538 "Could Social Alienation Among Some Trump Supporters Help Explain Why Polls Underestimated Trump

[NRaleighLiberal]

(posting with no personal comment - just something interesting to read as pollsters try to figure out why they sucked again- whoops - that was a personal comment!)

https://fivethirtyeight.com/features/could-social-alienation-among-some-trump-supporters-help-explain-why-polls-underestimated-trump-again/

Donald Trump was once again underestimated by the polls.

On the one hand, this polling error is fairly normal. We’re not talking about huge polling misses, and the polls still “called” the election correctly: Joe Biden won. But that said, it does seem as if polls are still failing to capture some of Trump’s support.

There are a number of possible explanations for this, and no definitive answers, but one thing I’ve come across in my public opinion research is that the share of Americans who are more socially disconnected from society is on the rise. And these voters disproportionately support Trump.

The idea that some of Trump’s supporters are more likely to be disconnected from civic life is hardly a new one. During the 2016 Republican primary, Yoni Appelbaum at The Atlantic noted that Trump was drawing support disproportionately from those who said they were civically disengaged. An analysis by Emily Ekins of the conservative-leaning Cato Institute found that despite Trump’s continued strong support among white evangelical Protestants, he was actually viewed more positively by supporters who weren’t involved in regular religious practice. Finally, research on the 2016 election by David Shor, a Democratic pollster, echoed what we found in our own pre-election 2020 survey: There was a large swing to Trump among white voters who had low levels of social trust — a group that researchers have found is also less likely to participate in telephone surveys.

In our pre-election survey on the strength of Americans’ social networks, we found that nearly one in five Americans (17 percent) reported having no one they were close with, marking a 9 percentage point increase from 2013.1 What’s more, we found that these socially disconnected voters were far more likely to view Trump positively and support his reelection than those with more robust personal networks. Biden was heavily favored by registered voters with larger social networks (53 percent to 37 percent), but it was Trump who had the edge among voters without any close social contacts (45 percent to 39 percent).

And this was especially true among white voters even after accounting for differences in income, education level, and racial attitudes. Sixty percent of white voters without anyone in their immediate social network favored Trump, compared to less than half (46 percent) of white voters with more robust social ties.



snip

last para

A lot of time will be spent over the next few years trying to explain the country’s growing social, economic and political problems, but we should not forget that political reforms and economic fixes are not going to address more fundamental problems of loneliness and social isolation in this country. And that may mean we increasingly have less insight into how a growing portion of the country feels.

[TexasBushwhacker]

1. Interesting n/t

[Ninga]

2. These "socially and civically" disengaged and ignorant

people need to learn that it wasn’t George Washington
that got them their Social Security and Medicare and eight hour workday and clean water and safe food and medicine and more, it was the Democrats!

[jimfields33]

3. It's funny. They will throw a lot of reasons why polls were off and see if something sticks.

I think the main reason is nobody answers their phones unless they know the person. That’s really the main reason.

[Silent3]

6. "nobody answers their phones" does NOT explain the difference unless...

...there's a political correlation with the likelihood of responding to a poll, and being truthful if you do respond. That political correlation is what this article is talking about.

If not wanting to answer your phone were a politically neutral behavior, it would have no effect on polling results at all.

[jimfields33]

7. That's interesting. Thank you for providing that explanation.

[Captain Zero]

22. The AP poll was conducted with a postcard contact first, not a phone call.

to the address of a registered voter. I know because we got one.
A big postcard addressed to any registered voter at the address.

I scratched off a pin number, then went online and entered the pin number to participate.
Another way to participate was to call an 800 number, use the pin, and take the survey over the phone.

So if people were too socially disengaged to do that, or too HOSTILE to pollsters to do that, then that could skew the results.
Socially engaged people will take the initiative to follow up and be surveyed, others, maybe not so much.

How many people let the obvious advertising mail like 6x9 postcards pile up somewhere too?
I do at times.

