General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region ForumsA Slogan does ***NOT*** excuse such ABYSMAL POLLING !! The US has WORST polling of ANY developed
Last edited Sat Nov 14, 2020, 10:07 AM - Edit history (2)
... country on the planet hands DOWN. Youtube can tell if I like Marvin Gaye more than Luther or Mrs. Baker but US polling firms can't poll citizens for political races?!?!
BULL FUCKIN SHIT !!!!
Make the polling firms explain why their polling is so bad, we at least need to start there.
EDIT: Improperly weighted voter suppression measures could be an empirical reason the polls are so off titled towards republicans.
No other country has had polling so bad for so long tilted towards one party in the developed world ... that's a fact. There are very few exceptions but not freakin 20 years of it.
No other country is putting up with poling that has multiple candidates in parliament races winning outside of the aggregate MOEs on LV polling for months then 7 out of ten of these candidates get defeated .... bullshit.
There's no polling (if one were to believe one anyway) that claims a damn slogan or anything else screwed up so many down ballot races either ... NOT ONE, the rest is guessing I don't care who it is.
Yes, the people who put a man on the moon before I was born, the country with the most prizes for science and innovation, the country that implemented the internet on a wide scale for the world .... can't poll its citizens worth a damn.
There's either horrible polling or some down ballot shit that's non democratic.
What does bad polling get us?
Jaime Harrison ... 100 million dollars put in a race that was a blow out for Graham ... did Harrison not have internals that saw what was happening?
That's funds that could've gone elsewhere and there were MULTITUDEs of polling for that race.
What about the Jan 5th runoff for the senate ? nope, Polling for that is already bunk and meaningless!!
We need to retune a message? Too bad, no polling !!
We need to canvas an area more ? Too bad no polling !!
We need to tell more truths about the opponent on radio? Too bad no polling !!
There's no reason to believe a letter of of the senate race polling and a good portion of it isn't out already.
Fuck a slogan screwing up the polling so much; DEMAND BETTER POLLING and its available!!!
DEMAND POLLSTERS EXPLAIN THE 20 YRS OF MISSES!!
These polling mess ups for the last 20 years has to stop ... period
patricia92243
(12,595 posts)uponit7771
(90,335 posts)JanMichael
(24,881 posts)The vast sums of $$ dumped into not even close races is a gross abomination of priorities. With $100,000,000 I could directly fund (no other sources) over 1,000 newly constructed, highly energy efficient, deeply affordable, housing units for the homeless, low income, and others that need it.
That's just one failed race. Ridiculous...
USALiberal
(10,877 posts)uponit7771
(90,335 posts)...pollsters explain down to the digit what went so wrong ... AGAIN
USALiberal
(10,877 posts)uponit7771
(90,335 posts)... on where to put the resources effectively.
There is data we just don't have polling science firms explaining themselves enough to know.
I'm not accepting a slogan threw their polling off nationwide by this much, that's BS
USALiberal
(10,877 posts)uponit7771
(90,335 posts)Make them explain, we at least need to start there
njhoneybadger
(3,910 posts)Joe Bidens campaign manager, Jen OMalley Dillon, says the race against President Trump is far closer than conventional wisdom suggests.
Despite polls showing the former vice president with a double-digit lead over Trump nationally and smaller but consistent leads in battleground states, Dillon warned over Twitter on Wednesday night that the contest is a lot closer.
Early voting is already underway in many states, Dillon said. Millions of voters have already cast their ballots. But there is still a long way to go in this campaign, and we think this race is far closer than folks on this website [Twitter] think. Like a lot closer.
(Posters here accused Jen O'Malley Dillon of not being truthful.)
Just saying
uponit7771
(90,335 posts)... in areas etc because that's their data.
The retail pollsters need to explain how they got down ballots so wrong for so long in so many races.
Dems need to know where to put resources not sink 100 million towards a blow out race regardless of dem or rep.
njhoneybadger
(3,910 posts)uponit7771
(90,335 posts)... evidence of it that even kGOP admits to.
