[EU] Brexit negotiations at a standstill

[Mayhem]

Democracy works by allowing everyone in the country who is an adult to vote and then checking the results.

Democracy does not work by taking a statistical projection from a supposedly representative sample of 1000 people or less and extrapolating across the whole nation, which is what Slant was implying-government by opinion poll, a ridiculous concept.

Your reading comprehension sucks.

[Lollis]

Democracy does not work by taking a statistical projection from a supposedly representative sample of 1000 people or less and extrapolating across the whole nation, which is what Slant was implying-government by opinion poll, a ridiculous concept.

Somebody doesn't understand statistics.

[tyoplapia]

Somebody doesn't understand statistics.

Uh-huh, I didn't know we were in the presence of an expert. Presumably can explain why the polls were well outside the margin of statistical error in the UK in 2015, failed to predict a Trump win in 2016, grossly understated Labour's share of the vote again in 2017 and have widely failed to be accurate in predicting the result in polls across the civilized world for years. You can also presumably explain why every major figure in polling from Nate Silver to Anthony Wells is at a loss to even explain why.

Let me guess: do you think the problem is the inherent non-linear dynamics in a complex process where the fringes of the distribution behave unpredictably due to a reporting bias? Do you think it may be down to a overcompensation on the part of the pollsters to correct prior errors and is simple regression to the mean?

I obviously know nothing about these things: it is not like I bet serious money on all those events and won every time because I study this shit 24/7.

So please, enlighten us, I mean it obviously isn't the case that you just don't know what you are talking about or anything...

[Lollis]

failed to predict a Trump win in 2016

Because the polls were based on votes, and Clinton won the popular vote?

1000 people is more than enough as a significant sample size.

[Kallisto]

Uh-huh, I didn't know we were in the presence of an expert. Presumably can explain why the polls were well outside the margin of statistical error in the UK in 2015

Shy Tory.

failed to predict a Trump win in 2016

Polls were mostly down to national level intention of voting which they got spot on. Each state in which Trump won was well within margin of error.

grossly understated Labour's share of the vote again in 2017

Polling was showing major gains for Labour in the build up to the election and was again within margin of error.

have widely failed to be accurate in predicting the result in polls across the civilized world for years.

Wrong they have been very close in pretty much spot on.

You can also presumably explain why every major figure in polling from Nate Silver to Anthony Wells is at a loss to even explain why.

The same Nate Silver that pretty much said "Hillary isn't a sure in and Trump was likely to win." even upto the moments before the voting closed on east coast.

[Slant]

The result has to be implemented before the decision can be reversed. We do not have another election before a government takes office.

That's not actually a requirement in democracy. Many things have been decided and then abandoned in democracies as their outcome was reevaluated and deemed undesired or too costly. Sometimes reality just intervened and made a decision obsolete.

[Shalcker]

Polling was showing major gains for Labour in the build up to the election and was again within margin of error.

Anything can be right if you make margin of error large enough.

[Shalcker]

The margin of error is determined by specific factors, it is not arbitrary.

If actual variation is below margin of error, then you need more precise methodology to get anything useful.

[Shalcker]

Implying that the margin of error is intentionally large enough to misrepresent the result.

I'm clarifying that i implied something else.

Yet what makes something right in statistics is not whether the desired result falls within the margin of error, it is the statistical significance.

Statistics matter in real world proportional to how right they are; not if they pass statistical significance threshold (which can also be somewhat arbitrary and prone to bias). They are all proxies for actual results, approximations.

If some significant result falls within margin of error (like Brexit), then relying on polls to predict outcome is unwise.

Conservatives did it twice, with Brexit and with last GE, and, as far as i see, lost both times.

[Butler to Baby Sloths]

The result has to be implemented before the decision can be reversed. We do not have another election before a government takes office.

Not necessarily.

After a general election the leader of the largest party in the Commons is asked by the Queen to put together a cabinet that can win the support of the House in a vote of confidence (simple majority). If the party majority is large enough, this is generally a formality, otherwise the government can take the form of a formal coalition (like the Tory/LD government in 2010), an informal coalition (like Tory/DUP government in 2017), or they can try a minority government.

