This forum is about wrong numbers in science, politics and the media. It respects good science and good English.
Correction - 6 out of the 9 polls predicted Remain. Leave was predicted by 3 polls, Opinium, TNS and ICM.
But did you know that Leave voting was caused by obesity?
You can see this particular bit of junk science (by the Univeristy of East Anglia - who else!) here:
It took me a while to understand the Top 10 districts tables in the Peter Ormosi blog post. The column headings are in percent, but the figures are actually decimal fractions, so need to be multiplied by 100 to turn them into percent.
One major problem with the "obese vote leave" theory is that it is obviously going to break down for Scotland, the land of the deep-fried Mars bar.
If we look at the overall obesity statistics for the four home nations of the UK, England, Scotland, Wales, Northern Ireland using a recent House of Commons briefing paper:
The overall obesity rate is: Scotland 28%, Northern Ireland 25%, England 24%, Wales 22%. The "obese vote leave" theory would suggest Scotland and Northern Ireland would be more likely to vote Leave out of the four nations, when they actually voted Remain. The two less obese home nations, England and Wales, should be more likely to vote Remain out of the four nations, but actually voted Leave.
The referendum betting was interesting.A number of media commentators stated that the bookmakers got it wrong. They should know better. Bookmakers have one objective and that is a profit regardless of the outcome.It is called "making a book". It can be tricky with price swings but most will have balanced the book + a profit.
On the Tuesday the best leave price was 3/1 & remain price 1/4.
This corresponds to 25% & 80% giving 105% or in bookie parlance 5% " over round".
and a 4.7% return on turnover..
Leave stood out as the value bet for the punter The polls generally predicted a remain win but by a far smaller margin than that suggested by the betting.
What was that economics student thinking when he wagered his £2000 student loan
seemingly at 1/8 odds?
The pro-Remain faction of the UK news media seemed to be quite keen on the idea of portraying the bookmakers as being some sort of 'streetwise' version of the opinion polling organisations in the runup to the EU referendum. The opinion poll results for the referendum were presenting a confusing picture, with telephone polls often suggesting a comfortable win for Remain, but the supposedly less representative online polls suggesting a much closer result. The bookmaker odds always suggested a comfortable win for Remain, so the pro-Remain journalists talked up the betting activity as they liked the look of it.
But the impression I have is that betting on political events is a bit different to normal betting in the UK. In political betting there seem to be a significant number of what might be called investor-type gamblers, who are prepared to bet much larger than average amounts and appear to come from the financial community. The economics student who squandered £2000 might be regarded as being one of these. The ultimate example of one of these investor-type gamblers would be an anonymous individual who bet £900K on a No vote in the 2014 Scottish Referendum, and managed to make a £200K profit. This BBC article describes the thinking that went behind the £900K bet:
It seems evident from the article that the high stakes punter believes that opinion polling is very accurate, and that opinion pollsters can be relied upon to carry out accurate sampling of the public. I suspect that in the 2016 EU Referendum, the investor-type gamblers were assuming that the opinion polls were accurate enough to be confident of a Remain win. The largest bet was £100K for Remain by a woman in Central London. I understand that about two thirds of the money being bet on the referendum was for Remain, whilst about 70% of betting transactions were actually for Leave. The bookmakers reacted to all the money being stacked on Remain by shortening the odds, making the probability of a Remain victory look unrealistically high.