Brexit Poll

[RandBlade]

Regardless of how you think the country should vote, how do you expect it will vote?

My gut feel is a narrow Remain win, approximately Remain 45% Leave 55%. At the start of the campaign I expected Remain 66% Leave 34% but since then a lot of people I know have changed their minds and all in one direction (myself included).

I don't trust the opinion polls, especially after the debacle last year, but I think the opinion polls will overestimate Leave. This is due to two factors, firstly in referenda polls almost inevitably overestimate the change option, secondly the polls cover Great Britain only and don't take into account Northern Ireland, Gibraltar and Overseas voters which will largely break Remain and could be another 1% for Remain. Finally I think online polls are most likely to overestimate Leave, phone polls will be most accurate.

Opinion Polls since Purdah began (27 May) have had a narrow Leave average lead, with a poll of polls putting it on 49% Remain to 51% Leave which I don't think is enough. I think Leave needs at least a 5% lead in the polls to be at "par". Having said that since purdah began there have been 10 Leave leads in the polls, 1 tie and 6 Remain leads. 6 Leave leads over 5% including the most recent 4 in a row
http://whatukthinks.org/eu/questions...he-eu/?removed

[Unheard Of]

I predict the most egregious liars and self-serving arseholes will win by approximately 5%.

[Steely Glint]

Wouldn't like to guess. Opinion polls are likely a crock for the same reason they were in the Scottish referendum. I don't think a 10% win counts as 'narrow', though.

[Aimless]

I think it will depend on who wins the river battle between Geldof and Farage, will return with an estimate after that.

[RandBlade]

Ipsos-MORI have just announced a massive swing from last month: Leave 53% (+10) Remain 47% (-10) which brings them in line with almost every other pollster. Apart from Comres (who went from Remain +11 to Remain +1 in their last poll) every other pollster now has Leave in the lead.

[Unheard Of]

We are completely fucked.

[Aimless]

Remain wins

53.8%

Subject to revision

[Aimless]

You should set a deadline for final guess.

[Hazir]

Given that Brits are buying much more gold than usual I am starting to believe the Out's may have it.

[Timbuk2]

We are completely fucked.

[Aimless]

Btw, what do you guys believe the turnout will be?

[Hazir]

Btw, what do you guys believe the turnout will be?

I guess in the 80s. There are always droves of people who really don't get it.

[Flixy]

I know from my coworkers that voting from abroad is definitely not easy, they didn't manage to do it in time for various reasons. Which is somewhat significant because Britains abroad would more likely support remain.

[Hazir]

I know from my coworkers that voting from abroad is definitely not easy, they didn't manage to do it in time for various reasons. Which is somewhat significant because Britains abroad would more likely support remain.

Actually I don't believe that, if the stakes are this high you could have started on the registration earlier and make sure you could vote

Interesting that people assume that murder yesterday is going to make a difference. I don't really believe that is the case. I do remember they thought the same when the Swedes were going to have their sham referendum about the euro.

[Aimless]

Actually I don't believe that, if the stakes are this high you could have started on the registration earlier and make sure you could vote

Interesting that people assume that murder yesterday is going to make a difference. I don't really believe that is the case. I do remember they thought the same when the Swedes were going to have their sham referendum about the euro.

I'm having this exact same discussion with an American poster on another forum. It reeks of the worst kind of amateur punditry and the problem is that the truth of his belief can't be objectively demonstrated in the event of a win for Remain, although a Brexit would at least show his belief (that Remain will win because of this tragedy) to be wrong.

There isn't much to go on. The closest example is indeed the murder of the much loved and strongly pro-Euro minister Anna Lindh days before the Swedish Euro-referendum, where many pundits believed it would lead to a victory for the pro-Euro side. Instead, they were soundly defeated. Not nearly as politically charged so not directly comparable, but the closest I can think of. In the US, as we've seen, tragic mass shootings that grab the public's attention rarely--if ever--lead to particularly impressive changes in firearms legislation or, I believe, in voter behavior.