[jimfields33]

28. That's actually not a bad way to do a survey.

[Drunken Irishman]

37. I think it's as simple as typical polling errors.

In 1996, polls had Bill Clinton winning reelection by a landslide. The final CNN/USA Today tracking poll had Clinton beating Dole 51-35. THAT was a monster polling error. Read this article and tell me if it couldn't have been lifted from 2020:

https://academic.oup.com/poq/article-abstract/62/2/230/1891100?redirectedFrom=PDF

Polls were also wrong in 1980. They had Carter and Reagan in a dead heat but Reagan won by a large margin.

Polls were off in 1980, 1996, some were off in 2000, 2012, 2016 and 2020.

The big difference with polls in 1980, 1996 and 2012, compared to 2020, was that the polling shift didn't change the expectations too much. Most thought Reagan would win - but not by the margin he did. Most thought Clinton would sail to reelection, but were surprised the polls had his margin much larger than it was (still, he won comfortably), and most thought Obama was going to win reelection but didn't fully expect it to be as easy as it turned out to be.

The difference with 2020? The way we counted the ballots.

Had it been a normal election, like 2016, Biden is declared the winner on election night and wins fairly handily across the board. So, the polling error doesn't seem as big of an issue. But because PA, WI and MI all forced a mirage come from behind win for Biden, it built this narrative the race was close.

it wasn't really close.

That's my take, anyway!

[RazzleCat]

4. Interesting, but I think all polls are "off"

because we no longer answer our phones. Think about it, polls are done via phone, when was the last time you answered you phone when it was "unknown caller"? I will posit that unless you were job hunting, or were expecting a call from an agency, persons etc that you don't know (say a doctor, or an appointment conformation), you never picked up your phone to any person conducting a poll.

[NRaleighLiberal]

5. Agreed - 538 did a podcast on this recently, noting the horrendous percentage of poll respondants vs

those contacted. To me, it makes the outcomes pretty useless.

And, sitting here at 64, my wife and I both noted neither one of us has ever been polled!

[Silent3]

8. Again, as I just said to some else, not answering the phone doesn't explain this...

...if not answering the phone is a politically neutral behavior. If lots of people won't answer, and that's all there is to it, you just make a lot more calls until you reach your desired sample size. Problem solved.

The whole point of the article quoted in the OP is that lack of response (and also possibly dishonesty in response) probably isn't politically neutral. Pollsters could adapt to this by adjusting their polling models, but the skewed lack of response which is the problem that needs to be understood is difficult to measure by its own nature.

[radius777]

41. If a very small group (5%) of people participate in polls,

then that group is atypical in at least that behavior, and likely to be atypical (and unrepresentative) politically as well.

[dalton99a]

9. +1

[uponit7771]

31. They're not off in other countries with more cell phone usage, no US POLLING SUCKS !!!

[Roisin Ni Fiachra]

10. This is an unsolvable problem. Too many unchangeable negative factors

contribute to this unfortunate phenomenon in the US.

"The most dangerous creation of any society is the person who has nothing to lose." - James Baldwin

[brooklynite]

11. If Trump kept telling you that CNN and the NY Times were "fake news"...

...why would you participate in their surveys?

[Mariana]

13. I think a lot of them participate in the surveys and lie. nt.

[brooklynite]

18. That would show in the raw data...

It's actually difficult to lie convincingly over time. It requires more than just saying "I'm voting for Biden"; you have to answer questions on personal opinion of candidates, office holders not up for re-election, past voting history, policy positions etc.

[Klaralven]

12. Anomie - caused by agricultural industrialization and the collapse of manufacturing

[TheBlackAdder]

14. They are very alienated. That's why they are living in a bubble.

.

.

[LonePirate]

15. Pollsters simply underestimated how likely the deplorables were to vote.

They had a sky high turnout rate, especially in rural and red areas. It explains how several Dem House members lost and how some Dem Senate candidates lost, not to mention the state legislature seat losses.