If this government fails to win the confidence of the Commons, the would-be Leader of the Opposition gets a go. If their government cannot win the support of the Commons either, then a second GE is called.

Thus a General Election can be called before the "Will of the People" that was expressed merely weeks earlier was implemented if no workable solution to their decision can be found. Which makes an elegant segue back to the situation with Brexit, where the "Will of the People" is not able to be implemented because there is no workable majority in Parliament (or the electorate, for that matter) for any of the outcomes (no deal, the deal, any deal, no brexit).

[Dhrizzle]

Wanting to overturn a democratic mandate makes you illiberal.

There's no democratic mandate fo anything except maybe a Brexit that manages to funnel £350m a week into the NHS without raising taxes or reducing funding elsewhere (including EU projects,) conjures trade deals which favour the UK more than our current arrangements through the EU and reduces net migration whilst keeping migrant workers for essential workers especially the NHS. If someone can ride in on their unicorn and deliver all that (whilst replacing all the rest of the legal framework based on EU institutions) I would accept there is a legal mandate based on the first referendum.

Closest thing to a democratic mandate at the moment is the EU election which shows a slightly higher preference for no Brexit or a second referendum than for a hard Brexit, with a 25% margin of error based on people voting Labour and Conservative with no clear message.

[Shalcker]

And the issue in both cases is explained by bias in sampling by people unwilling to admit their voting preferences. Which is why polls tend to be further weighted. Regardless, the EU elections are not a poll, they are an actual election.

We live in world of imperfect information. As far as i understand studies that actually can discover true voting preferences exist, but are a bit too big and largely incompatible with even monthly polls for random sampling. Kind of like actual fact-checking and deep investigative work is incompatible with "24 hour news" cycle.

Weighting is just introducing different bias; it doesn't necessarily improve things. There are examples of pollsters trying to fix bias in their polling and getting results even more wrong next time around.

And we weren't talking about EU elections.

[tyoplapia]

Because the polls were based on votes, and Clinton won the popular vote?

1000 people is more than enough as a significant sample size.

Please just stop, you are out of your depth. I listed three examples, you can't just cherry-pick one and ignore the other two, especially when trying to patronize someone else about your amazing knowledge of statistics. That is almost the definition of a biased sample.

- - - Updated - - -

Polls were mostly down to national level intention of voting which they got spot on. Each state in which Trump won was well within margin of error.

No, many individual states were way outside the margin of error.

- - - Updated - - -

Polling was showing major gains for Labour in the build up to the election and was again within margin of error.

So understating Labour's share of the vote by 6% is well within the margin of error? You do understand that many elections are decided by less than 1%?

- - - Updated - - -

The same Nate Silver that pretty much said "Hillary isn't a sure in and Trump was likely to win." even upto the moments before the voting closed on east coast.

...Despite the fact that his polling models were suggesting the exact opposite....again, after they completely fell apart in the UK 2015 election. That's your poster boy?

- - - Updated - - -

And the issue in both cases is explained by bias in sampling by people unwilling to admit their voting preferences. Which is why polls tend to be further weighted. Regardless, the EU elections are not a poll, they are an actual election.

I'd be extremely careful in wishing away polling error with simple explanations. Experts that appear on television tend to conjure up these simple narratives to explain away polls that fail to predict elections. You can see why they do that - a simple narrative is persuasive to most people and provides the interview with a satisfying explanation.

If you bet serious money into these things that isn't enough. Knowing the true reason for the failure of a poll or polls is very difficult. Bias is not going to be caused by a single factor, there will usually be multiple biases and how they interact with each other reason is important. Things as random as the weather are an issue: I doubt the 2015 & 2017 biases were much to do with people unwilling to admit their intentions-it was more likely to do with poor population sampling in the first place-people on the fringes tend not to vote so are ignored by pollsters. But it is almost impossible to be certain, it could be something no one has thought of.