[Khendraja'aro]

Actually I don't believe that, if the stakes are this high you could have started on the registration earlier and make sure you could vote

Interesting that people assume that murder yesterday is going to make a difference. I don't really believe that is the case. I do remember they thought the same when the Swedes were going to have their sham referendum about the euro.

Depends on how exactly voter registration works, Hazir. Over here, you can't really "start early" or something.

[Hazir]

Depends on how exactly voter registration works, Hazir. Over here, you can't really "start early" or something.

Appearantly you could register online.

[RandBlade]

Registration is done online, takes about 5 minutes and can be done any time with no starting deadline you can only register from if that makes sense.

[Flixy]

I forgot what the exact issue was. Had to do something with constituencies and previous addresses which was a bit weird as they moved quite a lot.

[Aimless]

So, any of you guys putting any real money on this?

[Timbuk2]

Y'know what, I think I might.

[Aimless]

Atta boy I haven't decided on whether or not to hedge my bets, leaning towards not hedging.

Afaict polls, esp. online polls, tend to underestimate the conservative vote and the status quo bias.

Typically, those same polls tend to have an unrepresentative sample where turnout--based on various models--is expected to be much higher than in the general population. YouGov at least may have tried to change their panels a little to adjust for that discrepancy but they've used data from general elections to do so and I believe that, all other things being equal,

Iirc, people who are undecided, or say that they don't know how they'll vote, are more likely to vote for the conservative option or the status quo, provided they show up at all.

Finally, iirc, these surveys tend to contain a disproportionately high number of responses from politically interested people with strong opinions.


The first should indicate a victory for Remain based on the averages of the most recent polls that put the two camps neck and neck.

Including the undecided vote and apportioning those with a slight bias towards Remain would make it a fairly comfortable victory.

I do believe that turnout will be fairly high for the referendum. The previous EU referendum in the UK had a turnout about 5% lower than in the preceding general election in spite of a politically interested electorate. However, I believe several things speak in favour of a higher turnout for this referendum: simple majority, perception of this being a close election, binding/semi-binding and a very important question that affects everyone. One study in Norway showed that, at the local level, semi-binding referendums had higher turnouts of about 11%. I think that's a little too much to expect from this election because Norwegians are odd and because those results may partly reflect the likelihood of semi-binding referendums to be held for more important questions and partly be an artefact of low interest in local referendums in general. At the same time, this is a very high-profile issue so I don't feel comfortable discounting the effect too heavily.

We should at minimum be able to expect a turnout over 60%. I'm gonna say 70% likelihood of turnout >/= 65% with my guess being 67% of registered voters. I think I'm more likely to be underestimating turnout than over-estimating it and frankly the scottish referendum's turnout (compared to turnout for general elections in Scotland) makes Hazir's estimate seem more appealing. Further revisions may follow later tonight

[Timbuk2]

£10 down on Remain at odds of 2/7

[Hazir]

I read this piece, I think on the economist website that also went into the difference you get if you try harder to get people to reply. Like not giving up at the first unanswered call, but simply try again untill you can ask them. There appearantly was a 4% difference.

[Aimless]

£10 down on Remain at odds of 2/7

I did the same with £23, no-one wanted to take my other bet It's the first time I've placed a bet online and I'm hoping the universe won't see fit to punish me for it.



Interesting approach described in an article from Jan: http://www.express.co.uk/news/uk/638...paign-WILL-win

Not sure I find it persuasive but will have to read more.

[Loki]

The length in office argument is problematic for an obvious reason: governments frequently use referendums when they don't think they have adequate support for an important decision or when they don't want to be blamed for a major mistake. Popular politicians don't need to call referendums. So it makes perfect sense that a politician who's been in power for a whole calling a referendum is more likely to be an act of desperation (which is likely to lose).

[LittleFuzzy]

I question the usefulness of an analysis relying on referendums from different countries (with their different political systems & structures, voter participation, party systems, etc) It said he got the Scottish referendum almost spot-on. How has his system fared with other votes?

