Could we please begin our examination of why "all the polls were wrong" by vowing

[Atticus]

not to use the phrase "that couldn't happen here"?

5 years ago Trump "couldn't happen here";

a total refusal of the US Senate to convict him despite overwhelming evidence of his criminality "couldn't happen here";

daily routine grifting by the POTUS and a total disregard of the Emoluments clause "couldn't happen here";

public presidential approval of violent white supremacy "couldn't happen here"

separating nursing infants from their mothers and storing them and their siblings in wire cages "couldn't happen here".

We all know that list could go on and on. Can we really say that widespread manipulation of voting machines, scanners and associated computers "can't happen here"?

We are currently being deluged with the ASSUMPTION that "the polls were all wrong". Maybe. Was there any significant difference between the PAPER ballots ( by mail or drop box ) and the polls? How about the machine votes?

These questions need to be answered.

[Johnny2X2X]

1. We don't even know if the polls were wrong or by how much yet

They're still counting votes everywhere and Biden is going to win by 7% or more.

[exboyfil]

6. Wisconsin +16, Michigan +8, Pennsylvania +6

Siena College/The New York Times Upshot needs to go into a different line of work.

[Johnny2X2X]

9. Average poll was WI +8 on 538, +6.7 on RCP.

So WI was way off, a couple other state polls were off a few points, most were within the margin of error.

[Happy Hoosier]

28. PA MAY be Biden +3 by the time all the votes are counted.

The 538 average there was around +5.

Yes, there was a problem with the polls in some states, but it is being WAY overstated IMO.

[Tribetime]

26. Ohio was 8% off on average..PA WI AND MI were way off

[Polybius]

53. PA wasn't way off

Most were 2-5%. Some polls even had Trump ahead.

[Tribetime]

59. I thought PA avg 6 or 7 %

[Nederland]

2. We can do that

However, I would start by saying exactly what I say to Trump supporters: show me the evidence.

[Atticus]

23. And I start by saying we can't refuse to investigate and then fold our arms and scratch our

heads and agree "Yup, there's no evidence!"

[Beakybird]

3. The polls told us Collins, Ernst, Graham, Tillis were in trouble. They weren't.

[Zing Zing Zingbah]

35. Collins was in trouble

51% isn't decisive for a 4 term incumbent, but it's enough to win. I think what happens is the undecideds just go with the incumbents at the last minute. Plus Collins made it look like she was still moderate to them by voting against going forward with supreme court confirmation before the election. That probably won over some of the those undecided people who seem to not have much memory of the past four years... only recent events stick in their minds. Collins sure knows how to play politics for sure. I can't stand her, but she knows what she is doing and she was able to distance herself from Trump.

[Beakybird]

36. If the polls were more accurate, I would have donated to Gideon and Cunningham

Instead of to Montana and South Carolina. But I am glad that I donated to Ossoff. He still has a chance.

[Zing Zing Zingbah]

40. Gideon didn't have a cash problem anyhow

Plus it probably hurt having all that money coming in from outside Maine. That's something that Mainers don't like. They take it as out-of-staters trying to influence their elections. I think that might have been a factor in Gideon losing too. It was all over the papers here how much money her campaign had. She had the largest campaign chest in Maine election history. So, maybe next time people from out state don't donate unless the Maine dem is really hurting for funds as compared to the repub. Gideon had way more money than Collins. That probably made Collins look like an underdog. People from other places don't realize that about Maine people. I can see an undecided breaking for Collins because of that obscene amount of money Gideon had. That makes people here think she isn't going to represent us because she's being bought by out of state interests. Too much money in a race here results in the opposite of the intended effect.

"Maine's race makes the top four largely because of the amount of money brought in by Gideon. With more than $68 million, Gideon has raised the fourth most of any candidate running in a national Senate race.

The most recent election data, which was released on Oct. 14, shows Gideon has raised $68,577,474, spent $47,908,382 and has $20,670,197 on hand.