Silent3
(15,190 posts)If you understand statistics, you can get quite a bit of good data out of "just" 1500 people.
Sample size is seldom the problem. It's how well the sample, no matter its size, represents the population. If the people you are capable of reaching when you conduct a poll, who are willing to cooperate once you reach them, and who answer truthfully to your questions, are different in aggregate political character than the people who actually go out and vote, that throws off your polling in a way where polling 15,000 or 150,000 people won't improve your accuracy.
I, personally, ignore all phone calls from people I don't know, and if I do happen to answer a pollster, I hang up. Is that behavior equally common among Democrats and Republicans, or does one group do that more than the other?
If I were to bother to answer a pollster, I'd at least answer honestly the questions I were asked. Does that average out to the same for Democrats and Republicans, or skew one way depending on political affiliation?
Who, even if they are willing to talk to a pollster, has the time to talk to a pollster when they get a call? If you're retired or working, have a long commute vs. a short commute, are very busy or not very busy during your off-hours, etc., can vary greatly from person to person, and it's not hard to imagine there are political correlations with those reasons for accessibility.
Again, increasing sample size helps with none of that. What's needed is modeling, which is a mathematical attempt to compensate for skewed sample representation, and that's what's difficult to get right.
I have a hunch (and it's just a hunch) that QAnon paranoia skewed polling these past couple of years, causing a distinctively Republican shift in who was unwilling to talk to pollsters, and tell the truth when they did, in a way that pollsters couldn't have predicted and failed to adjust their models for.
Mariana
(14,854 posts)They'll claim to be registered Democrats or Independents, they'll say they plan to vote for Democratic candidates, they'll lie about their age and their occupation and their income and everything else.
They believe the pollsters to be partisan and unfair, biased against Republicans in general and Trump in particular. Trump himself has said this many times. They want the polls to fail.
tinrobot
(10,893 posts)I don't have evidence, but it seems like the polls got worse the second we started voting on touchscreens.
uponit7771
(90,335 posts)... longer.
Somethings wrong and we need to demand more
bullimiami
(13,083 posts)They had non-verifiable dres previously.
spooky3
(34,434 posts)Its been reliably blue since 2008. I dont have evidence of cause and effect, but it does make you wonder...
uponit7771
(90,335 posts)... weighted in the polls because the pollsters have to expose the efficacy of republican treason
spooky3
(34,434 posts)passed last year, eg, no reason necessary for early and absentee https://www.governor.virginia.gov/newsroom/all-releases/2020/april/headline-856055-en.html
There will be less gerrymandering starting next yearvoters passed a proposition to set up a commission to redesign districts.
uponit7771
(90,335 posts)... ass accurate.
The pollsters want the money?
Why wouldn't they be screaming by now seeing its their necks on the line, Nate just posted the poling in 20 was worse than 16 but ... how in the world are these polling firms supposed to be trusted now.
bullimiami
(13,083 posts)We have the most massive voter suppression of any democratic country.
uponit7771
(90,335 posts)... bad that pollsters dismiss its effect or don't want to believe it.
Somethings wrong with the polling the US, no other country has polling this bad
AmericanCanuck
(1,102 posts)or arrived to late to be counted.
DeJoy had done this on purpose in FL, PA, MI, WI, ME, IA and AZ
What would be nice is a state-by-state tally of mail ballots requested/sent and mail ballots actually received and counted.
uponit7771
(90,335 posts)... and it got worse with kGOP voter suppression and digital voting
bullimiami
(13,083 posts)They have just kept blaming the polling but its not the polling which changed, except for trying to correct for the difference.
It was never the polling.
triron
(21,995 posts)forthemiddle
(1,379 posts)More and more people had cell phones. To the point today that land lines are almost non existent.
Same with caller ID. Most people dont answer calls from numbers they dont know.
