The Vote - the mechanics of it all....

kabayiri

tincans6 wrote: »

...

I'm just surprised no one does it.

You assume that people don't lie about their choices.

I do.

It's a temporary measure, they will soon analyse my voice pattern for the signs of stress lying can produce.

Having worked on mass marketing systems I find myself disliking the ever intrusive nature of these statistical based approaches.

You are not a voter, you are a voting unit !

If they can, they will cross-correlate your vote with your age/gender/sexual preference/education/holiday preferences/region/height/weight/hair loss proclivity.

We will be told that we won't need to vote in future, because our own centrally held Avatar will do it for us.

Generali

tincans6 wrote: »

While this is true, the couple of advantages an exit poll has are;

1) It captures just those who actually vote. Obviously ignores postal.

2) The sample size (9000+) is a magnitude bigger than most opinion polls (often around 1000).

There are plenty of fairly reliable indicators of who are more likely to vote one way or another.

I'm just surprised no one does it.

How do you set up your sample? That's the problem.

Normally you'd look at the last Brexit referendum and think about the age/income/wealth/education/geography of how Brexitiers and Remainiacs spread last time and try to reflect that in sampling this time.

Trouble is, most people that voted last time are long dead.

kabayiri

What happens when they analyse the results to work out the most Leave-happy town/village in Britain?

Will the Guardian readers lay siege to it?

I hear a crack parachuting squad are already being trained, headed by Mr Cumberbatch and Ms Thompson.

I bet there is an interview or 12 with a "Social Cohesion" consultant on breakfast telly next week.

If only the EU had a social cohesion fund we could call upon...

wotsthat

Generali wrote: »

How do you set up your sample? That's the problem.

Normally you'd look at the last Brexit referendum and think about the age/income/wealth/education/geography of how Brexitiers and Remainiacs spread last time and try to reflect that in sampling this time.

Trouble is, most people that voted last time are long dead.

Someone on Bloomberg was talking about Brexit polls this morning.

Apparently remainers are much more difficult to catch at home vs leavers which is making sample adjustment more difficult.

Don't know if that's true but my confirmation bias was fed by this little factoid.

Generali

wotsthat wrote: »

Someone on Bloomberg was talking about Brexit polls this morning.

Apparently remainers are much more difficult to catch at home vs leavers which is making sample adjustment more difficult.

Don't know if that's true but my confirmation bias was fed by this little factoid.

If you're sampling older vs younger then that is a very real sampling problem: older people are easier to reach and tend to be more inclined to answer questions. Changing your sample to try to resolve that can make matters worse in unpredictable ways.

tincans6

Generali wrote: »

How do you set up your sample? That's the problem.

Normally you'd look at the last Brexit referendum and think about the age/income/wealth/education/geography of how Brexitiers and Remainiacs spread last time and try to reflect that in sampling this time.

Trouble is, most people that voted last time are long dead.

The reason that just over 1000 is a usual sample size (alongside cost) is that gives you a 3% margin of error with a 95% confidence interval. Everything else pollsters do in these polls is to try and make sure they haven't got a biased sample.

I thought if you went to a sample of over 9000 your margin of error would reduce to 1% (again 95% confidence).

You then just need to try and ensure that your sample is not skewed by choices of polling stations and times that you ask people.

I have no idea however what the non response or response bias normally is.

It's not perfect by any means - and I appreciate that if there is unintended bias in the sample, the larger sample size doesn't get rid of it, but in a time of instant gratification, I'm still surprised it's not being done.

Shakethedisease

tincans6 wrote: »

The reason that just over 1000 is a usual sample size (alongside cost) is that gives you a 3% margin of error with a 95% confidence interval. Everything else pollsters do in these polls is to try and make sure they haven't got a biased sample.

I thought if you went to a sample of over 9000 your margin of error would reduce to 1% (again 95% confidence).

You then just need to try and ensure that your sample is not skewed by choices of polling stations and times that you ask people.

I have no idea however what the non response or response bias normally is.

It's not perfect by any means - and I appreciate that if there is unintended bias in the sample, the larger sample size doesn't get rid of it, but in a time of instant gratification, I'm still surprised it's not being done.

The lack of public exit polls bother me a bit I have to say ( as it did in the Scottish ref). I realise they're difficult to set up, but we have poll after poll daily based on filtering turnout and how people 'might' vote if at all. It would be useful to see how people actually HAVE voted in a few key areas once the polls close.

Also how come there's loads of rumours of hedge funds and banks doing 'private exit polls'.If so why can't they be done publically ? All I ever hear is that they're just too difficult to do. Obviously not difficult enough if these reports are true and some pretty hefty betting relies on the results of private exit polls.

Hedge funds and banks commission Brexit exit polls

Entrepreneurial traders are preparing to lay big bets on the result

By finding out the voting patterns early on June 23 and predicting the result, entrepreneurial traders can lay big bets on the result, hoping to be the first to benefit financially from a government-induced swing in sterling since George Soros bet against the pound when it crashed out of the then European exchange rate mechanism in 1992.

Early indications of the likely result in the referendum will be indirectly visible from foreign exchange and sterling derivative markets before the polls close, if big money is bet on the result.
The hedge funds are exploiting Electoral Commission rules that permit exit polls on the day of the referendum so long as they are not published until polls close at 10pm.

https://next.ft.com/content/7e26d896-241c-11e6-9d4d-c11776a5124d


I did a Yougov poll today that asked me what time I thought I would be voting/going to the polling station ( I said I was 100% certain to vote ). I wonder if they're planning on emailing people on Thursday based on the times given.

Mistermeaner

The results show is available from weds night on US netflicks. If you have a playstation or xbox change your location in settings to access it

wymondham

Happy voting people... see you on the other side!

ruggedtoast

vote no to ukip vote no to clapton vote no to xenophobia vote remain