The truth is whatever biases appear in each election can only be determined after the fact. Maybe there is a shy tory effect in one election: maybe that's just some bullshit someone made up to save their job. The problem is that even if it exists in one election it verifiably doesn't in others-indeed during the Blair years there was supposed to be a "shy labour" voter. Essentially it might as well just be random noise.

This is why I took exception to Slant's assertion that the mood had changed and the UK now wanted to remain in the EU. The polls got it wrong before the referendum, they could easily be wrong now, they've barely shifted since the actual vote took place.

[LeGin Tufnel]

So.... let's get this straight.

You're a statistician with hard left views who also gambles a lot and you think anyone further right than Kinnock is a cunt.

You must be fun to hang out with.

Shall we do Vegas?

[Shalcker]

Regardless, that does not invalidate polling. Some bias is inevitable. The truth remains that polls have not failed by nearly as large a margin as people keep claiming. And furthermore on the original topic, i.e. establishing whether a political mandate exists, the basis we are currently using is not a poll but a proper election. Yes, it has a low turnout (about 37% when some estimates of the GE turnout in 2017 is more than double at 78%) and yes there is a very implicit bias that favored the attendance of voters with a strong opinion on the EU. But it was still a monumental disaster for the Tories.

Brexit had 72% of registered voter turnout; using about half of that turnout to justify "Brexit not having mandate" is fraught with bias.

EP election was disaster for Tories because plenty of Tories voted Brexit Party there.

[LeGin Tufnel]

Brexit had 72% of registered voter turnout; using about half of that turnout to justify "Brexit not having mandate" is fraught with bias.

I didn't say that. I said there's no mandate for Hard Brexit. And there isn't. There never will be unless there's a second ref and it beats other options on the card.

[Shalcker]

That's an exaggeration. 37% of the electorate is still a massive sample size.

Yes, but it is sampling on different question. With plenty of people not considering those elections relevant to them at all.

Sure, some conjectures can be drawn; it is inadvisable to take them for the fact though.

- - - Updated - - -

I didn't say that. I said there's no mandate for Hard Brexit. And there isn't. There never will be unless there's a second ref and it beats other options on the card.

As far as i see result depends entirely on choice of questions in second ref.

Just Brexit vs Remain? Brexit probably wins again by small margin, just because "asking twice is stupid and undermines democracy, get on with it already".
Hard Brexit vs CU vs May's deal vs Remain? Remain obviously wins.
All kinds of Brexits vs all kinds of Remains? Toss up.

[Kallisto]

So.... let's get this straight.

You're a statistician with hard left views who also gambles a lot and you think anyone further right than Kinnock is a cunt.

You must be fun to hang out with.

Shall we do Vegas?

Kinnock?

Try Trotsky with this guy.

[Slant]

I've said it before, I'll say it again.. if you had set out to just destroy the UK as an entity, you can't do much better than the Brexiteers:

https://www.theguardian.com/business...print-contract

The UK banknote printer De La Rue is parting ways with its chief executive a year after he lost the battle for a £490m contract to print the post-Brexit blue passport.
[...]
The news follows a highly publicised battle for the contract to print the UK’s post-Brexit passport, which was given to Franco-Dutch company Gemalto.

Okay, so they hate foreigners and muh, sovereignity and blabla... but why do they keep giving away these huge Government contracts to foreign companies? What the heck is wrong with them?

"You're shite, we hate you, we're gonna fuck off now... oh, btw could you please make our new anti-Europe passports for us? We're really shit at doing that kind of stuff..."

It's like god finally found a place to shit dumbness into. And it's the brains of UK politicians.

[Kronik85]

Fun day today on Tory Island (Love Island starts soon right, I'm sure we can get a good analog going on between that and the leadership contest). Jeremy "420 Blaze It" Hunt and Rory "It Wasn't a Crack Pipe" Stewart making a bid for the youth vote (all 6 of then in the Conservative membership) by admitting to past drug use.

Then Sajid Javiid seems to have made the entirity of Scotland mad and is now being relentlessly trolled as Scots ask him for permission on everything.

This is going to be a fun contest.