Those two things are the reason polls are no longer accurate, in my opinion.
uponit7771
(90,335 posts)... line communication.
Bettie
(16,087 posts)Republicans are incapable of telling the truth.
They choose to lie.
Constantly.
It is the one thing we can be sure of, if a Republican is speaking, he isn't telling the truth. He's not even using weasel words, he's just lying.
Politicians, businessmen, random people on the street, liars correlate strongly with Republicanism.
Plus, at least half of people don't answer their phones when they don't know who is on the other end.
uponit7771
(90,335 posts)... factor of less truthful than democrats.
Humans aren't that unique in a day to day, 80-90% of what we do is very static its the 10-20% variance that make up our personalities etc.
I worked for one of the large search firms, they have general populous data that would blow the minds of a sociologist.
We need to have the polling companies explain this year and 2016 even weigted towards the kGOP.
Bettie
(16,087 posts)Hell, it is maddening to realize that a whole lot of people around us are irredeemably either stupid or malicious. Or both.
uponit7771
(90,335 posts)... but for some reason its not used?
I think its voter suppression measures that aren't weighted in like its supposed to
planetc
(7,804 posts)uponit7771
(90,335 posts)...polling were there's red state judges or red state legislatures that affect who can vote.
Once that's factored in it looks like we can have better polling
planetc
(7,804 posts)If polls are of people who think they can vote, and intend to, then having their votes changed magically by the machines they vote on would produce the same discrepancy.
triron
(21,995 posts)happened.
BootinUp
(47,139 posts)from 538 if anyone is interested
https://fivethirtyeight.com/features/politics-podcast-why-polls-were-off-in-2020-and-why-they-werent-that-bad/
uponit7771
(90,335 posts)... INCREASED by 4% !!
No, Nate ... love you like a play cousin but that doesn't explain the tilt toward republicans for the last couple of decades
BootinUp
(47,139 posts)(polling error) without a lot of information that is technical in nature. Ive listened to half of the podcast. Its very good to that point if you want to understand poll accuracy and sources of error.
uponit7771
(90,335 posts)... the tilt getting worse.
I'm thinking VSM is the one KNOWN and recognized factor that isn't accounted for, it would reveal a lot that people don't want to address
BootinUp
(47,139 posts)might not be right. Historical polling error, and what are likely explanations for errors.
uponit7771
(90,335 posts)The Genealogist
(4,723 posts)Almost impossible, I know, with numbers being thrown at you constantly. I actively tried to ignore them this time. I felt not nearly as let down with the results compared to how I felt in 2016. Just assume the polls are significantly slanted to look like Dems are ahead, then work with that assumption to get the vote out.
uponit7771
(90,335 posts)... could've gone elsewhere.
We need to GOTV for instance but not in a place that's already 95% voting for example
Data companies in the US need to explain themselves to both sides of the ailse but mostly to dems seeing 70 - 80% of the wrong polling has been tilted republican.
The Genealogist
(4,723 posts)I would guess that a LOT of the RumpHumpers refused to participate or just flat out lied. How do you do get accurate polling data with such cases? I think the same subset of the population has been doing it for years before the Dotard drug his Festivus pole onto the political scene. It is their right to do so, but if enough of these people behave in this manner, it is going to skew results. I also think Democrats are more honest, and are more willing to participate than the RumpHumper types. I see little reason to trust polls in such cases. I'd certainly like to hear a candid and earnest explanation from the polling companies as to why they keep messing up, it if is as I think, or if there are other factors involved.
uponit7771
(90,335 posts)... as amorphic as people think, polling sciences were fairly accurate without digitization but now forget about it.
The search firm I worked at had physicist working on variances in language and some of them were scary accurate in what people would do next.
We have the technology today be as accurate as other countries, Americans aren't that bad ... i hope
coti
(4,612 posts)These people lie to make themselves look stronger or better, not weaker. They're not lying to pollsters to screw with the pollsters. They hate that the polls make them look weak- think about Trump.