Susan Collins has raised more money than ever before while running for office with $26,511,555. According to the most recent data she has spent $23,010,418 and has $4,429,900 on hand. "

https://www.newscentermaine.com/article/news/politics/gideon-nets-4th-most-money-of-any-senate-candidate-nationally/97-598d151f-8818-4fb9-a79c-7e69048e8c63

[Polybius]

55. Yes but her opponent got only 42.9

Polls were hugely off on that one. Every one of them said Gideon was at least 5 points ahead. That's a 13 point swing.

[Zing Zing Zingbah]

60. Undecided independents who made up their minds at the last minute

and probably also were turned off by the money the Gideon campaign was bringing in.

[Polybius]

62. I think she fooled a lot of morons with her vote against ACB

She really did.

[Zing Zing Zingbah]

63. Yes, that too. n/t

[apnu]

4. Polls are unreliable when the subject lies

There is a percentage of Republicans who love getting poll calls and totally fucking with the pollster. I know a few, its the best entertainment they have.

The polls are dirty, not just because manipulative pollsters, but manipulative subjects as well.

[exboyfil]

7. No one has an obligation to be truthful

Pollsters get a click from me.

[uponit7771]

15. Americans aren't more dishonest than S Koreans, our polling is horrid compared to theirs

[dustyscamp]

42. Not just Pugs some Libertarians probably did this too

[coti]

61. I don't buy that at all. Trumpists want to project as

strong an image as possible.

[Shermann]

5. It will be interesting to see where the popular vote ends up at the end

Less than 3 points right now with all those mail-in votes still being counted.

I don't believe it will get close to those 10 point predictions, but maybe it will get within that margin of error?

[exboyfil]

8. No one should really care about national polls

As well as should be considered is the overall totals in the country. How much Presidential campaigning goes on in California for example? That alone skews any final result. We set up an obscene system that ignores most of the country for a few battleground states.

[ecstatic]

10. I think people are afraid to confront the topic because they don't want to discourage voting

I understand the fear, after all, how would we get people to stand in line for hours and hours if they think their votes won't count?

So our strategy for the past 12 years has been to try to overwhelm GOP cheating with massive turnout....

But you know what? That's not fair, and it's not right. We need to demand transparency regarding these ev machines. There definitely needs to be oversight and accountability. We need automatic audits of every election result.

In the meantime, since switching to paper ballots, I've never felt more confident about my vote being ACCURATELY counted.

[uponit7771]

19. +1

[Baitball Blogger]

11. Is it possible, especially in Florida and South Carolina, that

there was massive ballot fraud going on from the Republicans?

[YessirAtsaFact]

14. Absolutely

If they use touch screen machines, who examines the programming and validates that it works correctly? It's the company that built it, in conjunction with the same state officials who bought the system, who in SC and FL may very well be crooks.

How do you do an audit? It's impossible on a touch screen system, because there is no human marked paper to check vote totals against.

[Shermann]

17. FALSE

SC uses a hybrid system of electronic voting machines with printed ballots.

[YessirAtsaFact]

22. Glad to be corrected

Questions: 1 - Do people mark the ballots?
2 - Does the system retain the paper trail?

As long as there is a paper trail, cheating less likely.

The boards of elections won't do it unless they are forced, but I'd love to see hand audits of the inputs checked against the results.

[Shermann]

47. Linky

https://ballotpedia.org/Voting_methods_and_equipment_by_state

[Nederland]

30. Which technology is used is irrelevant

There is a phrase in the security business: Security is a process, not a product.

https://www.schneier.com/essays/archives/2000/04/the_process_of_secur.html#:~:text=Security%20is%20a%20process%2C%20not,of%20the%20products%20or%20patches

If you think using paper ballots protects you from fraud you are fooling yourself--it merely changes the type of fraud that is possible. The key to keeping our elections safe is to establish safe and reliable processes, and a way to verify that those processes were followed during the election. True, paper ballots have an advantage in that the technology involved in certain electronic voting machines is beyond the understanding of most people. As a result, your average person can observe the voting and counting of paper ballots and come away with a sense of certainty that is completely lacking with a 100% electronic system--and that sense of certainty is very important. If people lose faith in the system things unravel pretty quickly.