[Aimless]

The length in office argument is problematic for an obvious reason: governments frequently use referendums when they don't think they have adequate support for an important decision or when they don't want to be blamed for a major mistake. Popular politicians don't need to call referendums. So it makes perfect sense that a politician who's been in power for a whole calling a referendum is more likely to be an act of desperation (which is likely to lose).

I question the usefulness of an analysis relying on referendums from different countries (with their different political systems & structures, voter participation, party systems, etc) It said he got the Scottish referendum almost spot-on. How has his system fared with other votes?

I haven't looked closely at the research yet, I got the impression he's using data pertaining to 49 EU-related referendums and running some form of regression analysis on that data. Doesn't sound like he's had much opportunity to test the model on new EU-related referendums. I don't think his alleged success in forecasting the outcome of the Scottish referendum is necessarily relevant. Curiously, his research appears to have shown previously that referendums for independence strongly favour those who want to gain independence which would suggest that his other investigated variables must have had a disproportionate impact on the outcome of the Scottish referendum. Anyway I believe the sample is too small and heterogenous for this approach to be useful even apart from Loki's observation about referendums being the tool of impopular politicians. His first prediction 6 months ago was that Leave will def. win, now it's that Remain will probably win. Even if the economic variables may have changed summat that shouldn't have such a dramatic impact on the forecast unless his other variables (length in office etc) have a negligible impact.


That being said, if the model were to be accurate in forecasting desperate referendums called by impopular politicians who've almost overstayed their welcome, then this would be a prime example of such a referendum, wouldn't it?

[Aimless]

V sad and I don't like the implications of this for the near future of Swedish politics.

That said it's been very educational in many ways. Not only have I learned more about international law, EU law etc in the past few weeks of discussion than I have in years, I've also learned a valuable lesson about not making overly confident predictions based on dubious assumptions about dubious data. In reality the lack of comparable events and the problems with the samples and models used by pollsters should have led me to assign a much lower probability for Bremain.

Hard to say whether there's anything to learn from my failed forecast of turnout. While I predicted 67% and was initially quite certain I'd underestimated it I become more uncertain later and wanted to revise down to 65% because I still thought the previous EU referendum would be the best way to determine a baseline, extrapolating from a single data point. Viewed as a kind of independence referendum perhaps a higher turnout could have been more plausible.

Another thing that seems plausible is that aggregated poll data can be extremely misleading esp. when it mostly adjusts for things that are less directly relevant, such as demographic characteristics and self reported intentions, or when sample composition and analyses have been informed by findings from very different events (general election with low turnout, tactical voting, UKIP etc).

Finally, I believe I made the mistake of assuming that the conservative position would be either to preserve the status quo or to weigh the economical benefits etc of EU membership highly. I recognized but failed to take into account the atypical conservative, the kind that would a few decades ago have been Labour but had switched to Con and later UKIP. Or the kind that just want the French and Germans to fuck off or whatever, basically the traditional working class nationalist conservative. If the conservative position can just as well be to want to gtfo then my assumption about the polls was overly confident.

So I believe the confidence I assigned to my predictions did not accurately reflect the uncertainties I'd identified and in trying to arrive at a decision about what the outcome would most likely be I only considered the evidence and arguments that pointed in one direction (assumed pro-Remain bias among conservatives).

I don't belive I could have pegged the actual outcome based on the available data however. In this case I would have had to have more insight into attitudes and politics social aspects of life at the local level.

Another interesting observation is that the betting markets seem to have been very off. I expect most of the information in those markets came from polls and that this failure was thus a case of GIGO. Same for forex.

[Loki]

Well said. Though I don't think local politics are responsible for this. The British have an innate hatred of the EU. Do you remember the EU poll I posted recently, which had about a third of the people in most countries saying the EU should gain more power, while virtually no one Britain said the same. That should have been a big clue that the default position in Britain is to attack the EU, party politics be damned.

I see a lot of Labour strongholds voting to leave by the way (including most of Wales). It looks like it's mainly the areas with large number of immigrants and/or a highly educated work force that voted to remain.