Brexit looming

[Troll]

You forgot:

Condemning the barbaric acts of these terrorists.

That'll show 'em

[d000hg]

We already share Intel across borders

Intel outside?

[jamesbrown]

Linky please?

Some analysis here.

A specific example (from a recent ORB poll) is provided here (there are others, but I can't be arsed to look them out ).

Of course, the difficulty of making these distributional adjustments is that you need a larger sample size than the typical poll provides. It's all disputable. However, it's also intuitively reasonable that the net effects of various turnout factors will favour Leave, because the strongest predictor (IIRC, the only one with any significant bearing on voting preference) is age, and likelihood of turnout is strongly related to age.

I still don't believe it will be enough, but I expect the results will be much closer than the ~60/40 that many remainiacs expect, particularly when factoring in events, as there are no events that can really improve the odds of a Remain vote, but many events that can reduce those odds.

[sasguru]

Some analysis here.

A specific example (from a recent ORB poll) is provided here (there are others, but I can't be arsed to look them out ).

Of course, the difficulty of making these distributional adjustments is that you need a larger sample size than the typical poll provides. It's all disputable. However, it's also intuitively reasonable that the net effects of various turnout factors will favour Leave, because the strongest predictor (IIRC, the only one with any significant bearing on voting preference) is age, and likelihood of turnout is strongly related to age.

I still don't believe it will be enough, but I expect the results will be much closer than the ~60/40 that many remainiacs expect, particularly when factoring in events, as there are no events that can really improve the odds of a Remain vote, but many events that can reduce those odds.

There's been much discussion in the statistical press (Significance and others) about why the polls failed in the general election, with inconclusive results.
And until that issue is fixed I'm inclined not to trust them.
However the bookies who must use some (different) statos usually get it right

http://www.oddschecker.com/politics/...bership-result

[jamesbrown]

There's been much discussion in the statistical press (Significance and others) about why the polls failed in the general election, with inconclusive results.
And until that issue is fixed I'm inclined not to trust them.
However the bookies who must use some (different) statos usually get it right

Brexit Referendum Betting Odds | Oddschecker

Sometimes they do. OTOH, I'm not aware of any conclusive evidence on that. For example:

http://blogs.lse.ac.uk/politicsandpo...tting-markets/

When the implied probabilities are translated into discrete outcomes (i.e. max likelihood), the betting odds in the last GE were no better at predicting those outcomes than their poll-aggregator (electionforecast.co.uk).

[DimPrawn]

Sometimes they do. OTOH, I'm not aware of any conclusive evidence on that. For example:

http://blogs.lse.ac.uk/politicsandpo...tting-markets/

When the implied probabilities are translated into discrete outcomes (i.e. max likelihood), the betting odds in the last GE were no better at predicting those outcomes than their poll-aggregator (electionforecast.co.uk).

Bayesian or frequentist?

[jamesbrown]

BTW, I don't think there's anything particularly magic about how bookies do this. There's a lot of subjectivity (and many variables involved), and they also need to account for how bets are actually placed (tendency to back a favourite, for example), in order to avoid over-exposing themselves. Take an exchange like Betfair, which is essentially an opinion aggregator. They tend to offer similar odds to regular bookies. It just depends whether you prefer the gut instincts of the bookies, the betting public, or the general public.

[jamesbrown]

Bayesian or frequentist?

Who, me?

[DimPrawn]

Who, me?

Probably.

[jamesbrown]

Probably.

This is about the best I could find Sad lot, statisticians.