[YessirAtsaFact]

41. I partially agree

Process is essential, but we have to be able to get back to the source of the votes in order to do a real audit of an election.

You can't base an audit the placement of a finger on a touchscreen. Once the screen is cleared for the next voter, there is no independent record of the input, just data in a database.

Scanned ballots in locked boxes can be counted by human beings after the fact and compared to the captured data.

[Nederland]

48. This is the problem

The root problem in validating an election derives from the legal guarantee we all have that our vote is anonymous. The guarantee of anonymity serves two purposes. It protects people from threats like "if you don't vote for X we are going to burn your house down". It also prevents the buying of people's votes--if a person cannot prove who they voted for, bad actors are little inclined to offer them money to vote a certain way.

While the guarantee of anonymity is a crucial right in a democracy, it creates an intrinsic problem for validating elections. Regardless of what technology is used, anonymous voting means that once you vote, any information describing how you voted is irrevocably separated from you. It doesn't matter whether that vote was recorded by marks on a piece of paper or by voltage fluctuations on a piece of silicon, once you've voted, the record of how you voted goes into someone else's hands. That vote may be placed in a locked box that can only be opened by a select set of people holding metal keys, or it may be placed into a computer file that has been encrypted and can only be opened by selected set of people holding digital keys. It doesn't matter. Whatever technology is used to protect your vote from being changed, your confidence that your vote will eventually be counted correctly is based on a common set of assumptions. You have to trust the people holding the keys. You have to trust there is a process that protects those people from others trying to steal the keys from them. And once your vote has been "unlocked", you have to trust the people doing the counting--whether those people are volunteers sitting in chairs shuffling pieces of paper or programmers sitting in chairs shuffling bits and bytes. Ultimately, regardless of what method is used, you have to trust the system that was created and the people involved.

[YessirAtsaFact]

12. We need to eliminate any voting machine that doesn't scan a human marked ballot

In VA, I take a piece of paper with names and ovals on it.

I color in the ovals beside the names I like and feed that piece of paper into a scanner.

This year, the scanner read my votes for Joe Biden, Mark Warner and Abagail Spanberger into the elections database and dropped the paper in a locked bin.

If the board of elections wants to, they can count the votes on the paper and cross check those totals against the database totals.

We should do this everywhere. This setup, machine counting of human marked paper, should be the norm, so if the results diverge from the polling, humans can audit the returns by looking at the human marked inputs to determine if we have bad polling or fraud.

I'm convinced that touchscreen voting machines with no paper trail are giving incorrect results. In some cases it is deliberate fraud.

In others it is because they are built quickly, cheaply and badly by some fly-by-night company owned by a politician's brother-in-law and they suck as data processing systems.

[cbdo2007]

13. They polls weren't that far off....but they still have a lot to figure out on them.

One thing they need to take into account is "poll fatigue" and the number of people who actually participate in polls. I'm in Missouri and I would get 2 calls per day asking me to participate in a poll. One time I did because I thought, oh this will be fun, so I put in Biden as my choice and then it was just a big ad for why I should vote for Trump instead.

I don't know if any of them do in person polls anymore, but I definitely think call/text/internet based polls are not as accurate because people aren't as willing to participate.

[Jarqui]

16. Susan Collins led in only one poll in 2020 that I've seen.

But a whole bunch of pollsters are terribly wrong and should be condemned while her victory should not raise eyebrows? I don't think so.

We know the USPS was up to no good.
Who knows what else.

But I do not think throwing up our hands and saying "the pollsters were all wrong" is an answer I'm prepared to accept right now.

I also have questions about Miami-Dade for example. Biden beat Hillary nearly everywhere else it seems in FL. What happened in Miami-Dade? Was it really just the Cubans? I'd prefer facts over opinion.

