weeks ago when the Goodyear corporation officially declared that employees may wear BLM logos, but not MAGA hats to a company event?
Your reply to a pollster is supposed to be anonymous, but it will be recorded permanently in their data base, (possibly useful to them for future follow-up polls)–so now your phone number is defined as a Trump supporter. That data base is secret, and unlikely to be leaked.But , unlike the guy working on a factory floor I mentioned above, you didn’t get your job by standing in line and talking to the shift foreman. Your next job search will involve a fairly deep investigation into your personal history by a huge HR dept. That investigation will be thorough,and computerized. It is possible, (though unlikely) that they will stumble onto the pollster’s data base.
You supported Trump? You’re a deplorable racist.
You are blacklisted, and your career is ruined.
How many such “shy Trump voters” are there? Probably not many.
But it could be one half of one percent, maybe even one full percentage point. And since the polls massage the numbers by weighting for everything else except shy voters, their algorithms are probably wrong by more than the 1 percent which I mentioned.
…
Summary: my gut feeling is that shy voters exist, and the pollsters are unaware of them, so their formulas for weighting the data are wrong.
But if somebody has better info, I’d love to let your facts beat my guts. 