However, 2024 will likely return to more usual levels of turnout (like 2012 rates) without 45 on the ticket. The real question will be how 2022 turns out. Will it be more like 2006 or 2018 or more like 2014? Good Dem results in 2022 will foretell good results in 2024.

[gulliver]

16. I'm not sure why anyone would think random polls should even work today.

The sources of the data consume the data. The sources have a significant understanding of how their responses will play out in the media. Therefore, the answers are not random and independent. They are strategic. And the poll results feed back to the system, changing the strategies of the poll sources.

People can and do deliberately distort polls with their answers. Reporting on polls is abysmal, because it naively takes the validity of "scientific" polls at face value. Therefore, the polls just become another football in the information/misinformation football game.

It seems to me that real polling is going to require dedicated, honest, incentivized data sources. Maybe that will work. You agree to be polled on your cell. You agree to be honest. The pollster doesn't necessarily know your identity. You get money in your PayPal when you respond. You lose your spot in the polling lineup if statistics about your response record show you lie and play games.

[bullwinkle428]

17. I'm an introvert by nature, and the pandemic has increased the measure that

I've felt socially alienated to a greater degree than I've ever been in my life. Having said that, my disgust for that creature has only grown since the beginning of "abnormal life".

[Tribetime]

19. I still think voter suppression had a lot to do with it

My mail and had to vote in person and a lot of democratic areas that were more populated didn't have enough polling places for early voting. I know in Cleveland Cuyahoga County that Biden underperformed Hillary which is really hard for me to believe.

[BadDog40]

20. The polls are always off due to voter suppression

It's as simple as that.

[BannonsLiver]

27. It's actually not that simple, though.

That explains some of it. Maybe. But it’s well beyond the time that we all accept the fact there are millions of Americans who are rotten to their core and who have no redeeming qualities whatsoever. Totally unreachable and too far gone even if you could. They’re out there, just like the article says.

[triron]

21. The premise may be false. So any conclusion is invalidated.

Simple logic.

[crickets]

26. +1

[fearnobush]

23. Ths sounds like more excuse reaching. Simple answer is we don't know why yet.

And may never know. For example why did Obama outperform all the polls in 2012 by a similar margin? Why were 2018 midterms so accurate. Polls are just that polls and they will be off with in the margins of error.

That being said my own idea is that in 2020 With a pandemic and not a whole lot to be distracted by, voting became a ‘pandemic’ activity thus many more low propensity voters came out and voted possibly muddying up the expected outcome.

Joe and Kamala will won by 4.3 - 4.7 % nationwide. That’s well with in the expected margins with in their polling average of around 8%.

Dems won at least one Senate seat but lost at least 10 in the house.
This seems odd but not really. The wins in the 2018 midterms were at about the most they could actually win given the decades of puke gerrymandered House districts. So it kind of makes sense to lose races at the tightest of margins in 2020 with PresidentIal candidates on the tickets
.

[fearnobush]

25. Also I don't think defund had as big a roll as the M$M plays that out to be.

Thought it likely had some effect in some lost house races. I think this was an election where the electorate hated Trump but chose divided government due to over all distrust in government.

[Yavin4]

24. Trump's team knew about the Social Alienation variable

Late stage capitalism has increased social alienation and isolation. Our American culture is built around work. Work defines you and gives you an identity. Today, work does that less and less which leads to feelings of alienation. It's hard to find pride in yourself when you're an Uber driver.

[Dawson Leery]

29. Trump made his way in live appealing to those who believe they were given a bad hand in life.

It's the moniker of demagogues to place fault with some evil cabal that cannot be clearly defined.

[Midnight Writer]

30. Hate Radio has been telling listeners to lie to pollsters, over and over every day.

We underestimate the impact of Hate Radio in rural areas.

There are tens of millions of dedicated Hate Radio fans out there, who worship the words of cranks like Limbaugh, Hannity, etc.

[uponit7771]

32. **BULL FUCKIN SHIT!** NO Other Country Has Polling This Bad !!! Down Ballot Polling was HORRIBLE ...