The pollsters could be evidence of something else going on.
They revamped their approach after Hillary's election.
We have pollsters like Rasmussen who are not really pollsters.
But they're all bad and out to lunch? I doubt that.

We have the most corrupt President and Senate in the history of the country. Republicans have been stealing elections for much of my lifetime - all kinds of dirty tricks. There is probably a component of that going on here.

I was so convinced they were up to no good, I felt Joe had to win in a landslide just to eek out a win. That's what feels like happened. And Trump may well not leave and try to get Republican legislatures to steal it.

[Atticus]

18. Thank you. nt

[Tommymac]

39. +100000 I hope We The People decide to make some Good Trouble over this issue.

Thanks for the thoughts.

[Chakaconcarne]

20. Lets get paper ballots everywhere and then revisit the question.

There's no way Trump came this close without some voting machine shenanigans.

[Pobeka]

21. I'm baffled why people don't understand polls are not subject to suppression/tampering.

And that simple fact means a poll can give a margin of error that is different than an election result, not because the poll expressing the voter intent was wrong, but when it came to the voter not being able to cast a ballot due to suppression, or the actual count was altered, it *necessarily* means the poll will have a different estimate than the final result.

We cannot know for certain how large the tampering/suppression effect is.

We cannot know for certain how many people are gaming the pollster's and lying about intending to vote democratic, when they really intend to vote republican.

But when I see the Fox News poll results from a few days ago about how people feel about government policies around social security, health care, immigration, etc, it really makes me wonder, because I think people would be much less likely to lie to their own cult media station -- they want to see Fox News telling the story they believe to be true, not a story they artificially generate by lying.

A statistical generalization -- most polls give a 95% confidence interval, which means that we actually expect that 1 out of 20 times the voting result will exceed that margin (too low OR too high). When we have these polls so close to the election, this far off *on one side*, it means with very high certainty (much more the 95%) that either 1) people are actively lying to the pollsters, or 2) there is some sort of suppression or vote tampering occurring.

[Atticus]

24. THIS indeed! Thank you. nt

[Pobeka]

25. You're welcome.

30 years in a field with a lot of statistics use should be worth something

[Atticus]

27. I just read your post to my wife and commented "This poster is obviously a statistician

or in some closely related field".

[Nederland]

31. Or possibility #3

3) the sample in the poll was not representative of population that voted

[Pobeka]

32. That is built into the confidence interval. That's WHY there is a confidence interval. n/t

[Nederland]

45. It is not "built into" the confidence interval, it is an assumption of the confidence interval

Assumptions and Conditions
When constructing confidence intervals the assumptions and conditions of the central limit theorem must be met in order to use the normal model.

Randomization Condition: The data must be sampled randomly. Is one of the good sampling methodologies discussed in the Sampling and Data chapter being used?

https://cnx.org/contents/KnmPEWac

The data gathered by pollsters should never be considered a statistically random sample.

[Pobeka]

46. If you are going to argue that pollsters don't know how to take a random sample, good luck with that

Let me know what you find.

The whole reason for a confidence interval is that you *know* a sample is not likely to exactly match the true population mean, so you use a confidence interval to estimate the distribution of means generated by further samples, which is a good surrogate for the true population mean.

[Nederland]

54. Nate Silver says the problem lies in getting a random sample, good luck arguing with him

There was plenty of apprehension at the annual conference of the American Association for Public Opinion Research, which I attended last month, about the state of the polling industry.

The problem is simple but daunting. The foundation of opinion research has historically been the ability to draw a random sample of the population. That’s become much harder to do, at least in the United States. Response rates to telephone surveys have been declining for years and are often in the single digits, even for the highest-quality polls. The relatively few people who respond to polls may not be representative of the majority who don’t. Last week, the Federal Communications Commission proposed new guidelines that could make telephone polling even harder by enabling phone companies to block calls placed by automated dialers, a tool used in almost all surveys.

https://fivethirtyeight.com/features/polling-is-getting-harder-but-its-a-vital-check-on-power/

[Atticus]

33. It was not just "the" poll---it was ALL the polls. I can't believe they all chose unrepresentative

survey populations which were all skewed the same way.

