| What baffles me most is... 13:45 - Nov 4 with 898 views | monytowbray |
...how off polling has been in recent years. Why is this? |  |
| |  |
| What baffles me most is... on 13:49 - Nov 4 with 874 views | itfcjoe |
I think it's overplayed how off it is, bar Trump in 2016 most has been within margins of error.
Even now fivethirtyeight gave Biden an 89% chance of being elected, on Twitter Nate Silver, after the Omaha seat in Nebraska flips says it leaves him at approx 85%. |  |
| |
| What baffles me most is... on 13:49 - Nov 4 with 873 views | wkj |
Polling will always be less reliable with tight elections. They're generally a great predictor in a simple vote vs vote election, but in an electoral college vote they are very hard to take as gospel. Although the electoral college representation by the polls is way of, we may find the popular vote is closer to what the polls predicted. |  |
| |
| What baffles me most is... on 13:51 - Nov 4 with 862 views | clive_baker |
No real surprise that people are more likely to lie about being a bigot and offering their support to a genuine psychopath than being someone who cares for others and wants to live in a fairer society?
See also Brexit polls. [Post edited 4 Nov 2020 13:52]
|  |
| |
| What baffles me most is... on 13:55 - Nov 4 with 837 views | FoghornGleghorn |
| What baffles me most is... on 13:51 - Nov 4 by clive_baker |
No real surprise that people are more likely to lie about being a bigot and offering their support to a genuine psychopath than being someone who cares for others and wants to live in a fairer society?
See also Brexit polls. [Post edited 4 Nov 2020 13:52]
|
Thing is, this is the problem though isn't it? Trump's an utterly irretrievable c-note, but is branding everyone who votes for him a bigot going to persuade them not to, or is it just going to persuade them not tell the polls what they're intending to do? Reasoned debate's given way to insults, generalisation, labelling and memes. |  | | |
| What baffles me most is... on 13:55 - Nov 4 with 832 views | Ftnfwest |
more reliance on social media |  | | |
| What baffles me most is... on 13:56 - Nov 4 with 817 views | J2BLUE |
| What baffles me most is... on 13:51 - Nov 4 by clive_baker |
No real surprise that people are more likely to lie about being a bigot and offering their support to a genuine psychopath than being someone who cares for others and wants to live in a fairer society?
See also Brexit polls. [Post edited 4 Nov 2020 13:52]
|
Every time they deny that the shy Brexiteer, Trump voter or Tory exists. Then they're proven wrong.
All the predictions of Biden in the high 300s or 400s were always going to be way out but once you've had that opinion validated in your echo chamber for months people actually start to believe it. |  |
| |
| What baffles me most is... on 14:00 - Nov 4 with 795 views | C_HealyIsAPleasure |
2 reasons really:
- elections aren’t generally straight up one or other polls, things like constituencies in the UK or electoral colleges in the US mean a lot of very complicated modelling and you don’t have to get a lot wrong to end up looking further out than you are. The one obvious exception to this was the Brexit vote however that was always likely to be volatile as a one-off so a massive gap in historical data
- the world has arguably changed massively in a short space of time. Poll modelling is incredibly complex but by nature there is a reactive element to what has happened before so by nature behind the curve somewhat when there is a significant shift in expected behaviour
EDIT: there’s a lot of other complex stuff too and some polling models are just inherently flawed too, of course [Post edited 4 Nov 2020 14:02]
|  |
| |
| What baffles me most is... on 14:11 - Nov 4 with 742 views | Steve_M |
| What baffles me most is... on 14:00 - Nov 4 by C_HealyIsAPleasure |
2 reasons really:
- elections aren’t generally straight up one or other polls, things like constituencies in the UK or electoral colleges in the US mean a lot of very complicated modelling and you don’t have to get a lot wrong to end up looking further out than you are. The one obvious exception to this was the Brexit vote however that was always likely to be volatile as a one-off so a massive gap in historical data
- the world has arguably changed massively in a short space of time. Poll modelling is incredibly complex but by nature there is a reactive element to what has happened before so by nature behind the curve somewhat when there is a significant shift in expected behaviour
EDIT: there’s a lot of other complex stuff too and some polling models are just inherently flawed too, of course [Post edited 4 Nov 2020 14:02]
|
It might be too early to say but it does look that the state level polling in the US isn't as robust as the national one. Higher turnout - and that is the case this time in the US - also has an effect.