I don't much buy they're lying out of shame either. If they had shame, they wouldn't be voting for Trump.
LeftInTX
(25,224 posts)Or was he just really good a raising funds
I constantly got fundraising emails from him...Constantly!
uponit7771
(90,335 posts)triron
(21,995 posts)We do not know how many ballots never arrived or arrived too late.
And there's also widespread voter suppression in swing states.
We also have a polarized population and many in the electorate are either ignorantly dogmatic
or just plain gullible. A democracy needs enlightened people. Americans are largely NOT in that
category. I'm sure there are a multitude of factors.
uponit7771
(90,335 posts)... got so far off.
Even worse than 2016 Nate Silver explains
VSM is my first theory too, its' not weighted properly in KGOP controlled areas cause to do so would show how treasonous the kGOP schemes of suppression are.
triron
(21,995 posts)The pandemic and mail-in voting. VSMs were always there.
bbgrunt
(5,281 posts)Maybe with such bad polling results we will rely a little less on them to discuss elections in terms of horse races a put a little more emphasis on actual issues.
Maybe with such bad polling results we will rely a little less on their propaganda use and ability to shape people's opinions and get people to rely a little more on their own cognitive abilities.
I am sick and tired of seeing all "discussions" about supposed poll number results instead of substantive issues.
In fact, in the few times I have been polled, I have made an extra effort to screw up the results as much as possible either by refusing to answer or giving deliberately misleading answers. Perhaps others are doing the same. I would consider that a GOOD thing.
uponit7771
(90,335 posts)... valuable resources to lost causes like Harrison's race, that was a 100,000 million down the drain to no where.
One persons substantive issue is another persons slogan that turns them off for instance, we don't know that unless we ask people but it looks like since the firms that do the polling don't have to answer for their lack of consistency we'll never know
Fiendish Thingy
(15,569 posts)Your comparison of software algorithms to live polling is not even apples to oranges, its apples to wombats. Also, polls of values or preferences (Medicare for all, gun control, abortion rights) are different than polls trying to predict behaviour (voting).
You also dont take into consideration the pandemic and voter suppression.
Most of the 2016 polls (averages) were within the margin of error.
Many 2020 polls (averages) were also within the margin of error, but that error was all in Trumps favor.
Because of low participation, a 600 respondent state poll generally has a 4-5% MOE. In a two candidate race, you must have a margin DOUBLE the MOE to get beyond the range of possible results that include the other candidate beating the apparent winner. For example:
A poll with Biden 50-Trump 45 with a 4 point MOE, could also be, at its extremes, Trump 49- Biden 46, or Biden 54- Trump 41. The question is: why have the errors all been in Trumps favor?
Lastly, America is changing; the Rust Belt is no longer The Blue Wall, and the Sun Belt is gaining swing states at a rapid rate. Over the last 20 years, VA, & CO have gone from reliably red to dependably blue.
uponit7771
(90,335 posts)Fiendish Thingy
(15,569 posts)Any sample of 600 is going to have a larger MOE than one with 1,000-1,500, and that is what most people dont understand or take into consideration, both pre- and post-election.
Getting a 1,500 sample size for a state poll would be very expensive and logistically challenging.
Response to uponit7771 (Original post)
ahoysrcsm This message was self-deleted by its author.
gulliver
(13,180 posts)Otherwise, you simply can't get honest, reliable responses, imo. I don't see how it's even possible to claim that polling works these days if it's based on random attempts to contact people, most of whom don't answer or who give answers they don't believe just to mess with pollsters deliberately. The election proves that political polling doesn't work. It's that simple.
If you want political polling to work, pay people to be polled. Even then, don't assume that polls are predictive. They're in a feedback loop with voters. One group sees they're behind and suddenly they become more likely to vote. If we had mandatory participation, then that problem wouldn't be there. But we are far from mandatory participation. We live, instead, with a ridiculous level of voter suppression.
How are we still having this problem in 2020?