[Nederland]

37. Why is it so hard to believe?

The representative sample a poll uses is an estimate or guess as to who will be voting in the upcoming election--and that estimate is in no small part decided by looking at who voted in the last election. Given that this election had record turnout in which millions of people that did not vote last time (or perhaps ever) showed up at the polls, why is it hard to believe pollsters constructed their samples wrong?

[Atticus]

49. Your response discusses "a" poll. My post discussed "all" polls. Big difference. nt

[Nederland]

50. All polls start with the same data

They ALL look at the data from past elections to figure out what a representative sample should look like. As a result, they ALL suffered from the same problem: millions who didn't vote last time voted this time.

[Atticus]

51. We disagree. You make assumptions I consider unsupported and unsupportable. You can

have the last word as I doubt further back-and-forth will be productive or beneficial for either of us.

I do appreciate your civility and obvious intellect. I am sure we agree on much more than might be apparent from just this exchange.

[Cicada]

29. Here are several reasons they were wrong

I read an op Ed a couple weeks before Election Day by a woman who argued that the polls might end up wrong because 2020 is a very weird year. Covid, massive economic disruption, so our normal rules about likely voter turnout were very iffy. She said we couldn’t really project turnout rates by different groups. Also with so many absentee ballots a bigger share of dem votes would be excluded. Absentee ballots get rejected, so that suppressed the dem vote from normal. Post Office was late so in some states that reduced dem votes. Voter suppression was up. Only one drop box in Harris county, reduced number of polling places. Something like 15 million votes still haven’t been counted. The Blue shift will add maybe one percent to Biden’s margin of victory. So RCP had Biden leading by 7 points. After the Blue Shift his margin might end up four points. The extra rejections due to so many more absentee votes, ans other voter suppression are probably worth another one percent of margin. So on a national level the error after those adjustments might be not far from the historic average.but individual states? The historic average error for a state poll is six points. State polls always suck. Unless the pollster is that Iowa lady Seltzer. And Trump is unique. His voters do seem to be paranoid nuts who figure pollsters are really trickster pedophiles or something so they don’t respond.

[GoCubsGo]

38. Were they wrong, or were they an indicator of foul play?

My money is on the latter.

[dustyscamp]

43. Well, Biden and his team were right to tell us not to trust them

Link to tweet

Link to tweet

?lang=en

Link to tweet

Link to tweet

?lang=en

Link to tweet

?lang=en


“Those are inflated national public polling numbers.”
"The reality is that this race is far closer than some of the punditry we're seeing on Twitter and on TV would suggest," O'Malley Dillon wrote

I think the polls made a lot of us super cocky and got us to think we were going to get the Senate too

[Demsrule86]

44. No, let's get through this. It seems to me after examining the polls there was a change very

close to the election where the economy replaced Covid as the big issue. I saw this reflected in exit polls. and that may have made the difference as we know Trump has and edge on this issue...although he shouldn't.

[Polybius]

56. I think I know exactly why they were wrong

Many first-time young voters (who overwhelmingly said they would choose Biden) chose to request a mail-in ballot. Well, when the time came to send it in, for whatever reason they couldn't be bothered. Many probably don't even know how to mail a letter, since they have never had to.

[Buckeyeblue]

57. I don't think the polls anticipated the higher turn out for Trump

I think his numbers seemed flat or even down a little. If he would have had similar numbers as 2016, many polls would have been closer.

Why was that missed? I'm not sure. I honestly think the increase turnout for Trump was people who don't want another covid shutdown.

We have two years to swing voters around to have a strong mid-term. Which we need to start working on now.

[MoonlitKnight]

58. It's a bit of both

Or maybe a lot of both in some states.