As much as anything else, polls are probability models and have a margin of error inherent to them. I wish I could remember more of the statistics stuff I have done but a basic grasp of probability would be helpful for many people. |  |
| |
Login to get fewer ads
| What baffles me most is... on 14:18 - Nov 4 with 720 views | gordon |
Not sure the polls are that far off in this case. US presidential elections always swing a lot on small margins in key states, so outcomes are difficult to predict based on vote share. |  | | |
| What baffles me most is... on 14:27 - Nov 4 with 696 views | BloomBlue |
| What baffles me most is... on 13:51 - Nov 4 by clive_baker |
No real surprise that people are more likely to lie about being a bigot and offering their support to a genuine psychopath than being someone who cares for others and wants to live in a fairer society?
See also Brexit polls. [Post edited 4 Nov 2020 13:52]
|
But in the last two GE's in the UK the pre polling has been wrong by a lot in some cases but the exit poll the BBC have published immediately after voting has closed at 10pm has been spot-on at both GE. Which would indicate people are happy to be honest about who they voted for after they've voted.
So is it more of a case of people unsure or changing their minds before voting day? |  | | |
| What baffles me most is... on 16:46 - Nov 4 with 582 views | Guthrum |
| What baffles me most is... on 14:00 - Nov 4 by C_HealyIsAPleasure |
2 reasons really:
- elections aren’t generally straight up one or other polls, things like constituencies in the UK or electoral colleges in the US mean a lot of very complicated modelling and you don’t have to get a lot wrong to end up looking further out than you are. The one obvious exception to this was the Brexit vote however that was always likely to be volatile as a one-off so a massive gap in historical data
- the world has arguably changed massively in a short space of time. Poll modelling is incredibly complex but by nature there is a reactive element to what has happened before so by nature behind the curve somewhat when there is a significant shift in expected behaviour
EDIT: there’s a lot of other complex stuff too and some polling models are just inherently flawed too, of course [Post edited 4 Nov 2020 14:02]
|
Also people telling polling companies one thing, then doing another once in the booth. Especially voters considering switching parties, but with emotional ties to their old home. |  |
| |
| What baffles me most is... on 17:06 - Nov 4 with 554 views | tractordownsouth |
US polling always has more potential to be volatile, because there's usually such low turnout that there's a large pool of potential voters for the candidates to energise
Florida is probably the biggest error so far, but I don't understand why so many polls were so bullish about it being blue. It trended right in 2018 despite the Democrats doing well nationally - I never had it down for Biden. |  |
| |
| What baffles me most is... on 17:09 - Nov 4 with 545 views | Mullet |
| What baffles me most is... on 17:06 - Nov 4 by tractordownsouth |
US polling always has more potential to be volatile, because there's usually such low turnout that there's a large pool of potential voters for the candidates to energise
Florida is probably the biggest error so far, but I don't understand why so many polls were so bullish about it being blue. It trended right in 2018 despite the Democrats doing well nationally - I never had it down for Biden. |
Hasn't there been some controversy over postal ballots in Florida being dumped or ignored, or is that just conspiracy nonsense seeping out?
It seems that like a tinpot dictator Trump won't accept anything going against him and people will get hurt in the coming days because of it. Some of the radicalisation evident on twitter is frightening. |  |
| |
| What baffles me most is... on 17:11 - Nov 4 with 539 views | eireblue |
https://en.wikipedia.org/wiki/Social-desirability_bias
The Trafalgar Group polling company, that seems to be quite accurate, claims to use social desirability, anonymity and short questionnaires to get more accurate results.
E.g. people may not same something that is perceived to be bad, especially when they think their responses are not secret, and they get bored and don’t really answer lots of questions accurately, just quickly. |  | | |
| |
By continuing to use the site, you agree to our use of cookies and to abide by our Terms and Conditions.