Last edited Wed Nov 25, 2020, 05:42 PM - Edit history (1)

... and other countries have cell phone usage in higher percentages than Americans and polling is more accurate so its not that either.

Either our polling is off or our voting is off PERIOD!!!

Silver is trying to save his job with baseless guessing vs going after the polling companies for explanations on how they got down ballots so fucked up.

No ... fuck no, there's something wrong with something in the US in regards to our elections figures and the shit didn't start to get highly irregular until after 2004 ish.

GET THE POLLING COMPANIES IN THE HOT SEAT and stop the baseless guessing

[mathematic]

33. This is literally an explanation for how polling companies "fucked up"

Somehow you managed to turn an explanation for why the polls were wrong into a demand for explanations on why the polls were wrong.

Furthermore, this is not "baseless" guessing (nor was it even written by nate silver or 538 staff). At its core, this explanation is that poll non-responders are overwhelmingly supportive of trump. Accounting for this would represent a meaningful change in how polls are conducted. However there's no clear way to generally account for the political preferences of people that don't respond to polls.

[uponit7771]

34. No its not, no polling company gave this explanation to Silvers he's guessing. PERIOD

Also

So ignore the "other countries" part right?

All variables we're dealing with other countries are dealing with also and don't get it this off for this long tilted towards one party.

I literally made the point in the reply that the country that had the most prizes for sciences is most off in polling ... science.... for nearly a generation of humans and social alienation doesn't explain it either seeing it could've been weighted for PROPERLY from get.

None of what Silvers proffers explains how down ballot polling was so ... VERY VERY VERY ... off either all over the country.

The guess is baseless seeing we're the only developed country facing those variables on the planet with down ballot polling. Of course Silvers is going to focus on useless national polling vs the HORRID swing state polling that was exhibited.

"We're "that" unique" logic would be the SAME logic MAGAs use in regards to US COVID response and amount of cases too so the company this logic sits in is not good from the get.

No, its this countries polling and we're not bad country full of stupid liars either ... that's giving the polling companies a pass too.

[applegrove]

35. A campaign to lie continuously to MSM polling firms and say they were

voting for Biden instead of what they did which was vote for Trump, well that would mess up the polls twice as fast or take half as many liars.

[JI7]

36. Polls were not all off. Biden campaign manager said that things were close

in swing states. That they were not ahead by a lot.

[Drunken Irishman]

38. National polls have a history of being off.

In 1980, the national polls, at best, had Reagan slightly leading Carter. He won in a surprise landslide.

In 1996, the national polls had Clinton beating Dole 51-35 (the last CNN/USA Today tracking poll of that election). He won convincingly but nowhere near the size the polls suggested.

In 2012, the national polls had Obama basically tied with Romney and Obama won by a fairly comfortable four-points.

In 2016, the national polls were actually fairly correct.

It's time we realize there's a consistent polling error and while it doesn't show up in every election, it has proven to be there for previous elections.

[Buckeyeblue]

39. I'm still surprised at how wrong Florida was. And how Biden underperformed compared to Hillary

Has Florida become a state that can no longer be polled? Did slow mail (undelivered mail) cause that much of a difference? Or have Cuban-American's fallen off the deep end? Will Cuban-Americans now vote more consistently with hillbillies than other Spanish speaking groups? If so, how odd.

The polls in Michigan were fairly close. Maybe polling has to be considered/measured different in each region.

I was consistently polled via text from mid-summer through the last weekend of the election. The questions each time were pretty much the same, as were my responses.

[scrabblequeen40]

40. It's shame, not alienation

For a lot of people, it's not "polite" to admit that you voted for a racist. You're too ashamed to admit it or have your Christianity or morality be undermined by your voting choices. It's the hypocrisy they're afraid others will find out about. It's not a good look for a christian to hate the black man.

They'll tell pollsters 'God loves all his children, even the black ones' and "I would never vote for a racist."

Then in the secrecy of the polling booth, they'll pull the lever for Jesse Helms.

These are fucking cowards.